id	sid	tid	token	lemma	pos
easat-2167	1	1	edelweiss	edelweiss	PROPN
easat-2167	1	2	applied	apply	VERB
easat-2167	1	3	science	science	NOUN
easat-2167	1	4	and	and	CCONJ
easat-2167	1	5	technology	technology	NOUN
easat-2167	1	6	issn	issn	PROPN
easat-2167	1	7	:	:	PUNCT
easat-2167	1	8	2576	2576	NUM
easat-2167	1	9	-	-	SYM
easat-2167	1	10	8484	8484	NUM
easat-2167	1	11	vol	vol	NOUN
easat-2167	1	12	.	.	PROPN
easat-2167	1	13	8	8	NUM
easat-2167	1	14	,	,	PUNCT
easat-2167	1	15	no	no	INTJ
easat-2167	1	16	.	.	NOUN
easat-2167	1	17	6	6	NUM
easat-2167	1	18	,	,	PUNCT
easat-2167	1	19	793	793	NUM
easat-2167	1	20	-	-	SYM
easat-2167	1	21	813	813	NUM
easat-2167	1	22	2024	2024	NUM
easat-2167	1	23	publisher	publisher	NOUN
easat-2167	1	24	:	:	PUNCT
easat-2167	1	25	learning	learn	VERB
easat-2167	1	26	gate	gate	NOUN
easat-2167	1	27	doi	doi	PROPN
easat-2167	1	28	:	:	PUNCT
easat-2167	1	29	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	1	30	©	©	NOUN
easat-2167	1	31	2024	2024	NUM
easat-2167	1	32	by	by	ADP
easat-2167	1	33	the	the	DET
easat-2167	1	34	authors	author	NOUN
easat-2167	1	35	;	;	PUNCT
easat-2167	1	36	licensee	licensee	PROPN
easat-2167	1	37	learning	learning	NOUN
easat-2167	1	38	gate	gate	NOUN
easat-2167	1	39	©	©	PROPN
easat-2167	1	40	2024	2024	NUM
easat-2167	1	41	by	by	ADP
easat-2167	1	42	the	the	DET
easat-2167	1	43	authors	author	NOUN
easat-2167	1	44	;	;	PUNCT
easat-2167	1	45	licensee	licensee	PROPN
easat-2167	1	46	learning	learn	VERB
easat-2167	1	47	gate	gate	NOUN
easat-2167	1	48	*	*	PUNCT
easat-2167	1	49	correspondence	correspondence	NOUN
easat-2167	1	50	:	:	PUNCT
easat-2167	1	51	ajay.khunteta@poornima.edu.in	ajay.khunteta@poornima.edu.in	PROPN
easat-2167	1	52	"	"	PUNCT
easat-2167	1	53	a	a	DET
easat-2167	1	54	novel	novel	ADJ
easat-2167	1	55	hybrid	hybrid	ADJ
easat-2167	1	56	deep	deep	ADJ
easat-2167	1	57	learning	learning	NOUN
easat-2167	1	58	framework	framework	NOUN
easat-2167	1	59	for	for	ADP
easat-2167	1	60	enhanced	enhanced	ADJ
easat-2167	1	61	segmentation	segmentation	NOUN
easat-2167	1	62	of	of	ADP
easat-2167	1	63	rice	rice	NOUN
easat-2167	1	64	leaf	leaf	NOUN
easat-2167	1	65	diseases	disease	NOUN
easat-2167	1	66	using	use	VERB
easat-2167	1	67	attention	attention	NOUN
easat-2167	1	68	-	-	PUNCT
easat-2167	1	69	driven	drive	VERB
easat-2167	1	70	efficientnet	efficientnet	NOUN
easat-2167	1	71	models	model	NOUN
easat-2167	1	72	"	"	PUNCT
easat-2167	1	73	pooja	pooja	PROPN
easat-2167	1	74	sharma1	sharma1	PROPN
easat-2167	1	75	,	,	PUNCT
easat-2167	1	76	ajay	ajay	PROPN
easat-2167	1	77	khunteta2	khunteta2	PROPN
easat-2167	1	78	*	*	PROPN
easat-2167	1	79	1,2poornima	1,2poornima	PROPN
easat-2167	1	80	university	university	NOUN
easat-2167	1	81	,	,	PUNCT
easat-2167	1	82	jaipur	jaipur	PROPN
easat-2167	1	83	;	;	PUNCT
easat-2167	1	84	er9.pooja@gmail.com	er9.pooja@gmail.com	PROPN
easat-2167	1	85	(	(	PUNCT
easat-2167	1	86	p.s	p.s	PROPN
easat-2167	1	87	.	.	PUNCT
easat-2167	1	88	)	)	PUNCT
easat-2167	2	1	ajay.khunteta@poornima.edu.in	ajay.khunteta@poornima.edu.in	PROPN
easat-2167	2	2	(	(	PUNCT
easat-2167	2	3	a.k	a.k	PROPN
easat-2167	2	4	.	.	PROPN
easat-2167	2	5	)	)	PUNCT
easat-2167	2	6	.	.	PUNCT
easat-2167	3	1	abstract	abstract	ADJ
easat-2167	3	2	:	:	PUNCT
easat-2167	4	1	rice	rice	NOUN
easat-2167	4	2	diseases	disease	NOUN
easat-2167	4	3	such	such	ADJ
easat-2167	4	4	as	as	ADP
easat-2167	4	5	bacterial	bacterial	ADJ
easat-2167	4	6	leaf	leaf	NOUN
easat-2167	4	7	blight	blight	NOUN
easat-2167	4	8	,	,	PUNCT
easat-2167	4	9	brown	brown	ADJ
easat-2167	4	10	spot	spot	NOUN
easat-2167	4	11	,	,	PUNCT
easat-2167	4	12	and	and	CCONJ
easat-2167	4	13	blast	blast	NOUN
easat-2167	4	14	pose	pose	VERB
easat-2167	4	15	significant	significant	ADJ
easat-2167	4	16	threats	threat	NOUN
easat-2167	4	17	to	to	ADP
easat-2167	4	18	global	global	ADJ
easat-2167	4	19	food	food	NOUN
easat-2167	4	20	security	security	NOUN
easat-2167	4	21	by	by	ADP
easat-2167	4	22	reducing	reduce	VERB
easat-2167	4	23	crop	crop	NOUN
easat-2167	4	24	yields	yield	NOUN
easat-2167	4	25	,	,	PUNCT
easat-2167	4	26	making	make	VERB
easat-2167	4	27	early	early	ADJ
easat-2167	4	28	and	and	CCONJ
easat-2167	4	29	accurate	accurate	ADJ
easat-2167	4	30	detection	detection	NOUN
easat-2167	4	31	crucial	crucial	ADJ
easat-2167	4	32	.	.	PUNCT
easat-2167	5	1	this	this	DET
easat-2167	5	2	study	study	NOUN
easat-2167	5	3	aims	aim	VERB
easat-2167	5	4	to	to	PART
easat-2167	5	5	improve	improve	VERB
easat-2167	5	6	automated	automate	VERB
easat-2167	5	7	segmentation	segmentation	NOUN
easat-2167	5	8	of	of	ADP
easat-2167	5	9	rice	rice	NOUN
easat-2167	5	10	leaf	leaf	NOUN
easat-2167	5	11	diseases	disease	NOUN
easat-2167	5	12	using	use	VERB
easat-2167	5	13	advanced	advanced	ADJ
easat-2167	5	14	deep	deep	ADJ
easat-2167	5	15	learning	learning	NOUN
easat-2167	5	16	techniques	technique	NOUN
easat-2167	5	17	,	,	PUNCT
easat-2167	5	18	specifically	specifically	ADV
easat-2167	5	19	u	u	NOUN
easat-2167	5	20	-	-	NOUN
easat-2167	5	21	net	net	ADJ
easat-2167	5	22	with	with	ADP
easat-2167	5	23	mobilenetv2	mobilenetv2	PROPN
easat-2167	5	24	,	,	PUNCT
easat-2167	5	25	deeplabv3	deeplabv3	ADJ
easat-2167	5	26	+	+	X
easat-2167	5	27	with	with	ADP
easat-2167	5	28	efficientnetb4	efficientnetb4	PROPN
easat-2167	5	29	,	,	PUNCT
easat-2167	5	30	and	and	CCONJ
easat-2167	5	31	u	u	NOUN
easat-2167	5	32	-	-	NOUN
easat-2167	5	33	net	net	ADJ
easat-2167	5	34	with	with	ADP
easat-2167	5	35	efficientnetb7	efficientnetb7	NOUN
easat-2167	5	36	integrated	integrate	VERB
easat-2167	5	37	with	with	ADP
easat-2167	5	38	attention	attention	NOUN
easat-2167	5	39	gates	gate	NOUN
easat-2167	5	40	.	.	PUNCT
easat-2167	6	1	the	the	DET
easat-2167	6	2	models	model	NOUN
easat-2167	6	3	were	be	AUX
easat-2167	6	4	evaluated	evaluate	VERB
easat-2167	6	5	on	on	ADP
easat-2167	6	6	a	a	DET
easat-2167	6	7	dataset	dataset	NOUN
easat-2167	6	8	of	of	ADP
easat-2167	6	9	diseased	diseased	ADJ
easat-2167	6	10	rice	rice	NOUN
easat-2167	6	11	leaves	leave	NOUN
easat-2167	6	12	for	for	ADP
easat-2167	6	13	segmentation	segmentation	NOUN
easat-2167	6	14	accuracy	accuracy	NOUN
easat-2167	6	15	,	,	PUNCT
easat-2167	6	16	computational	computational	ADJ
easat-2167	6	17	efficiency	efficiency	NOUN
easat-2167	6	18	,	,	PUNCT
easat-2167	6	19	and	and	CCONJ
easat-2167	6	20	robustness	robustness	NOUN
easat-2167	6	21	in	in	ADP
easat-2167	6	22	real	real	ADJ
easat-2167	6	23	-	-	PUNCT
easat-2167	6	24	world	world	NOUN
easat-2167	6	25	conditions	condition	NOUN
easat-2167	6	26	.	.	PUNCT
easat-2167	7	1	results	result	NOUN
easat-2167	7	2	show	show	VERB
easat-2167	7	3	that	that	SCONJ
easat-2167	7	4	the	the	DET
easat-2167	7	5	u	u	NOUN
easat-2167	7	6	-	-	NOUN
easat-2167	7	7	net	net	ADJ
easat-2167	7	8	with	with	ADP
easat-2167	7	9	efficientnetb7	efficientnetb7	NOUN
easat-2167	7	10	and	and	CCONJ
easat-2167	7	11	attention	attention	NOUN
easat-2167	7	12	gates	gate	NOUN
easat-2167	7	13	outperforms	outperform	VERB
easat-2167	7	14	the	the	DET
easat-2167	7	15	other	other	ADJ
easat-2167	7	16	models	model	NOUN
easat-2167	7	17	,	,	PUNCT
easat-2167	7	18	particularly	particularly	ADV
easat-2167	7	19	for	for	ADP
easat-2167	7	20	complex	complex	ADJ
easat-2167	7	21	leaf	leaf	NOUN
easat-2167	7	22	images	image	NOUN
easat-2167	7	23	,	,	PUNCT
easat-2167	7	24	achieving	achieve	VERB
easat-2167	7	25	superior	superior	ADJ
easat-2167	7	26	accuracy	accuracy	NOUN
easat-2167	7	27	and	and	CCONJ
easat-2167	7	28	generalization	generalization	NOUN
easat-2167	7	29	.	.	PUNCT
easat-2167	8	1	this	this	DET
easat-2167	8	2	research	research	NOUN
easat-2167	8	3	provides	provide	VERB
easat-2167	8	4	a	a	DET
easat-2167	8	5	practical	practical	ADJ
easat-2167	8	6	,	,	PUNCT
easat-2167	8	7	real	real	ADJ
easat-2167	8	8	-	-	PUNCT
easat-2167	8	9	time	time	NOUN
easat-2167	8	10	solution	solution	NOUN
easat-2167	8	11	for	for	ADP
easat-2167	8	12	early	early	ADJ
easat-2167	8	13	disease	disease	NOUN
easat-2167	8	14	detection	detection	NOUN
easat-2167	8	15	in	in	ADP
easat-2167	8	16	rice	rice	NOUN
easat-2167	8	17	,	,	PUNCT
easat-2167	8	18	contributing	contribute	VERB
easat-2167	8	19	to	to	PART
easat-2167	8	20	precision	precision	VERB
easat-2167	8	21	agriculture	agriculture	NOUN
easat-2167	8	22	by	by	ADP
easat-2167	8	23	helping	help	VERB
easat-2167	8	24	reduce	reduce	VERB
easat-2167	8	25	crop	crop	NOUN
easat-2167	8	26	losses	loss	NOUN
easat-2167	8	27	and	and	CCONJ
easat-2167	8	28	optimize	optimize	VERB
easat-2167	8	29	the	the	DET
easat-2167	8	30	use	use	NOUN
easat-2167	8	31	of	of	ADP
easat-2167	8	32	agrochemicals	agrochemical	NOUN
easat-2167	8	33	,	,	PUNCT
easat-2167	8	34	ultimately	ultimately	ADV
easat-2167	8	35	promoting	promote	VERB
easat-2167	8	36	sustainable	sustainable	ADJ
easat-2167	8	37	crop	crop	NOUN
easat-2167	8	38	management	management	NOUN
easat-2167	8	39	.	.	PUNCT
easat-2167	9	1	keywords	keyword	NOUN
easat-2167	9	2	:	:	PUNCT
easat-2167	9	3	attention	attention	NOUN
easat-2167	9	4	mechanisms	mechanism	NOUN
easat-2167	9	5	,	,	PUNCT
easat-2167	9	6	computational	computational	ADJ
easat-2167	9	7	efficiency	efficiency	NOUN
easat-2167	9	8	,	,	PUNCT
easat-2167	9	9	crop	crop	NOUN
easat-2167	9	10	management	management	NOUN
easat-2167	9	11	,	,	PUNCT
easat-2167	9	12	deep	deep	ADJ
easat-2167	9	13	learning	learning	NOUN
easat-2167	9	14	,	,	PUNCT
easat-2167	9	15	deeplabv3	deeplabv3	PROPN
easat-2167	9	16	+	+	ADJ
easat-2167	9	17	,	,	PUNCT
easat-2167	9	18	disease	disease	NOUN
easat-2167	9	19	detection	detection	NOUN
easat-2167	9	20	,	,	PUNCT
easat-2167	9	21	image	image	NOUN
easat-2167	9	22	segmentation	segmentation	NOUN
easat-2167	9	23	,	,	PUNCT
easat-2167	9	24	mobilenetv2	mobilenetv2	PROPN
easat-2167	9	25	,	,	PUNCT
easat-2167	9	26	precision	precision	NOUN
easat-2167	9	27	agriculture	agriculture	NOUN
easat-2167	9	28	,	,	PUNCT
easat-2167	9	29	rice	rice	NOUN
easat-2167	9	30	leaf	leaf	NOUN
easat-2167	9	31	diseases	disease	NOUN
easat-2167	9	32	,	,	PUNCT
easat-2167	9	33	u	u	NOUN
easat-2167	9	34	-	-	NOUN
easat-2167	9	35	net	net	ADJ
easat-2167	9	36	,	,	PUNCT
easat-2167	9	37	efficient	efficient	ADJ
easat-2167	9	38	net	net	NOUN
easat-2167	9	39	.	.	PUNCT
easat-2167	10	1	1	1	X
easat-2167	10	2	.	.	X
easat-2167	10	3	background	background	NOUN
easat-2167	10	4	rice	rice	NOUN
easat-2167	10	5	is	be	AUX
easat-2167	10	6	a	a	DET
easat-2167	10	7	staple	staple	ADJ
easat-2167	10	8	food	food	NOUN
easat-2167	10	9	for	for	ADP
easat-2167	10	10	a	a	DET
easat-2167	10	11	significant	significant	ADJ
easat-2167	10	12	portion	portion	NOUN
easat-2167	10	13	of	of	ADP
easat-2167	10	14	the	the	DET
easat-2167	10	15	global	global	ADJ
easat-2167	10	16	population	population	NOUN
easat-2167	10	17	,	,	PUNCT
easat-2167	10	18	especially	especially	ADV
easat-2167	10	19	in	in	ADP
easat-2167	10	20	asia	asia	PROPN
easat-2167	10	21	.	.	PUNCT
easat-2167	11	1	rice	rice	NOUN
easat-2167	11	2	diseases	disease	NOUN
easat-2167	11	3	,	,	PUNCT
easat-2167	11	4	such	such	ADJ
easat-2167	11	5	as	as	ADP
easat-2167	11	6	bacterial	bacterial	ADJ
easat-2167	11	7	leaf	leaf	NOUN
easat-2167	11	8	blight	blight	NOUN
easat-2167	11	9	,	,	PUNCT
easat-2167	11	10	brown	brown	ADJ
easat-2167	11	11	spot	spot	NOUN
easat-2167	11	12	,	,	PUNCT
easat-2167	11	13	and	and	CCONJ
easat-2167	11	14	blast	blast	NOUN
easat-2167	11	15	,	,	PUNCT
easat-2167	11	16	pose	pose	VERB
easat-2167	11	17	severe	severe	ADJ
easat-2167	11	18	threats	threat	NOUN
easat-2167	11	19	to	to	ADP
easat-2167	11	20	crop	crop	NOUN
easat-2167	11	21	yields	yield	NOUN
easat-2167	11	22	,	,	PUNCT
easat-2167	11	23	impacting	impact	VERB
easat-2167	11	24	food	food	NOUN
easat-2167	11	25	security	security	NOUN
easat-2167	11	26	and	and	CCONJ
easat-2167	11	27	farmers	farmer	NOUN
easat-2167	11	28	'	'	PART
easat-2167	11	29	livelihoods	livelihood	NOUN
easat-2167	11	30	.	.	PUNCT
easat-2167	12	1	early	early	ADJ
easat-2167	12	2	and	and	CCONJ
easat-2167	12	3	accurate	accurate	ADJ
easat-2167	12	4	detection	detection	NOUN
easat-2167	12	5	of	of	ADP
easat-2167	12	6	these	these	DET
easat-2167	12	7	diseases	disease	NOUN
easat-2167	12	8	is	be	AUX
easat-2167	12	9	crucial	crucial	ADJ
easat-2167	12	10	for	for	ADP
easat-2167	12	11	effective	effective	ADJ
easat-2167	12	12	management	management	NOUN
easat-2167	12	13	and	and	CCONJ
easat-2167	12	14	prevention	prevention	NOUN
easat-2167	12	15	of	of	ADP
easat-2167	12	16	crop	crop	NOUN
easat-2167	12	17	loss	loss	NOUN
easat-2167	12	18	.	.	PUNCT
easat-2167	13	1	while	while	SCONJ
easat-2167	13	2	numerous	numerous	ADJ
easat-2167	13	3	methods	method	NOUN
easat-2167	13	4	have	have	AUX
easat-2167	13	5	been	be	AUX
easat-2167	13	6	developed	develop	VERB
easat-2167	13	7	for	for	ADP
easat-2167	13	8	rice	rice	NOUN
easat-2167	13	9	leaf	leaf	NOUN
easat-2167	13	10	disease	disease	NOUN
easat-2167	13	11	detection	detection	NOUN
easat-2167	13	12	,	,	PUNCT
easat-2167	13	13	challenges	challenge	NOUN
easat-2167	13	14	such	such	ADJ
easat-2167	13	15	as	as	ADP
easat-2167	13	16	segmentation	segmentation	NOUN
easat-2167	13	17	accuracy	accuracy	NOUN
easat-2167	13	18	,	,	PUNCT
easat-2167	13	19	handling	handle	VERB
easat-2167	13	20	complex	complex	ADJ
easat-2167	13	21	backgrounds	background	NOUN
easat-2167	13	22	,	,	PUNCT
easat-2167	13	23	and	and	CCONJ
easat-2167	13	24	the	the	DET
easat-2167	13	25	need	need	NOUN
easat-2167	13	26	for	for	ADP
easat-2167	13	27	large	large	ADJ
easat-2167	13	28	,	,	PUNCT
easat-2167	13	29	labelled	label	VERB
easat-2167	13	30	datasets	dataset	NOUN
easat-2167	13	31	persist	persist	VERB
easat-2167	13	32	.	.	PUNCT
easat-2167	14	1	image	image	NOUN
easat-2167	14	2	segmentation	segmentation	NOUN
easat-2167	14	3	is	be	AUX
easat-2167	14	4	a	a	DET
easat-2167	14	5	vital	vital	ADJ
easat-2167	14	6	technique	technique	NOUN
easat-2167	14	7	in	in	ADP
easat-2167	14	8	the	the	DET
easat-2167	14	9	field	field	NOUN
easat-2167	14	10	of	of	ADP
easat-2167	14	11	computer	computer	NOUN
easat-2167	14	12	vision	vision	NOUN
easat-2167	14	13	and	and	CCONJ
easat-2167	14	14	image	image	NOUN
easat-2167	14	15	processing	processing	NOUN
easat-2167	14	16	,	,	PUNCT
easat-2167	14	17	particularly	particularly	ADV
easat-2167	14	18	for	for	ADP
easat-2167	14	19	agricultural	agricultural	ADJ
easat-2167	14	20	applications	application	NOUN
easat-2167	14	21	.	.	PUNCT
easat-2167	15	1	it	it	PRON
easat-2167	15	2	involves	involve	VERB
easat-2167	15	3	partitioning	partition	VERB
easat-2167	15	4	an	an	DET
easat-2167	15	5	image	image	NOUN
easat-2167	15	6	into	into	ADP
easat-2167	15	7	different	different	ADJ
easat-2167	15	8	segments	segment	NOUN
easat-2167	15	9	or	or	CCONJ
easat-2167	15	10	regions	region	NOUN
easat-2167	15	11	,	,	PUNCT
easat-2167	15	12	each	each	PRON
easat-2167	15	13	representing	represent	VERB
easat-2167	15	14	a	a	DET
easat-2167	15	15	specific	specific	ADJ
easat-2167	15	16	part	part	NOUN
easat-2167	15	17	of	of	ADP
easat-2167	15	18	the	the	DET
easat-2167	15	19	image	image	NOUN
easat-2167	15	20	,	,	PUNCT
easat-2167	15	21	such	such	ADJ
easat-2167	15	22	as	as	ADP
easat-2167	15	23	the	the	DET
easat-2167	15	24	background	background	NOUN
easat-2167	15	25	,	,	PUNCT
easat-2167	15	26	healthy	healthy	ADJ
easat-2167	15	27	plant	plant	NOUN
easat-2167	15	28	tissue	tissue	NOUN
easat-2167	15	29	,	,	PUNCT
easat-2167	15	30	or	or	CCONJ
easat-2167	15	31	diseased	diseased	ADJ
easat-2167	15	32	regions	region	NOUN
easat-2167	15	33	.	.	PUNCT
easat-2167	16	1	in	in	ADP
easat-2167	16	2	the	the	DET
easat-2167	16	3	context	context	NOUN
easat-2167	16	4	of	of	ADP
easat-2167	16	5	leaf	leaf	NOUN
easat-2167	16	6	disease	disease	NOUN
easat-2167	16	7	detection	detection	NOUN
easat-2167	16	8	,	,	PUNCT
easat-2167	16	9	segmentation	segmentation	NOUN
easat-2167	16	10	is	be	AUX
easat-2167	16	11	the	the	DET
easat-2167	16	12	process	process	NOUN
easat-2167	16	13	of	of	ADP
easat-2167	16	14	isolating	isolate	VERB
easat-2167	16	15	and	and	CCONJ
easat-2167	16	16	identifying	identify	VERB
easat-2167	16	17	the	the	DET
easat-2167	16	18	diseased	diseased	ADJ
easat-2167	16	19	areas	area	NOUN
easat-2167	16	20	on	on	ADP
easat-2167	16	21	a	a	DET
easat-2167	16	22	leaf	leaf	NOUN
easat-2167	16	23	from	from	ADP
easat-2167	16	24	the	the	DET
easat-2167	16	25	healthy	healthy	ADJ
easat-2167	16	26	areas	area	NOUN
easat-2167	16	27	and	and	CCONJ
easat-2167	16	28	the	the	DET
easat-2167	16	29	background	background	NOUN
easat-2167	16	30	.	.	PUNCT
easat-2167	17	1	1.1	1.1	NUM
easat-2167	17	2	.	.	PUNCT
easat-2167	17	3	problems	problem	NOUN
easat-2167	17	4	addressed	address	VERB
easat-2167	17	5	accurate	accurate	ADJ
easat-2167	17	6	disease	disease	NOUN
easat-2167	17	7	localization	localization	NOUN
easat-2167	17	8	:	:	PUNCT
easat-2167	17	9	image	image	NOUN
easat-2167	17	10	segmentation	segmentation	NOUN
easat-2167	17	11	allows	allow	VERB
easat-2167	17	12	for	for	ADP
easat-2167	17	13	precise	precise	ADJ
easat-2167	17	14	localization	localization	NOUN
easat-2167	17	15	of	of	ADP
easat-2167	17	16	diseased	diseased	ADJ
easat-2167	17	17	regions	region	NOUN
easat-2167	17	18	on	on	ADP
easat-2167	17	19	leaves	leave	NOUN
easat-2167	17	20	,	,	PUNCT
easat-2167	17	21	distinguishing	distinguish	VERB
easat-2167	17	22	between	between	ADP
easat-2167	17	23	healthy	healthy	ADJ
easat-2167	17	24	and	and	CCONJ
easat-2167	17	25	infected	infected	ADJ
easat-2167	17	26	tissues	tissue	NOUN
easat-2167	17	27	.	.	PUNCT
easat-2167	18	1	this	this	DET
easat-2167	18	2	precision	precision	NOUN
easat-2167	18	3	is	be	AUX
easat-2167	18	4	crucial	crucial	ADJ
easat-2167	18	5	for	for	ADP
easat-2167	18	6	early	early	ADJ
easat-2167	18	7	detection	detection	NOUN
easat-2167	18	8	and	and	CCONJ
easat-2167	18	9	accurate	accurate	ADJ
easat-2167	18	10	diagnosis	diagnosis	NOUN
easat-2167	18	11	of	of	ADP
easat-2167	18	12	plant	plant	NOUN
easat-2167	18	13	diseases	disease	NOUN
easat-2167	18	14	.	.	PUNCT
easat-2167	19	1	handling	handle	VERB
easat-2167	19	2	complex	complex	ADJ
easat-2167	19	3	backgrounds	background	NOUN
easat-2167	19	4	:	:	PUNCT
easat-2167	19	5	in	in	ADP
easat-2167	19	6	field	field	NOUN
easat-2167	19	7	conditions	condition	NOUN
easat-2167	19	8	,	,	PUNCT
easat-2167	19	9	leaf	leaf	NOUN
easat-2167	19	10	images	image	NOUN
easat-2167	19	11	often	often	ADV
easat-2167	19	12	have	have	VERB
easat-2167	19	13	complex	complex	ADJ
easat-2167	19	14	backgrounds	background	NOUN
easat-2167	19	15	,	,	PUNCT
easat-2167	19	16	such	such	ADJ
easat-2167	19	17	as	as	ADP
easat-2167	19	18	soil	soil	NOUN
easat-2167	19	19	,	,	PUNCT
easat-2167	19	20	other	other	ADJ
easat-2167	19	21	plants	plant	NOUN
easat-2167	19	22	,	,	PUNCT
easat-2167	19	23	and	and	CCONJ
easat-2167	19	24	varying	vary	VERB
easat-2167	19	25	lighting	lighting	NOUN
easat-2167	19	26	conditions	condition	NOUN
easat-2167	19	27	.	.	PUNCT
easat-2167	20	1	segmentation	segmentation	NOUN
easat-2167	20	2	techniques	technique	NOUN
easat-2167	20	3	can	can	AUX
easat-2167	20	4	effectively	effectively	ADV
easat-2167	20	5	separate	separate	VERB
easat-2167	20	6	the	the	DET
easat-2167	20	7	leaf	leaf	NOUN
easat-2167	20	8	from	from	ADP
easat-2167	20	9	the	the	DET
easat-2167	20	10	background	background	NOUN
easat-2167	20	11	,	,	PUNCT
easat-2167	20	12	ensuring	ensure	VERB
easat-2167	20	13	that	that	SCONJ
easat-2167	20	14	the	the	DET
easat-2167	20	15	analysis	analysis	NOUN
easat-2167	20	16	focuses	focus	VERB
easat-2167	20	17	solely	solely	ADV
easat-2167	20	18	on	on	ADP
easat-2167	20	19	the	the	DET
easat-2167	20	20	leaf	leaf	NOUN
easat-2167	20	21	's	's	PART
easat-2167	20	22	surface	surface	NOUN
easat-2167	20	23	.	.	PUNCT
easat-2167	21	1	quantification	quantification	NOUN
easat-2167	21	2	of	of	ADP
easat-2167	21	3	disease	disease	NOUN
easat-2167	21	4	severity	severity	NOUN
easat-2167	21	5	:	:	PUNCT
easat-2167	21	6	by	by	ADP
easat-2167	21	7	segmenting	segment	VERB
easat-2167	21	8	the	the	DET
easat-2167	21	9	diseased	diseased	ADJ
easat-2167	21	10	areas	area	NOUN
easat-2167	21	11	,	,	PUNCT
easat-2167	21	12	it	it	PRON
easat-2167	21	13	is	be	AUX
easat-2167	21	14	possible	possible	ADJ
easat-2167	21	15	to	to	PART
easat-2167	21	16	quantify	quantify	VERB
easat-2167	21	17	the	the	DET
easat-2167	21	18	extent	extent	NOUN
easat-2167	21	19	of	of	ADP
easat-2167	21	20	the	the	DET
easat-2167	21	21	infection	infection	NOUN
easat-2167	21	22	,	,	PUNCT
easat-2167	21	23	providing	provide	VERB
easat-2167	21	24	valuable	valuable	ADJ
easat-2167	21	25	information	information	NOUN
easat-2167	21	26	on	on	ADP
easat-2167	21	27	the	the	DET
easat-2167	21	28	severity	severity	NOUN
easat-2167	21	29	of	of	ADP
easat-2167	21	30	the	the	DET
easat-2167	21	31	disease	disease	NOUN
easat-2167	21	32	.	.	PUNCT
easat-2167	22	1	this	this	DET
easat-2167	22	2	data	datum	NOUN
easat-2167	22	3	can	can	AUX
easat-2167	22	4	help	help	VERB
easat-2167	22	5	794	794	NUM
easat-2167	22	6	edelweiss	edelweiss	PROPN
easat-2167	22	7	applied	apply	VERB
easat-2167	22	8	science	science	NOUN
easat-2167	22	9	and	and	CCONJ
easat-2167	22	10	technology	technology	NOUN
easat-2167	22	11	issn	issn	PROPN
easat-2167	22	12	:	:	PUNCT
easat-2167	22	13	2576	2576	NUM
easat-2167	22	14	-	-	SYM
easat-2167	22	15	8484	8484	NUM
easat-2167	22	16	vol	vol	NOUN
easat-2167	22	17	.	.	PROPN
easat-2167	22	18	8	8	NUM
easat-2167	22	19	,	,	PUNCT
easat-2167	22	20	no	no	INTJ
easat-2167	22	21	.	.	NOUN
easat-2167	23	1	6	6	NUM
easat-2167	23	2	:	:	PUNCT
easat-2167	23	3	793	793	NUM
easat-2167	23	4	-	-	NUM
easat-2167	23	5	813	813	NUM
easat-2167	23	6	,	,	PUNCT
easat-2167	23	7	2024	2024	NUM
easat-2167	23	8	doi	doi	NOUN
easat-2167	23	9	:	:	PUNCT
easat-2167	23	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	23	11	©	©	NOUN
easat-2167	23	12	2024	2024	NUM
easat-2167	23	13	by	by	ADP
easat-2167	23	14	the	the	DET
easat-2167	23	15	authors	author	NOUN
easat-2167	23	16	;	;	PUNCT
easat-2167	23	17	licensee	licensee	PROPN
easat-2167	23	18	learning	learning	NOUN
easat-2167	23	19	gate	gate	NOUN
easat-2167	23	20	in	in	ADP
easat-2167	23	21	determining	determine	VERB
easat-2167	23	22	the	the	DET
easat-2167	23	23	appropriate	appropriate	ADJ
easat-2167	23	24	level	level	NOUN
easat-2167	23	25	of	of	ADP
easat-2167	23	26	intervention	intervention	NOUN
easat-2167	23	27	needed	need	VERB
easat-2167	23	28	.	.	PUNCT
easat-2167	24	1	automated	automate	VERB
easat-2167	24	2	monitoring	monitoring	NOUN
easat-2167	24	3	and	and	CCONJ
easat-2167	24	4	early	early	ADJ
easat-2167	24	5	detection	detection	NOUN
easat-2167	24	6	:	:	PUNCT
easat-2167	24	7	automated	automate	VERB
easat-2167	24	8	segmentation	segmentation	NOUN
easat-2167	24	9	allows	allow	VERB
easat-2167	24	10	for	for	ADP
easat-2167	24	11	continuous	continuous	ADJ
easat-2167	24	12	monitoring	monitoring	NOUN
easat-2167	24	13	of	of	ADP
easat-2167	24	14	crops	crop	NOUN
easat-2167	24	15	using	use	VERB
easat-2167	24	16	imaging	imaging	NOUN
easat-2167	24	17	devices	device	NOUN
easat-2167	24	18	,	,	PUNCT
easat-2167	24	19	such	such	ADJ
easat-2167	24	20	as	as	ADP
easat-2167	24	21	drones	drone	NOUN
easat-2167	24	22	or	or	CCONJ
easat-2167	24	23	cameras	camera	NOUN
easat-2167	24	24	installed	instal	VERB
easat-2167	24	25	in	in	ADP
easat-2167	24	26	fields	field	NOUN
easat-2167	24	27	.	.	PUNCT
easat-2167	25	1	early	early	ADJ
easat-2167	25	2	detection	detection	NOUN
easat-2167	25	3	of	of	ADP
easat-2167	25	4	diseases	disease	NOUN
easat-2167	25	5	through	through	ADP
easat-2167	25	6	such	such	ADJ
easat-2167	25	7	automated	automate	VERB
easat-2167	25	8	systems	system	NOUN
easat-2167	25	9	can	can	AUX
easat-2167	25	10	prevent	prevent	VERB
easat-2167	25	11	the	the	DET
easat-2167	25	12	spread	spread	NOUN
easat-2167	25	13	of	of	ADP
easat-2167	25	14	the	the	DET
easat-2167	25	15	disease	disease	NOUN
easat-2167	25	16	and	and	CCONJ
easat-2167	25	17	reduce	reduce	VERB
easat-2167	25	18	crop	crop	NOUN
easat-2167	25	19	losses	loss	NOUN
easat-2167	25	20	.	.	PUNCT
easat-2167	26	1	improving	improve	VERB
easat-2167	26	2	decision	decision	NOUN
easat-2167	26	3	-	-	PUNCT
easat-2167	26	4	making	making	NOUN
easat-2167	26	5	:	:	PUNCT
easat-2167	26	6	segmentation	segmentation	NOUN
easat-2167	26	7	helps	help	VERB
easat-2167	26	8	in	in	ADP
easat-2167	26	9	providing	provide	VERB
easat-2167	26	10	precise	precise	ADJ
easat-2167	26	11	data	datum	NOUN
easat-2167	26	12	that	that	PRON
easat-2167	26	13	can	can	AUX
easat-2167	26	14	be	be	AUX
easat-2167	26	15	used	use	VERB
easat-2167	26	16	to	to	PART
easat-2167	26	17	make	make	VERB
easat-2167	26	18	informed	informed	ADJ
easat-2167	26	19	decisions	decision	NOUN
easat-2167	26	20	regarding	regard	VERB
easat-2167	26	21	the	the	DET
easat-2167	26	22	timing	timing	NOUN
easat-2167	26	23	and	and	CCONJ
easat-2167	26	24	type	type	NOUN
easat-2167	26	25	of	of	ADP
easat-2167	26	26	interventions	intervention	NOUN
easat-2167	26	27	,	,	PUNCT
easat-2167	26	28	such	such	ADJ
easat-2167	26	29	as	as	ADP
easat-2167	26	30	pesticide	pesticide	NOUN
easat-2167	26	31	application	application	NOUN
easat-2167	26	32	,	,	PUNCT
easat-2167	26	33	to	to	PART
easat-2167	26	34	manage	manage	VERB
easat-2167	26	35	the	the	DET
easat-2167	26	36	disease	disease	NOUN
easat-2167	26	37	effectively	effectively	ADV
easat-2167	26	38	.	.	PUNCT
easat-2167	27	1	introduction	introduction	NOUN
easat-2167	27	2	to	to	ADP
easat-2167	27	3	techniques	technique	NOUN
easat-2167	27	4	:	:	PUNCT
easat-2167	27	5	several	several	ADJ
easat-2167	27	6	advanced	advanced	ADJ
easat-2167	27	7	techniques	technique	NOUN
easat-2167	27	8	have	have	AUX
easat-2167	27	9	been	be	AUX
easat-2167	27	10	developed	develop	VERB
easat-2167	27	11	for	for	ADP
easat-2167	27	12	leaf	leaf	NOUN
easat-2167	27	13	disease	disease	NOUN
easat-2167	27	14	image	image	NOUN
easat-2167	27	15	segmentation	segmentation	NOUN
easat-2167	27	16	,	,	PUNCT
easat-2167	27	17	leveraging	leverage	VERB
easat-2167	27	18	the	the	DET
easat-2167	27	19	power	power	NOUN
easat-2167	27	20	of	of	ADP
easat-2167	27	21	machine	machine	NOUN
easat-2167	27	22	learning	learning	NOUN
easat-2167	27	23	,	,	PUNCT
easat-2167	27	24	deep	deep	ADJ
easat-2167	27	25	learning	learning	NOUN
easat-2167	27	26	,	,	PUNCT
easat-2167	27	27	and	and	CCONJ
easat-2167	27	28	image	image	NOUN
easat-2167	27	29	processing	processing	NOUN
easat-2167	27	30	algorithms	algorithm	NOUN
easat-2167	27	31	.	.	PUNCT
easat-2167	28	1	traditional	traditional	ADJ
easat-2167	28	2	methods	method	NOUN
easat-2167	28	3	,	,	PUNCT
easat-2167	28	4	such	such	ADJ
easat-2167	28	5	as	as	ADP
easat-2167	28	6	thresholding	thresholding	NOUN
easat-2167	28	7	and	and	CCONJ
easat-2167	28	8	edge	edge	NOUN
easat-2167	28	9	detection	detection	NOUN
easat-2167	28	10	,	,	PUNCT
easat-2167	28	11	have	have	AUX
easat-2167	28	12	been	be	AUX
easat-2167	28	13	supplemented	supplement	VERB
easat-2167	28	14	by	by	ADP
easat-2167	28	15	more	more	ADV
easat-2167	28	16	sophisticated	sophisticated	ADJ
easat-2167	28	17	approaches	approach	NOUN
easat-2167	28	18	like	like	ADP
easat-2167	28	19	:	:	PUNCT
easat-2167	28	20	•	•	NUM
easat-2167	28	21	k	k	X
easat-2167	28	22	-	-	PUNCT
easat-2167	28	23	means	means	NOUN
easat-2167	28	24	clustering	clustering	NOUN
easat-2167	28	25	and	and	CCONJ
easat-2167	28	26	superpixel	superpixel	ADJ
easat-2167	28	27	segmentation	segmentation	NOUN
easat-2167	28	28	:	:	PUNCT
easat-2167	28	29	these	these	DET
easat-2167	28	30	methods	method	NOUN
easat-2167	28	31	group	group	NOUN
easat-2167	28	32	pixels	pixel	VERB
easat-2167	28	33	with	with	ADP
easat-2167	28	34	similar	similar	ADJ
easat-2167	28	35	features	feature	NOUN
easat-2167	28	36	,	,	PUNCT
easat-2167	28	37	allowing	allow	VERB
easat-2167	28	38	for	for	ADP
easat-2167	28	39	better	well	ADJ
easat-2167	28	40	differentiation	differentiation	NOUN
easat-2167	28	41	between	between	ADP
easat-2167	28	42	diseased	diseased	ADJ
easat-2167	28	43	and	and	CCONJ
easat-2167	28	44	healthy	healthy	ADJ
easat-2167	28	45	regions	region	NOUN
easat-2167	28	46	.	.	PUNCT
easat-2167	29	1	•	•	NUM
easat-2167	29	2	deep	deep	ADJ
easat-2167	29	3	learning	learning	NOUN
easat-2167	29	4	models	model	NOUN
easat-2167	29	5	:	:	PUNCT
easat-2167	29	6	convolutional	convolutional	ADJ
easat-2167	29	7	neural	neural	ADJ
easat-2167	29	8	networks	network	NOUN
easat-2167	29	9	(	(	PUNCT
easat-2167	29	10	cnns	cnns	PROPN
easat-2167	29	11	)	)	PUNCT
easat-2167	29	12	and	and	CCONJ
easat-2167	29	13	other	other	ADJ
easat-2167	29	14	deep	deep	ADJ
easat-2167	29	15	learning	learning	NOUN
easat-2167	29	16	architectures	architecture	NOUN
easat-2167	29	17	have	have	AUX
easat-2167	29	18	revolutionized	revolutionize	VERB
easat-2167	29	19	image	image	NOUN
easat-2167	29	20	segmentation	segmentation	NOUN
easat-2167	29	21	.	.	PUNCT
easat-2167	30	1	these	these	DET
easat-2167	30	2	models	model	NOUN
easat-2167	30	3	can	can	AUX
easat-2167	30	4	learn	learn	VERB
easat-2167	30	5	complex	complex	ADJ
easat-2167	30	6	patterns	pattern	NOUN
easat-2167	30	7	and	and	CCONJ
easat-2167	30	8	features	feature	NOUN
easat-2167	30	9	from	from	ADP
easat-2167	30	10	large	large	ADJ
easat-2167	30	11	datasets	dataset	NOUN
easat-2167	30	12	,	,	PUNCT
easat-2167	30	13	improving	improve	VERB
easat-2167	30	14	the	the	DET
easat-2167	30	15	accuracy	accuracy	NOUN
easat-2167	30	16	of	of	ADP
easat-2167	30	17	disease	disease	NOUN
easat-2167	30	18	detection	detection	NOUN
easat-2167	30	19	even	even	ADV
easat-2167	30	20	in	in	ADP
easat-2167	30	21	challenging	challenging	ADJ
easat-2167	30	22	conditions	condition	NOUN
easat-2167	30	23	.	.	PUNCT
easat-2167	31	1	•	•	NUM
easat-2167	31	2	hybrid	hybrid	ADJ
easat-2167	31	3	techniques	technique	NOUN
easat-2167	31	4	:	:	PUNCT
easat-2167	31	5	combining	combine	VERB
easat-2167	31	6	different	different	ADJ
easat-2167	31	7	segmentation	segmentation	NOUN
easat-2167	31	8	methods	method	NOUN
easat-2167	31	9	,	,	PUNCT
easat-2167	31	10	such	such	ADJ
easat-2167	31	11	as	as	ADP
easat-2167	31	12	using	use	VERB
easat-2167	31	13	both	both	CCONJ
easat-2167	31	14	color	color	NOUN
easat-2167	31	15	and	and	CCONJ
easat-2167	31	16	texture	texture	ADJ
easat-2167	31	17	analysis	analysis	NOUN
easat-2167	31	18	,	,	PUNCT
easat-2167	31	19	can	can	AUX
easat-2167	31	20	enhance	enhance	VERB
easat-2167	31	21	the	the	DET
easat-2167	31	22	robustness	robustness	NOUN
easat-2167	31	23	of	of	ADP
easat-2167	31	24	the	the	DET
easat-2167	31	25	segmentation	segmentation	NOUN
easat-2167	31	26	process	process	NOUN
easat-2167	31	27	,	,	PUNCT
easat-2167	31	28	particularly	particularly	ADV
easat-2167	31	29	in	in	ADP
easat-2167	31	30	diverse	diverse	ADJ
easat-2167	31	31	field	field	NOUN
easat-2167	31	32	conditions	condition	NOUN
easat-2167	31	33	.	.	PUNCT
easat-2167	32	1	2	2	X
easat-2167	32	2	.	.	X
easat-2167	32	3	literature	literature	NOUN
easat-2167	32	4	survey	survey	NOUN
easat-2167	32	5	on	on	ADP
easat-2167	32	6	rice	rice	NOUN
easat-2167	32	7	diseases	disease	NOUN
easat-2167	32	8	image	image	NOUN
easat-2167	32	9	segmentation	segmentation	VERB
easat-2167	32	10	the	the	DET
easat-2167	32	11	identification	identification	NOUN
easat-2167	32	12	and	and	CCONJ
easat-2167	32	13	segmentation	segmentation	NOUN
easat-2167	32	14	of	of	ADP
easat-2167	32	15	rice	rice	NOUN
easat-2167	32	16	diseases	disease	NOUN
easat-2167	32	17	using	use	VERB
easat-2167	32	18	image	image	NOUN
easat-2167	32	19	processing	processing	NOUN
easat-2167	32	20	and	and	CCONJ
easat-2167	32	21	deep	deep	ADJ
easat-2167	32	22	learning	learning	NOUN
easat-2167	32	23	techniques	technique	NOUN
easat-2167	32	24	have	have	AUX
easat-2167	32	25	been	be	AUX
easat-2167	32	26	a	a	DET
easat-2167	32	27	significant	significant	ADJ
easat-2167	32	28	focus	focus	NOUN
easat-2167	32	29	in	in	ADP
easat-2167	32	30	agricultural	agricultural	ADJ
easat-2167	32	31	research	research	NOUN
easat-2167	32	32	over	over	ADP
easat-2167	32	33	the	the	DET
easat-2167	32	34	past	past	ADJ
easat-2167	32	35	decade	decade	NOUN
easat-2167	32	36	.	.	PUNCT
easat-2167	33	1	this	this	DET
easat-2167	33	2	survey	survey	NOUN
easat-2167	33	3	summarizes	summarize	VERB
easat-2167	33	4	key	key	ADJ
easat-2167	33	5	studies	study	NOUN
easat-2167	33	6	,	,	PUNCT
easat-2167	33	7	focusing	focus	VERB
easat-2167	33	8	on	on	ADP
easat-2167	33	9	methods	method	NOUN
easat-2167	33	10	used	use	VERB
easat-2167	33	11	,	,	PUNCT
easat-2167	33	12	findings	finding	NOUN
easat-2167	33	13	,	,	PUNCT
easat-2167	33	14	and	and	CCONJ
easat-2167	33	15	their	their	PRON
easat-2167	33	16	contributions	contribution	NOUN
easat-2167	33	17	to	to	ADP
easat-2167	33	18	the	the	DET
easat-2167	33	19	field	field	NOUN
easat-2167	33	20	.	.	PUNCT
easat-2167	34	1	2.1	2.1	NUM
easat-2167	34	2	.	.	PUNCT
easat-2167	35	1	key	key	ADJ
easat-2167	35	2	findings	finding	NOUN
easat-2167	35	3	from	from	ADP
easat-2167	35	4	major	major	ADJ
easat-2167	35	5	papers	paper	NOUN
easat-2167	35	6	automatic	automatic	ADJ
easat-2167	35	7	rice	rice	NOUN
easat-2167	35	8	leaf	leaf	NOUN
easat-2167	35	9	disease	disease	NOUN
easat-2167	35	10	segmentation	segmentation	NOUN
easat-2167	35	11	:	:	PUNCT
easat-2167	35	12	a	a	DET
easat-2167	35	13	study	study	NOUN
easat-2167	35	14	proposed	propose	VERB
easat-2167	35	15	an	an	DET
easat-2167	35	16	algorithm	algorithm	NOUN
easat-2167	35	17	for	for	ADP
easat-2167	35	18	automatic	automatic	ADJ
easat-2167	35	19	detection	detection	NOUN
easat-2167	35	20	and	and	CCONJ
easat-2167	35	21	segmentation	segmentation	NOUN
easat-2167	35	22	of	of	ADP
easat-2167	35	23	bacterial	bacterial	ADJ
easat-2167	35	24	leaf	leaf	NOUN
easat-2167	35	25	blight	blight	NOUN
easat-2167	35	26	and	and	CCONJ
easat-2167	35	27	brown	brown	ADJ
easat-2167	35	28	spot	spot	NOUN
easat-2167	35	29	in	in	ADP
easat-2167	35	30	rice	rice	NOUN
easat-2167	35	31	leaves	leave	VERB
easat-2167	35	32	under	under	ADP
easat-2167	35	33	varying	vary	VERB
easat-2167	35	34	environmental	environmental	ADJ
easat-2167	35	35	conditions	condition	NOUN
easat-2167	35	36	.	.	PUNCT
easat-2167	36	1	the	the	DET
easat-2167	36	2	method	method	NOUN
easat-2167	36	3	demonstrated	demonstrate	VERB
easat-2167	36	4	robust	robust	ADJ
easat-2167	36	5	performance	performance	NOUN
easat-2167	36	6	in	in	ADP
easat-2167	36	7	diverse	diverse	ADJ
easat-2167	36	8	conditions	condition	NOUN
easat-2167	36	9	[	[	X
easat-2167	36	10	1	1	NUM
easat-2167	36	11	]	]	PUNCT
easat-2167	36	12	.	.	PUNCT
easat-2167	37	1	image	image	NOUN
easat-2167	37	2	processing	processing	NOUN
easat-2167	37	3	techniques	technique	NOUN
easat-2167	37	4	survey	survey	NOUN
easat-2167	37	5	:	:	PUNCT
easat-2167	37	6	a	a	DET
easat-2167	37	7	comprehensive	comprehensive	ADJ
easat-2167	37	8	survey	survey	NOUN
easat-2167	37	9	reviewed	review	VERB
easat-2167	37	10	various	various	ADJ
easat-2167	37	11	image	image	NOUN
easat-2167	37	12	segmentation	segmentation	NOUN
easat-2167	37	13	and	and	CCONJ
easat-2167	37	14	classification	classification	NOUN
easat-2167	37	15	techniques	technique	NOUN
easat-2167	37	16	for	for	ADP
easat-2167	37	17	diagnosing	diagnose	VERB
easat-2167	37	18	rice	rice	NOUN
easat-2167	37	19	plant	plant	NOUN
easat-2167	37	20	diseases	disease	NOUN
easat-2167	37	21	.	.	PUNCT
easat-2167	38	1	it	it	PRON
easat-2167	38	2	highlighted	highlight	VERB
easat-2167	38	3	achievements	achievement	NOUN
easat-2167	38	4	and	and	CCONJ
easat-2167	38	5	limitations	limitation	NOUN
easat-2167	38	6	in	in	ADP
easat-2167	38	7	the	the	DET
easat-2167	38	8	field	field	NOUN
easat-2167	38	9	,	,	PUNCT
easat-2167	38	10	suggesting	suggest	VERB
easat-2167	38	11	future	future	ADJ
easat-2167	38	12	research	research	NOUN
easat-2167	38	13	directions	direction	NOUN
easat-2167	38	14	[	[	X
easat-2167	38	15	2	2	NUM
easat-2167	38	16	]	]	PUNCT
easat-2167	38	17	.	.	PUNCT
easat-2167	39	1	deep	deep	ADJ
easat-2167	39	2	learning	learning	NOUN
easat-2167	39	3	for	for	ADP
easat-2167	39	4	rice	rice	NOUN
easat-2167	39	5	disease	disease	NOUN
easat-2167	39	6	detection	detection	NOUN
easat-2167	39	7	:	:	PUNCT
easat-2167	39	8	an	an	DET
easat-2167	39	9	approach	approach	NOUN
easat-2167	39	10	combining	combine	VERB
easat-2167	39	11	deep	deep	ADJ
easat-2167	39	12	learning	learning	NOUN
easat-2167	39	13	with	with	ADP
easat-2167	39	14	level	level	NOUN
easat-2167	39	15	set	set	NOUN
easat-2167	39	16	methods	method	NOUN
easat-2167	39	17	was	be	AUX
easat-2167	39	18	proposed	propose	VERB
easat-2167	39	19	to	to	PART
easat-2167	39	20	improve	improve	VERB
easat-2167	39	21	the	the	DET
easat-2167	39	22	detection	detection	NOUN
easat-2167	39	23	and	and	CCONJ
easat-2167	39	24	classification	classification	NOUN
easat-2167	39	25	of	of	ADP
easat-2167	39	26	rice	rice	NOUN
easat-2167	39	27	diseases	disease	NOUN
easat-2167	39	28	.	.	PUNCT
easat-2167	40	1	the	the	DET
easat-2167	40	2	method	method	NOUN
easat-2167	40	3	achieved	achieve	VERB
easat-2167	40	4	high	high	ADJ
easat-2167	40	5	accuracy	accuracy	NOUN
easat-2167	40	6	even	even	ADV
easat-2167	40	7	with	with	ADP
easat-2167	40	8	limited	limited	ADJ
easat-2167	40	9	labeled	label	VERB
easat-2167	40	10	data	datum	NOUN
easat-2167	40	11	[	[	X
easat-2167	40	12	3	3	NUM
easat-2167	40	13	]	]	PUNCT
easat-2167	40	14	.	.	PUNCT
easat-2167	41	1	bi	bi	ADJ
easat-2167	41	2	-	-	NOUN
easat-2167	41	3	level	level	NOUN
easat-2167	41	4	thresholding	thresholding	NOUN
easat-2167	41	5	for	for	ADP
easat-2167	41	6	disease	disease	NOUN
easat-2167	41	7	diagnosis	diagnosis	NOUN
easat-2167	41	8	:	:	PUNCT
easat-2167	41	9	a	a	DET
easat-2167	41	10	portable	portable	ADJ
easat-2167	41	11	tool	tool	NOUN
easat-2167	41	12	using	use	VERB
easat-2167	41	13	bi	bi	ADJ
easat-2167	41	14	-	-	ADJ
easat-2167	41	15	level	level	ADJ
easat-2167	41	16	thresholding	thresholding	NOUN
easat-2167	41	17	was	be	AUX
easat-2167	41	18	developed	develop	VERB
easat-2167	41	19	to	to	PART
easat-2167	41	20	predict	predict	VERB
easat-2167	41	21	rice	rice	NOUN
easat-2167	41	22	diseases	disease	NOUN
easat-2167	41	23	.	.	PUNCT
easat-2167	42	1	the	the	DET
easat-2167	42	2	method	method	NOUN
easat-2167	42	3	achieved	achieve	VERB
easat-2167	42	4	better	well	ADJ
easat-2167	42	5	accuracy	accuracy	NOUN
easat-2167	42	6	than	than	ADP
easat-2167	42	7	traditional	traditional	ADJ
easat-2167	42	8	methods	method	NOUN
easat-2167	42	9	by	by	ADP
easat-2167	42	10	effectively	effectively	ADV
easat-2167	42	11	segmenting	segment	VERB
easat-2167	42	12	regions	region	NOUN
easat-2167	42	13	of	of	ADP
easat-2167	42	14	interest	interest	NOUN
easat-2167	42	15	[	[	X
easat-2167	42	16	4	4	NUM
easat-2167	42	17	]	]	PUNCT
easat-2167	42	18	.	.	PUNCT
easat-2167	43	1	lesion	lesion	NOUN
easat-2167	43	2	segmentation	segmentation	NOUN
easat-2167	43	3	with	with	ADP
easat-2167	43	4	random	random	ADJ
easat-2167	43	5	forest	forest	NOUN
easat-2167	43	6	:	:	PUNCT
easat-2167	43	7	this	this	DET
easat-2167	43	8	study	study	NOUN
easat-2167	43	9	presented	present	VERB
easat-2167	43	10	a	a	DET
easat-2167	43	11	method	method	NOUN
easat-2167	43	12	combining	combine	VERB
easat-2167	43	13	superpixel	superpixel	ADJ
easat-2167	43	14	segmentation	segmentation	NOUN
easat-2167	43	15	and	and	CCONJ
easat-2167	43	16	random	random	ADJ
easat-2167	43	17	forest	forest	NOUN
easat-2167	43	18	classifiers	classifier	NOUN
easat-2167	43	19	to	to	PART
easat-2167	43	20	segment	segment	VERB
easat-2167	43	21	lesions	lesion	NOUN
easat-2167	43	22	from	from	ADP
easat-2167	43	23	rice	rice	NOUN
easat-2167	43	24	leaf	leaf	NOUN
easat-2167	43	25	blast	blast	NOUN
easat-2167	43	26	images	image	NOUN
easat-2167	43	27	,	,	PUNCT
easat-2167	43	28	achieving	achieve	VERB
easat-2167	43	29	good	good	ADJ
easat-2167	43	30	segmentation	segmentation	NOUN
easat-2167	43	31	performance	performance	NOUN
easat-2167	43	32	[	[	X
easat-2167	43	33	5	5	NUM
easat-2167	43	34	]	]	PUNCT
easat-2167	43	35	.	.	PUNCT
easat-2167	44	1	comparison	comparison	NOUN
easat-2167	44	2	of	of	ADP
easat-2167	44	3	segmentation	segmentation	NOUN
easat-2167	44	4	methods	method	NOUN
easat-2167	44	5	:	:	PUNCT
easat-2167	44	6	various	various	ADJ
easat-2167	44	7	image	image	NOUN
easat-2167	44	8	segmentation	segmentation	NOUN
easat-2167	44	9	algorithms	algorithm	NOUN
easat-2167	44	10	were	be	AUX
easat-2167	44	11	compared	compare	VERB
easat-2167	44	12	to	to	PART
easat-2167	44	13	identify	identify	VERB
easat-2167	44	14	the	the	DET
easat-2167	44	15	most	most	ADV
easat-2167	44	16	effective	effective	ADJ
easat-2167	44	17	method	method	NOUN
easat-2167	44	18	for	for	ADP
easat-2167	44	19	segmenting	segment	VERB
easat-2167	44	20	diseased	diseased	ADJ
easat-2167	44	21	rice	rice	NOUN
easat-2167	44	22	leaves	leave	NOUN
easat-2167	44	23	,	,	PUNCT
easat-2167	44	24	emphasizing	emphasize	VERB
easat-2167	44	25	the	the	DET
easat-2167	44	26	need	need	NOUN
easat-2167	44	27	for	for	ADP
easat-2167	44	28	a	a	DET
easat-2167	44	29	unique	unique	ADJ
easat-2167	44	30	segmentation	segmentation	NOUN
easat-2167	44	31	method	method	NOUN
easat-2167	44	32	[	[	X
easat-2167	44	33	6	6	NUM
easat-2167	44	34	]	]	PUNCT
easat-2167	44	35	.	.	PUNCT
easat-2167	45	1	machine	machine	NOUN
easat-2167	45	2	vision	vision	NOUN
easat-2167	45	3	for	for	ADP
easat-2167	45	4	disease	disease	NOUN
easat-2167	45	5	spots	spot	VERB
easat-2167	45	6	extraction	extraction	NOUN
easat-2167	45	7	:	:	PUNCT
easat-2167	45	8	a	a	DET
easat-2167	45	9	method	method	NOUN
easat-2167	45	10	based	base	VERB
easat-2167	45	11	on	on	ADP
easat-2167	45	12	machine	machine	NOUN
easat-2167	45	13	vision	vision	NOUN
easat-2167	45	14	was	be	AUX
easat-2167	45	15	developed	develop	VERB
easat-2167	45	16	to	to	PART
easat-2167	45	17	segment	segment	VERB
easat-2167	45	18	and	and	CCONJ
easat-2167	45	19	extract	extract	VERB
easat-2167	45	20	disease	disease	NOUN
easat-2167	45	21	spots	spot	NOUN
easat-2167	45	22	in	in	ADP
easat-2167	45	23	rice	rice	NOUN
easat-2167	45	24	leaves	leave	NOUN
easat-2167	45	25	.	.	PUNCT
easat-2167	46	1	the	the	DET
easat-2167	46	2	technique	technique	NOUN
easat-2167	46	3	involved	involve	VERB
easat-2167	46	4	gray	gray	ADJ
easat-2167	46	5	-	-	PUNCT
easat-2167	46	6	level	level	NOUN
easat-2167	46	7	transformation	transformation	NOUN
easat-2167	46	8	followed	follow	VERB
easat-2167	46	9	by	by	ADP
easat-2167	46	10	segmentation	segmentation	NOUN
easat-2167	46	11	[	[	X
easat-2167	46	12	7	7	NUM
easat-2167	46	13	]	]	PUNCT
easat-2167	46	14	.	.	PUNCT
easat-2167	47	1	detection	detection	NOUN
easat-2167	47	2	using	use	VERB
easat-2167	47	3	svm	svm	PROPN
easat-2167	47	4	and	and	CCONJ
easat-2167	47	5	random	random	ADJ
easat-2167	47	6	forest	forest	NOUN
easat-2167	47	7	:	:	PUNCT
easat-2167	47	8	an	an	DET
easat-2167	47	9	automated	automate	VERB
easat-2167	47	10	method	method	NOUN
easat-2167	47	11	for	for	ADP
easat-2167	47	12	detecting	detect	VERB
easat-2167	47	13	rice	rice	NOUN
easat-2167	47	14	diseases	disease	NOUN
easat-2167	47	15	using	use	VERB
easat-2167	47	16	795	795	NUM
easat-2167	47	17	edelweiss	edelweiss	PROPN
easat-2167	47	18	applied	apply	VERB
easat-2167	47	19	science	science	NOUN
easat-2167	47	20	and	and	CCONJ
easat-2167	47	21	technology	technology	NOUN
easat-2167	47	22	issn	issn	PROPN
easat-2167	47	23	:	:	PUNCT
easat-2167	47	24	2576	2576	NUM
easat-2167	47	25	-	-	SYM
easat-2167	47	26	8484	8484	NUM
easat-2167	47	27	vol	vol	NOUN
easat-2167	47	28	.	.	PROPN
easat-2167	47	29	8	8	NUM
easat-2167	47	30	,	,	PUNCT
easat-2167	47	31	no	no	INTJ
easat-2167	47	32	.	.	NOUN
easat-2167	48	1	6	6	NUM
easat-2167	48	2	:	:	PUNCT
easat-2167	48	3	793	793	NUM
easat-2167	48	4	-	-	NUM
easat-2167	48	5	813	813	NUM
easat-2167	48	6	,	,	PUNCT
easat-2167	48	7	2024	2024	NUM
easat-2167	48	8	doi	doi	NOUN
easat-2167	48	9	:	:	PUNCT
easat-2167	48	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	48	11	©	©	NOUN
easat-2167	48	12	2024	2024	NUM
easat-2167	48	13	by	by	ADP
easat-2167	48	14	the	the	DET
easat-2167	48	15	authors	author	NOUN
easat-2167	48	16	;	;	PUNCT
easat-2167	48	17	licensee	licensee	PROPN
easat-2167	48	18	learning	learn	VERB
easat-2167	48	19	gate	gate	NOUN
easat-2167	48	20	svm	svm	PROPN
easat-2167	48	21	and	and	CCONJ
easat-2167	48	22	random	random	ADJ
easat-2167	48	23	forest	forest	NOUN
easat-2167	48	24	classifiers	classifier	NOUN
easat-2167	48	25	was	be	AUX
easat-2167	48	26	proposed	propose	VERB
easat-2167	48	27	.	.	PUNCT
easat-2167	49	1	the	the	DET
easat-2167	49	2	method	method	NOUN
easat-2167	49	3	showed	show	VERB
easat-2167	49	4	superior	superior	ADJ
easat-2167	49	5	performance	performance	NOUN
easat-2167	49	6	in	in	ADP
easat-2167	49	7	identifying	identify	VERB
easat-2167	49	8	diseases	disease	NOUN
easat-2167	49	9	compared	compare	VERB
easat-2167	49	10	to	to	ADP
easat-2167	49	11	traditional	traditional	ADJ
easat-2167	49	12	approaches	approach	NOUN
easat-2167	49	13	[	[	X
easat-2167	49	14	8	8	NUM
easat-2167	49	15	]	]	PUNCT
easat-2167	49	16	.	.	PUNCT
easat-2167	50	1	copy	copy	PROPN
easat-2167	50	2	paste	paste	NOUN
easat-2167	50	3	and	and	CCONJ
easat-2167	50	4	semantic	semantic	ADJ
easat-2167	50	5	segmentation	segmentation	NOUN
easat-2167	50	6	:	:	PUNCT
easat-2167	50	7	a	a	DET
easat-2167	50	8	lightweight	lightweight	ADJ
easat-2167	50	9	network	network	NOUN
easat-2167	50	10	using	use	VERB
easat-2167	50	11	copy	copy	NOUN
easat-2167	50	12	paste	paste	NOUN
easat-2167	50	13	and	and	CCONJ
easat-2167	50	14	semantic	semantic	ADJ
easat-2167	50	15	segmentation	segmentation	NOUN
easat-2167	50	16	was	be	AUX
easat-2167	50	17	introduced	introduce	VERB
easat-2167	50	18	for	for	ADP
easat-2167	50	19	accurate	accurate	ADJ
easat-2167	50	20	disease	disease	NOUN
easat-2167	50	21	region	region	NOUN
easat-2167	50	22	segmentation	segmentation	NOUN
easat-2167	50	23	and	and	CCONJ
easat-2167	50	24	severity	severity	NOUN
easat-2167	50	25	assessment	assessment	NOUN
easat-2167	50	26	in	in	ADP
easat-2167	50	27	rice	rice	NOUN
easat-2167	50	28	leaves	leave	NOUN
easat-2167	50	29	[	[	X
easat-2167	50	30	9	9	NUM
easat-2167	50	31	]	]	PUNCT
easat-2167	50	32	.	.	PUNCT
easat-2167	51	1	texture	texture	NOUN
easat-2167	51	2	and	and	CCONJ
easat-2167	51	3	deep	deep	ADJ
easat-2167	51	4	features	feature	NOUN
easat-2167	51	5	for	for	ADP
easat-2167	51	6	disease	disease	NOUN
easat-2167	51	7	detection	detection	NOUN
easat-2167	51	8	:	:	PUNCT
easat-2167	51	9	a	a	DET
easat-2167	51	10	deep	deep	ADJ
easat-2167	51	11	learning	learning	NOUN
easat-2167	51	12	model	model	NOUN
easat-2167	51	13	using	use	VERB
easat-2167	51	14	segnet	segnet	NOUN
easat-2167	51	15	for	for	ADP
easat-2167	51	16	segmentation	segmentation	NOUN
easat-2167	51	17	and	and	CCONJ
easat-2167	51	18	deep	deep	ADJ
easat-2167	51	19	recurrent	recurrent	ADJ
easat-2167	51	20	neural	neural	ADJ
easat-2167	51	21	networks	network	NOUN
easat-2167	51	22	(	(	PUNCT
easat-2167	51	23	deep	deep	ADJ
easat-2167	51	24	rnn	rnn	PROPN
easat-2167	51	25	)	)	PUNCT
easat-2167	51	26	for	for	ADP
easat-2167	51	27	detection	detection	NOUN
easat-2167	51	28	was	be	AUX
easat-2167	51	29	proposed	propose	VERB
easat-2167	51	30	.	.	PUNCT
easat-2167	52	1	the	the	DET
easat-2167	52	2	model	model	NOUN
easat-2167	52	3	achieved	achieve	VERB
easat-2167	52	4	high	high	ADJ
easat-2167	52	5	accuracy	accuracy	NOUN
easat-2167	52	6	and	and	CCONJ
easat-2167	52	7	specificity	specificity	NOUN
easat-2167	52	8	[	[	X
easat-2167	52	9	10	10	NUM
easat-2167	52	10	]	]	PUNCT
easat-2167	52	11	.	.	PUNCT
easat-2167	53	1	transfer	transfer	NOUN
easat-2167	53	2	learning	learn	VERB
easat-2167	53	3	with	with	ADP
easat-2167	53	4	cnn	cnn	PROPN
easat-2167	53	5	:	:	PUNCT
easat-2167	53	6	a	a	DET
easat-2167	53	7	cnn	cnn	PROPN
easat-2167	53	8	-	-	PUNCT
easat-2167	53	9	based	base	VERB
easat-2167	53	10	model	model	NOUN
easat-2167	53	11	using	use	VERB
easat-2167	53	12	transfer	transfer	NOUN
easat-2167	53	13	learning	learning	NOUN
easat-2167	53	14	was	be	AUX
easat-2167	53	15	developed	develop	VERB
easat-2167	53	16	for	for	ADP
easat-2167	53	17	predicting	predict	VERB
easat-2167	53	18	rice	rice	NOUN
easat-2167	53	19	leaf	leaf	NOUN
easat-2167	53	20	diseases	disease	NOUN
easat-2167	53	21	,	,	PUNCT
easat-2167	53	22	achieving	achieve	VERB
easat-2167	53	23	95.67	95.67	NUM
easat-2167	53	24	%	%	NOUN
easat-2167	53	25	accuracy	accuracy	NOUN
easat-2167	54	1	[	[	X
easat-2167	54	2	11	11	NUM
easat-2167	54	3	]	]	PUNCT
easat-2167	54	4	.	.	PUNCT
easat-2167	55	1	otsu	otsu	PROPN
easat-2167	55	2	’s	’s	PART
easat-2167	55	3	method	method	NOUN
easat-2167	55	4	for	for	ADP
easat-2167	55	5	segmentation	segmentation	NOUN
easat-2167	55	6	:	:	PUNCT
easat-2167	55	7	a	a	DET
easat-2167	55	8	method	method	NOUN
easat-2167	55	9	using	use	VERB
easat-2167	55	10	otsu	otsu	NOUN
easat-2167	55	11	’s	’s	PART
easat-2167	55	12	thresholding	thresholding	NOUN
easat-2167	55	13	for	for	ADP
easat-2167	55	14	segmentation	segmentation	NOUN
easat-2167	55	15	combined	combine	VERB
easat-2167	55	16	with	with	ADP
easat-2167	55	17	local	local	ADJ
easat-2167	55	18	binary	binary	ADJ
easat-2167	55	19	patterns	pattern	NOUN
easat-2167	55	20	(	(	PUNCT
easat-2167	55	21	lbp	lbp	PROPN
easat-2167	55	22	)	)	PUNCT
easat-2167	55	23	and	and	CCONJ
easat-2167	55	24	histogram	histogram	NOUN
easat-2167	55	25	of	of	ADP
easat-2167	55	26	oriented	orient	VERB
easat-2167	55	27	gradients	gradient	NOUN
easat-2167	55	28	(	(	PUNCT
easat-2167	55	29	hog	hog	PROPN
easat-2167	55	30	)	)	PUNCT
easat-2167	55	31	achieved	achieve	VERB
easat-2167	55	32	94.6	94.6	NUM
easat-2167	55	33	%	%	NOUN
easat-2167	55	34	accuracy	accuracy	NOUN
easat-2167	55	35	for	for	ADP
easat-2167	55	36	rice	rice	NOUN
easat-2167	55	37	disease	disease	NOUN
easat-2167	55	38	classification	classification	NOUN
easat-2167	55	39	[	[	X
easat-2167	55	40	12	12	NUM
easat-2167	55	41	]	]	PUNCT
easat-2167	55	42	.	.	PUNCT
easat-2167	56	1	fusion	fusion	NOUN
easat-2167	56	2	of	of	ADP
easat-2167	56	3	fcm	fcm	NOUN
easat-2167	56	4	-	-	NOUN
easat-2167	56	5	km	km	NOUN
easat-2167	56	6	and	and	CCONJ
easat-2167	56	7	faster	fast	ADJ
easat-2167	56	8	r	r	NOUN
easat-2167	56	9	-	-	PUNCT
easat-2167	56	10	cnn	cnn	NOUN
easat-2167	56	11	:	:	PUNCT
easat-2167	56	12	a	a	DET
easat-2167	56	13	fusion	fusion	NOUN
easat-2167	56	14	method	method	NOUN
easat-2167	56	15	using	use	VERB
easat-2167	56	16	fcm	fcm	ADJ
easat-2167	56	17	-	-	ADJ
easat-2167	56	18	km	km	NOUN
easat-2167	56	19	clustering	clustering	NOUN
easat-2167	56	20	and	and	CCONJ
easat-2167	56	21	faster	fast	ADJ
easat-2167	56	22	rcnn	rcnn	NOUN
easat-2167	56	23	for	for	ADP
easat-2167	56	24	rapid	rapid	ADJ
easat-2167	56	25	detection	detection	NOUN
easat-2167	56	26	of	of	ADP
easat-2167	56	27	rice	rice	NOUN
easat-2167	56	28	diseases	disease	NOUN
easat-2167	56	29	demonstrated	demonstrate	VERB
easat-2167	56	30	high	high	ADJ
easat-2167	56	31	accuracy	accuracy	NOUN
easat-2167	56	32	and	and	CCONJ
easat-2167	56	33	reduced	reduced	ADJ
easat-2167	56	34	detection	detection	NOUN
easat-2167	56	35	time	time	NOUN
easat-2167	56	36	[	[	X
easat-2167	56	37	13	13	NUM
easat-2167	56	38	]	]	PUNCT
easat-2167	56	39	.	.	PUNCT
easat-2167	57	1	deep	deep	ADJ
easat-2167	57	2	cnn	cnn	PROPN
easat-2167	57	3	for	for	ADP
easat-2167	57	4	disease	disease	NOUN
easat-2167	57	5	identification	identification	NOUN
easat-2167	57	6	:	:	PUNCT
easat-2167	57	7	a	a	DET
easat-2167	57	8	cnn	cnn	PROPN
easat-2167	57	9	-	-	PUNCT
easat-2167	57	10	based	base	VERB
easat-2167	57	11	method	method	NOUN
easat-2167	57	12	achieved	achieve	VERB
easat-2167	57	13	95.48	95.48	NUM
easat-2167	57	14	%	%	NOUN
easat-2167	57	15	accuracy	accuracy	NOUN
easat-2167	57	16	in	in	ADP
easat-2167	57	17	identifying	identify	VERB
easat-2167	57	18	10	10	NUM
easat-2167	57	19	common	common	ADJ
easat-2167	57	20	rice	rice	NOUN
easat-2167	57	21	diseases	disease	NOUN
easat-2167	57	22	,	,	PUNCT
easat-2167	57	23	significantly	significantly	ADV
easat-2167	57	24	outperforming	outperform	VERB
easat-2167	57	25	traditional	traditional	ADJ
easat-2167	57	26	machine	machine	NOUN
easat-2167	57	27	learning	learning	NOUN
easat-2167	57	28	models	model	NOUN
easat-2167	57	29	[	[	X
easat-2167	57	30	14	14	NUM
easat-2167	57	31	]	]	PUNCT
easat-2167	57	32	.	.	PUNCT
easat-2167	58	1	attention	attention	NOUN
easat-2167	58	2	networks	network	NOUN
easat-2167	58	3	for	for	ADP
easat-2167	58	4	rice	rice	NOUN
easat-2167	58	5	disease	disease	NOUN
easat-2167	58	6	detection	detection	NOUN
easat-2167	58	7	:	:	PUNCT
easat-2167	58	8	lightweight	lightweight	ADJ
easat-2167	58	9	attention	attention	NOUN
easat-2167	58	10	networks	network	NOUN
easat-2167	58	11	based	base	VERB
easat-2167	58	12	on	on	ADP
easat-2167	58	13	mobilenet	mobilenet	NOUN
easat-2167	58	14	-	-	PUNCT
easat-2167	58	15	v2	v2	NOUN
easat-2167	58	16	achieved	achieve	VERB
easat-2167	58	17	99.67	99.67	NUM
easat-2167	58	18	%	%	NOUN
easat-2167	58	19	accuracy	accuracy	NOUN
easat-2167	58	20	in	in	ADP
easat-2167	58	21	identifying	identify	VERB
easat-2167	58	22	rice	rice	NOUN
easat-2167	58	23	plant	plant	NOUN
easat-2167	58	24	diseases	disease	NOUN
easat-2167	58	25	[	[	X
easat-2167	58	26	15	15	NUM
easat-2167	58	27	]	]	PUNCT
easat-2167	58	28	.	.	PUNCT
easat-2167	59	1	backpropagation	backpropagation	NOUN
easat-2167	59	2	neural	neural	ADJ
easat-2167	59	3	network	network	NOUN
easat-2167	59	4	:	:	PUNCT
easat-2167	59	5	a	a	DET
easat-2167	59	6	method	method	NOUN
easat-2167	59	7	using	use	VERB
easat-2167	59	8	backpropagation	backpropagation	NOUN
easat-2167	59	9	neural	neural	ADJ
easat-2167	59	10	networks	network	NOUN
easat-2167	59	11	for	for	ADP
easat-2167	59	12	detecting	detect	VERB
easat-2167	59	13	three	three	NUM
easat-2167	59	14	common	common	ADJ
easat-2167	59	15	rice	rice	NOUN
easat-2167	59	16	diseases	disease	NOUN
easat-2167	59	17	achieved	achieve	VERB
easat-2167	59	18	100	100	NUM
easat-2167	59	19	%	%	NOUN
easat-2167	59	20	accuracy	accuracy	NOUN
easat-2167	60	1	[	[	X
easat-2167	60	2	16	16	NUM
easat-2167	60	3	]	]	PUNCT
easat-2167	60	4	.	.	PUNCT
easat-2167	61	1	survey	survey	NOUN
easat-2167	61	2	on	on	ADP
easat-2167	61	3	rice	rice	NOUN
easat-2167	61	4	disease	disease	NOUN
easat-2167	61	5	detection	detection	NOUN
easat-2167	61	6	:	:	PUNCT
easat-2167	61	7	a	a	DET
easat-2167	61	8	survey	survey	NOUN
easat-2167	61	9	reviewed	review	VERB
easat-2167	61	10	various	various	ADJ
easat-2167	61	11	image	image	NOUN
easat-2167	61	12	processing	processing	NOUN
easat-2167	61	13	and	and	CCONJ
easat-2167	61	14	machine	machine	NOUN
easat-2167	61	15	learning	learn	VERB
easat-2167	61	16	techniques	technique	NOUN
easat-2167	61	17	for	for	ADP
easat-2167	61	18	detecting	detect	VERB
easat-2167	61	19	rice	rice	NOUN
easat-2167	61	20	diseases	disease	NOUN
easat-2167	61	21	,	,	PUNCT
easat-2167	61	22	highlighting	highlight	VERB
easat-2167	61	23	strengths	strength	NOUN
easat-2167	61	24	and	and	CCONJ
easat-2167	61	25	limitations	limitation	NOUN
easat-2167	61	26	[	[	X
easat-2167	61	27	17	17	NUM
easat-2167	61	28	]	]	PUNCT
easat-2167	61	29	.	.	PUNCT
easat-2167	62	1	rice	rice	NOUN
easat-2167	62	2	-	-	PUNCT
easat-2167	62	3	fusion	fusion	NOUN
easat-2167	62	4	framework	framework	NOUN
easat-2167	62	5	:	:	PUNCT
easat-2167	62	6	a	a	DET
easat-2167	62	7	multimodality	multimodality	NOUN
easat-2167	62	8	data	datum	NOUN
easat-2167	62	9	fusion	fusion	NOUN
easat-2167	62	10	framework	framework	NOUN
easat-2167	62	11	combining	combine	VERB
easat-2167	62	12	agro	agro	ADJ
easat-2167	62	13	-	-	PUNCT
easat-2167	62	14	meteorological	meteorological	ADJ
easat-2167	62	15	data	datum	NOUN
easat-2167	62	16	and	and	CCONJ
easat-2167	62	17	image	image	NOUN
easat-2167	62	18	features	feature	NOUN
easat-2167	62	19	achieved	achieve	VERB
easat-2167	62	20	95.31	95.31	NUM
easat-2167	62	21	%	%	NOUN
easat-2167	62	22	accuracy	accuracy	NOUN
easat-2167	62	23	for	for	ADP
easat-2167	62	24	rice	rice	NOUN
easat-2167	62	25	disease	disease	NOUN
easat-2167	62	26	diagnosis	diagnosis	NOUN
easat-2167	63	1	[	[	X
easat-2167	63	2	18	18	NUM
easat-2167	63	3	]	]	PUNCT
easat-2167	63	4	.	.	PUNCT
easat-2167	64	1	2.2	2.2	NUM
easat-2167	64	2	.	.	PUNCT
easat-2167	65	1	image	image	NOUN
easat-2167	65	2	analysis	analysis	NOUN
easat-2167	65	3	for	for	ADP
easat-2167	65	4	disease	disease	NOUN
easat-2167	65	5	detection	detection	NOUN
easat-2167	65	6	an	an	DET
easat-2167	65	7	image	image	NOUN
easat-2167	65	8	processing	processing	NOUN
easat-2167	65	9	method	method	NOUN
easat-2167	65	10	for	for	ADP
easat-2167	65	11	classifying	classify	VERB
easat-2167	65	12	rice	rice	NOUN
easat-2167	65	13	diseases	disease	NOUN
easat-2167	65	14	based	base	VERB
easat-2167	65	15	on	on	ADP
easat-2167	65	16	leaf	leaf	NOUN
easat-2167	65	17	and	and	CCONJ
easat-2167	65	18	lesion	lesion	NOUN
easat-2167	65	19	attributes	attribute	NOUN
easat-2167	65	20	was	be	AUX
easat-2167	65	21	evaluated	evaluate	VERB
easat-2167	65	22	using	use	VERB
easat-2167	65	23	a	a	DET
easat-2167	65	24	publicly	publicly	ADV
easat-2167	65	25	available	available	ADJ
easat-2167	65	26	database	database	NOUN
easat-2167	66	1	[	[	X
easat-2167	66	2	19	19	NUM
easat-2167	66	3	]	]	PUNCT
easat-2167	66	4	.	.	PUNCT
easat-2167	67	1	automated	automate	VERB
easat-2167	67	2	system	system	NOUN
easat-2167	67	3	for	for	ADP
easat-2167	67	4	disease	disease	NOUN
easat-2167	67	5	monitoring	monitoring	NOUN
easat-2167	67	6	:	:	PUNCT
easat-2167	67	7	an	an	DET
easat-2167	67	8	automated	automate	VERB
easat-2167	67	9	system	system	NOUN
easat-2167	67	10	using	use	VERB
easat-2167	67	11	image	image	NOUN
easat-2167	67	12	processing	processing	NOUN
easat-2167	67	13	for	for	ADP
easat-2167	67	14	identifying	identify	VERB
easat-2167	67	15	four	four	NUM
easat-2167	67	16	common	common	ADJ
easat-2167	67	17	rice	rice	NOUN
easat-2167	67	18	diseases	disease	NOUN
easat-2167	67	19	achieved	achieve	VERB
easat-2167	67	20	high	high	ADJ
easat-2167	67	21	classification	classification	NOUN
easat-2167	67	22	accuracy	accuracy	NOUN
easat-2167	67	23	with	with	ADP
easat-2167	67	24	k	k	PROPN
easat-2167	67	25	-	-	PUNCT
easat-2167	67	26	nn	nn	ADJ
easat-2167	67	27	and	and	CCONJ
easat-2167	67	28	minimum	minimum	NOUN
easat-2167	67	29	distance	distance	NOUN
easat-2167	67	30	classifier	classifier	NOUN
easat-2167	67	31	(	(	PUNCT
easat-2167	67	32	mdc	mdc	NOUN
easat-2167	67	33	)	)	PUNCT
easat-2167	68	1	[	[	X
easat-2167	68	2	(	(	PUNCT
easat-2167	68	3	joshi	joshi	PROPN
easat-2167	68	4	&	&	CCONJ
easat-2167	68	5	jadhav	jadhav	PROPN
easat-2167	68	6	,	,	PUNCT
easat-2167	68	7	2016	2016	NUM
easat-2167	68	8	)	)	PUNCT
easat-2167	68	9	]	]	PUNCT
easat-2167	69	1	[	[	X
easat-2167	69	2	20	20	NUM
easat-2167	69	3	]	]	PUNCT
easat-2167	69	4	.	.	PUNCT
easat-2167	70	1	3	3	X
easat-2167	70	2	.	.	X
easat-2167	70	3	research	research	NOUN
easat-2167	70	4	problem	problem	NOUN
easat-2167	70	5	&	&	CCONJ
easat-2167	70	6	objectives	objective	NOUN
easat-2167	70	7	despite	despite	SCONJ
easat-2167	70	8	advances	advance	NOUN
easat-2167	70	9	in	in	ADP
easat-2167	70	10	image	image	NOUN
easat-2167	70	11	segmentation	segmentation	NOUN
easat-2167	70	12	and	and	CCONJ
easat-2167	70	13	deep	deep	ADJ
easat-2167	70	14	learning	learning	NOUN
easat-2167	70	15	,	,	PUNCT
easat-2167	70	16	existing	exist	VERB
easat-2167	70	17	methods	method	NOUN
easat-2167	70	18	for	for	ADP
easat-2167	70	19	rice	rice	NOUN
easat-2167	70	20	leaf	leaf	NOUN
easat-2167	70	21	disease	disease	NOUN
easat-2167	70	22	detection	detection	NOUN
easat-2167	70	23	face	face	VERB
easat-2167	70	24	several	several	ADJ
easat-2167	70	25	limitations	limitation	NOUN
easat-2167	70	26	:	:	PUNCT
easat-2167	70	27	•	•	NUM
easat-2167	70	28	segmentation	segmentation	NOUN
easat-2167	70	29	accuracy	accuracy	NOUN
easat-2167	70	30	:	:	PUNCT
easat-2167	70	31	accurately	accurately	ADV
easat-2167	70	32	segmenting	segment	VERB
easat-2167	70	33	diseased	diseased	ADJ
easat-2167	70	34	regions	region	NOUN
easat-2167	70	35	in	in	ADP
easat-2167	70	36	rice	rice	NOUN
easat-2167	70	37	leaves	leave	NOUN
easat-2167	70	38	remains	remain	VERB
easat-2167	70	39	challenging	challenge	VERB
easat-2167	70	40	due	due	ADP
easat-2167	70	41	to	to	ADP
easat-2167	70	42	the	the	DET
easat-2167	70	43	complex	complex	ADJ
easat-2167	70	44	and	and	CCONJ
easat-2167	70	45	varying	vary	VERB
easat-2167	70	46	backgrounds	background	NOUN
easat-2167	70	47	in	in	ADP
easat-2167	70	48	field	field	NOUN
easat-2167	70	49	images	image	NOUN
easat-2167	70	50	,	,	PUNCT
easat-2167	70	51	which	which	PRON
easat-2167	70	52	often	often	ADV
easat-2167	70	53	resemble	resemble	VERB
easat-2167	70	54	the	the	DET
easat-2167	70	55	color	color	NOUN
easat-2167	70	56	and	and	CCONJ
easat-2167	70	57	texture	texture	NOUN
easat-2167	70	58	of	of	ADP
easat-2167	70	59	the	the	DET
easat-2167	70	60	leaves	leave	NOUN
easat-2167	70	61	.	.	PUNCT
easat-2167	71	1	•	•	NUM
easat-2167	71	2	data	datum	NOUN
easat-2167	71	3	scarcity	scarcity	NOUN
easat-2167	71	4	:	:	PUNCT
easat-2167	71	5	deep	deep	ADJ
easat-2167	71	6	learning	learning	NOUN
easat-2167	71	7	models	model	NOUN
easat-2167	71	8	typically	typically	ADV
easat-2167	71	9	require	require	VERB
easat-2167	71	10	large	large	ADJ
easat-2167	71	11	,	,	PUNCT
easat-2167	71	12	annotated	annotate	VERB
easat-2167	71	13	datasets	dataset	NOUN
easat-2167	71	14	for	for	ADP
easat-2167	71	15	training	training	NOUN
easat-2167	71	16	.	.	PUNCT
easat-2167	72	1	however	however	ADV
easat-2167	72	2	,	,	PUNCT
easat-2167	72	3	such	such	ADJ
easat-2167	72	4	datasets	dataset	NOUN
easat-2167	72	5	are	be	AUX
easat-2167	72	6	often	often	ADV
easat-2167	72	7	limited	limit	VERB
easat-2167	72	8	,	,	PUNCT
easat-2167	72	9	particularly	particularly	ADV
easat-2167	72	10	for	for	ADP
easat-2167	72	11	rice	rice	NOUN
easat-2167	72	12	diseases	disease	NOUN
easat-2167	72	13	.	.	PUNCT
easat-2167	73	1	•	•	NUM
easat-2167	73	2	model	model	NOUN
easat-2167	73	3	generalization	generalization	NOUN
easat-2167	73	4	:	:	PUNCT
easat-2167	73	5	models	model	NOUN
easat-2167	73	6	trained	train	VERB
easat-2167	73	7	on	on	ADP
easat-2167	73	8	controlled	control	VERB
easat-2167	73	9	environments	environment	NOUN
easat-2167	73	10	may	may	AUX
easat-2167	73	11	not	not	PART
easat-2167	73	12	generalize	generalize	VERB
easat-2167	73	13	well	well	ADV
easat-2167	73	14	to	to	ADP
easat-2167	73	15	real	real	ADJ
easat-2167	73	16	-	-	PUNCT
easat-2167	73	17	world	world	NOUN
easat-2167	73	18	field	field	NOUN
easat-2167	73	19	conditions	condition	NOUN
easat-2167	73	20	,	,	PUNCT
easat-2167	73	21	where	where	SCONJ
easat-2167	73	22	lighting	lighting	NOUN
easat-2167	73	23	,	,	PUNCT
easat-2167	73	24	shadows	shadow	NOUN
easat-2167	73	25	,	,	PUNCT
easat-2167	73	26	and	and	CCONJ
easat-2167	73	27	occlusions	occlusion	NOUN
easat-2167	73	28	vary	vary	VERB
easat-2167	73	29	significantly	significantly	ADV
easat-2167	73	30	.	.	PUNCT
easat-2167	74	1	•	•	NUM
easat-2167	74	2	computation	computation	NOUN
easat-2167	74	3	efficiency	efficiency	NOUN
easat-2167	74	4	:	:	PUNCT
easat-2167	74	5	real	real	ADJ
easat-2167	74	6	-	-	PUNCT
easat-2167	74	7	time	time	NOUN
easat-2167	74	8	application	application	NOUN
easat-2167	74	9	of	of	ADP
easat-2167	74	10	these	these	DET
easat-2167	74	11	models	model	NOUN
easat-2167	74	12	in	in	ADP
easat-2167	74	13	the	the	DET
easat-2167	74	14	field	field	NOUN
easat-2167	74	15	requires	require	VERB
easat-2167	74	16	computationally	computationally	ADV
easat-2167	74	17	efficient	efficient	ADJ
easat-2167	74	18	algorithms	algorithm	NOUN
easat-2167	74	19	that	that	PRON
easat-2167	74	20	can	can	AUX
easat-2167	74	21	process	process	VERB
easat-2167	74	22	images	image	NOUN
easat-2167	74	23	quickly	quickly	ADV
easat-2167	74	24	and	and	CCONJ
easat-2167	74	25	accurately	accurately	ADV
easat-2167	74	26	on	on	ADP
easat-2167	74	27	resource	resource	NOUN
easat-2167	74	28	-	-	PUNCT
easat-2167	74	29	limited	limit	VERB
easat-2167	74	30	devices	device	NOUN
easat-2167	74	31	.	.	PUNCT
easat-2167	75	1	3.1	3.1	NUM
easat-2167	75	2	.	.	X
easat-2167	75	3	objective	objective	NOUN
easat-2167	75	4	the	the	DET
easat-2167	75	5	primary	primary	ADJ
easat-2167	75	6	objective	objective	NOUN
easat-2167	75	7	of	of	ADP
easat-2167	75	8	this	this	DET
easat-2167	75	9	research	research	NOUN
easat-2167	75	10	is	be	AUX
easat-2167	75	11	to	to	PART
easat-2167	75	12	develop	develop	VERB
easat-2167	75	13	a	a	DET
easat-2167	75	14	robust	robust	ADJ
easat-2167	75	15	,	,	PUNCT
easat-2167	75	16	efficient	efficient	ADJ
easat-2167	75	17	,	,	PUNCT
easat-2167	75	18	and	and	CCONJ
easat-2167	75	19	accurate	accurate	ADJ
easat-2167	75	20	method	method	NOUN
easat-2167	75	21	for	for	ADP
easat-2167	75	22	the	the	DET
easat-2167	75	23	796	796	NUM
easat-2167	75	24	edelweiss	edelweiss	PROPN
easat-2167	75	25	applied	apply	VERB
easat-2167	75	26	science	science	NOUN
easat-2167	75	27	and	and	CCONJ
easat-2167	75	28	technology	technology	NOUN
easat-2167	75	29	issn	issn	PROPN
easat-2167	75	30	:	:	PUNCT
easat-2167	75	31	2576	2576	NUM
easat-2167	75	32	-	-	SYM
easat-2167	75	33	8484	8484	NUM
easat-2167	75	34	vol	vol	NOUN
easat-2167	75	35	.	.	PROPN
easat-2167	75	36	8	8	NUM
easat-2167	75	37	,	,	PUNCT
easat-2167	75	38	no	no	INTJ
easat-2167	75	39	.	.	NOUN
easat-2167	76	1	6	6	NUM
easat-2167	76	2	:	:	PUNCT
easat-2167	76	3	793	793	NUM
easat-2167	76	4	-	-	NUM
easat-2167	76	5	813	813	NUM
easat-2167	76	6	,	,	PUNCT
easat-2167	76	7	2024	2024	NUM
easat-2167	76	8	doi	doi	NOUN
easat-2167	76	9	:	:	PUNCT
easat-2167	76	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	76	11	©	©	NOUN
easat-2167	76	12	2024	2024	NUM
easat-2167	76	13	by	by	ADP
easat-2167	76	14	the	the	DET
easat-2167	76	15	authors	author	NOUN
easat-2167	76	16	;	;	PUNCT
easat-2167	76	17	licensee	licensee	PROPN
easat-2167	76	18	learning	learn	VERB
easat-2167	76	19	gate	gate	NOUN
easat-2167	76	20	segmentation	segmentation	NOUN
easat-2167	76	21	and	and	CCONJ
easat-2167	76	22	classification	classification	NOUN
easat-2167	76	23	of	of	ADP
easat-2167	76	24	rice	rice	NOUN
easat-2167	76	25	leaf	leaf	NOUN
easat-2167	76	26	diseases	disease	NOUN
easat-2167	76	27	that	that	PRON
easat-2167	76	28	can	can	AUX
easat-2167	76	29	overcome	overcome	VERB
easat-2167	76	30	the	the	DET
easat-2167	76	31	limitations	limitation	NOUN
easat-2167	76	32	of	of	ADP
easat-2167	76	33	current	current	ADJ
easat-2167	76	34	approaches	approach	NOUN
easat-2167	76	35	.	.	PUNCT
easat-2167	77	1	this	this	PRON
easat-2167	77	2	includes	include	VERB
easat-2167	77	3	:	:	PUNCT
easat-2167	77	4	1	1	X
easat-2167	77	5	.	.	X
easat-2167	77	6	enhancing	enhance	VERB
easat-2167	77	7	segmentation	segmentation	NOUN
easat-2167	77	8	accuracy	accuracy	NOUN
easat-2167	77	9	:	:	PUNCT
easat-2167	77	10	developing	develop	VERB
easat-2167	77	11	advanced	advanced	ADJ
easat-2167	77	12	segmentation	segmentation	NOUN
easat-2167	77	13	techniques	technique	NOUN
easat-2167	77	14	that	that	PRON
easat-2167	77	15	effectively	effectively	ADV
easat-2167	77	16	differentiate	differentiate	VERB
easat-2167	77	17	between	between	ADP
easat-2167	77	18	diseased	diseased	ADJ
easat-2167	77	19	and	and	CCONJ
easat-2167	77	20	healthy	healthy	ADJ
easat-2167	77	21	regions	region	NOUN
easat-2167	77	22	in	in	ADP
easat-2167	77	23	complex	complex	ADJ
easat-2167	77	24	field	field	NOUN
easat-2167	77	25	images	image	NOUN
easat-2167	77	26	.	.	PUNCT
easat-2167	78	1	2	2	X
easat-2167	78	2	.	.	X
easat-2167	78	3	data	datum	NOUN
easat-2167	78	4	augmentation	augmentation	NOUN
easat-2167	78	5	and	and	CCONJ
easat-2167	78	6	synthetic	synthetic	ADJ
easat-2167	78	7	data	data	NOUN
easat-2167	78	8	generation	generation	NOUN
easat-2167	78	9	:	:	PUNCT
easat-2167	78	10	utilizing	utilize	VERB
easat-2167	78	11	data	datum	NOUN
easat-2167	78	12	augmentation	augmentation	NOUN
easat-2167	78	13	techniques	technique	NOUN
easat-2167	78	14	and	and	CCONJ
easat-2167	78	15	synthetic	synthetic	ADJ
easat-2167	78	16	data	data	NOUN
easat-2167	78	17	generation	generation	NOUN
easat-2167	78	18	(	(	PUNCT
easat-2167	78	19	e.g.	e.g.	ADV
easat-2167	78	20	,	,	PUNCT
easat-2167	78	21	copy	copy	NOUN
easat-2167	78	22	-	-	PUNCT
easat-2167	78	23	paste	paste	NOUN
easat-2167	78	24	techniques	technique	NOUN
easat-2167	78	25	)	)	PUNCT
easat-2167	78	26	to	to	PART
easat-2167	78	27	expand	expand	VERB
easat-2167	78	28	the	the	DET
easat-2167	78	29	training	training	NOUN
easat-2167	78	30	dataset	dataset	NOUN
easat-2167	78	31	and	and	CCONJ
easat-2167	78	32	improve	improve	VERB
easat-2167	78	33	model	model	NOUN
easat-2167	78	34	robustness	robustness	NOUN
easat-2167	78	35	.	.	PUNCT
easat-2167	79	1	3	3	X
easat-2167	79	2	.	.	PUNCT
easat-2167	79	3	model	model	NOUN
easat-2167	79	4	generalization	generalization	NOUN
easat-2167	79	5	:	:	PUNCT
easat-2167	79	6	implementing	implement	VERB
easat-2167	79	7	and	and	CCONJ
easat-2167	79	8	testing	testing	NOUN
easat-2167	79	9	models	model	NOUN
easat-2167	79	10	in	in	ADP
easat-2167	79	11	diverse	diverse	ADJ
easat-2167	79	12	real	real	ADJ
easat-2167	79	13	-	-	PUNCT
easat-2167	79	14	world	world	NOUN
easat-2167	79	15	conditions	condition	NOUN
easat-2167	79	16	to	to	PART
easat-2167	79	17	ensure	ensure	VERB
easat-2167	79	18	generalization	generalization	NOUN
easat-2167	79	19	across	across	ADP
easat-2167	79	20	different	different	ADJ
easat-2167	79	21	environmental	environmental	ADJ
easat-2167	79	22	settings	setting	NOUN
easat-2167	79	23	.	.	PUNCT
easat-2167	80	1	4	4	X
easat-2167	80	2	.	.	X
easat-2167	80	3	computation	computation	NOUN
easat-2167	80	4	efficiency	efficiency	NOUN
easat-2167	80	5	:	:	PUNCT
easat-2167	80	6	optimizing	optimize	VERB
easat-2167	80	7	the	the	DET
easat-2167	80	8	model	model	NOUN
easat-2167	80	9	to	to	PART
easat-2167	80	10	run	run	VERB
easat-2167	80	11	efficiently	efficiently	ADV
easat-2167	80	12	on	on	ADP
easat-2167	80	13	portable	portable	ADJ
easat-2167	80	14	devices	device	NOUN
easat-2167	80	15	for	for	ADP
easat-2167	80	16	realtime	realtime	NOUN
easat-2167	80	17	disease	disease	NOUN
easat-2167	80	18	detection	detection	NOUN
easat-2167	80	19	in	in	ADP
easat-2167	80	20	the	the	DET
easat-2167	80	21	field	field	NOUN
easat-2167	80	22	.	.	PUNCT
easat-2167	81	1	3.2	3.2	NUM
easat-2167	81	2	.	.	PUNCT
easat-2167	81	3	expected	expect	VERB
easat-2167	81	4	outcomes	outcome	NOUN
easat-2167	81	5	•	•	ADP
easat-2167	81	6	a	a	DET
easat-2167	81	7	novel	novel	ADJ
easat-2167	81	8	segmentation	segmentation	NOUN
easat-2167	81	9	algorithm	algorithm	NOUN
easat-2167	81	10	that	that	PRON
easat-2167	81	11	significantly	significantly	ADV
easat-2167	81	12	improves	improve	VERB
easat-2167	81	13	the	the	DET
easat-2167	81	14	accuracy	accuracy	NOUN
easat-2167	81	15	of	of	ADP
easat-2167	81	16	disease	disease	NOUN
easat-2167	81	17	region	region	NOUN
easat-2167	81	18	detection	detection	NOUN
easat-2167	81	19	in	in	ADP
easat-2167	81	20	rice	rice	NOUN
easat-2167	81	21	leaf	leaf	NOUN
easat-2167	81	22	images	image	NOUN
easat-2167	81	23	.	.	PUNCT
easat-2167	82	1	•	•	NUM
easat-2167	82	2	a	a	DET
easat-2167	82	3	deep	deep	ADJ
easat-2167	82	4	learning	learning	NOUN
easat-2167	82	5	model	model	NOUN
easat-2167	82	6	that	that	PRON
easat-2167	82	7	generalizes	generalize	VERB
easat-2167	82	8	well	well	ADV
easat-2167	82	9	across	across	ADP
easat-2167	82	10	different	different	ADJ
easat-2167	82	11	field	field	NOUN
easat-2167	82	12	conditions	condition	NOUN
easat-2167	82	13	,	,	PUNCT
easat-2167	82	14	ensuring	ensure	VERB
easat-2167	82	15	reliable	reliable	ADJ
easat-2167	82	16	performance	performance	NOUN
easat-2167	82	17	in	in	ADP
easat-2167	82	18	real	real	ADJ
easat-2167	82	19	-	-	PUNCT
easat-2167	82	20	world	world	NOUN
easat-2167	82	21	applications	application	NOUN
easat-2167	82	22	.	.	PUNCT
easat-2167	83	1	•	•	NUM
easat-2167	83	2	an	an	DET
easat-2167	83	3	efficient	efficient	ADJ
easat-2167	83	4	,	,	PUNCT
easat-2167	83	5	portable	portable	ADJ
easat-2167	83	6	system	system	NOUN
easat-2167	83	7	for	for	ADP
easat-2167	83	8	real	real	ADJ
easat-2167	83	9	-	-	PUNCT
easat-2167	83	10	time	time	NOUN
easat-2167	83	11	rice	rice	NOUN
easat-2167	83	12	disease	disease	NOUN
easat-2167	83	13	detection	detection	NOUN
easat-2167	83	14	,	,	PUNCT
easat-2167	83	15	aiding	aid	VERB
easat-2167	83	16	farmers	farmer	NOUN
easat-2167	83	17	in	in	ADP
easat-2167	83	18	timely	timely	ADJ
easat-2167	83	19	and	and	CCONJ
easat-2167	83	20	accurate	accurate	ADJ
easat-2167	83	21	disease	disease	NOUN
easat-2167	83	22	management	management	NOUN
easat-2167	83	23	.	.	PUNCT
easat-2167	84	1	impact	impact	NOUN
easat-2167	84	2	:	:	PUNCT
easat-2167	84	3	this	this	DET
easat-2167	84	4	research	research	NOUN
easat-2167	84	5	will	will	AUX
easat-2167	84	6	contribute	contribute	VERB
easat-2167	84	7	to	to	ADP
easat-2167	84	8	the	the	DET
easat-2167	84	9	field	field	NOUN
easat-2167	84	10	of	of	ADP
easat-2167	84	11	precision	precision	NOUN
easat-2167	84	12	agriculture	agriculture	NOUN
easat-2167	84	13	by	by	ADP
easat-2167	84	14	providing	provide	VERB
easat-2167	84	15	a	a	DET
easat-2167	84	16	reliable	reliable	ADJ
easat-2167	84	17	and	and	CCONJ
easat-2167	84	18	practical	practical	ADJ
easat-2167	84	19	tool	tool	NOUN
easat-2167	84	20	for	for	ADP
easat-2167	84	21	early	early	ADJ
easat-2167	84	22	disease	disease	NOUN
easat-2167	84	23	detection	detection	NOUN
easat-2167	84	24	in	in	ADP
easat-2167	84	25	rice	rice	NOUN
easat-2167	84	26	crops	crop	NOUN
easat-2167	84	27	.	.	PUNCT
easat-2167	85	1	the	the	DET
easat-2167	85	2	outcomes	outcome	NOUN
easat-2167	85	3	could	could	AUX
easat-2167	85	4	lead	lead	VERB
easat-2167	85	5	to	to	ADP
easat-2167	85	6	improved	improved	ADJ
easat-2167	85	7	crop	crop	NOUN
easat-2167	85	8	management	management	NOUN
easat-2167	85	9	practices	practice	NOUN
easat-2167	85	10	,	,	PUNCT
easat-2167	85	11	reduced	reduce	VERB
easat-2167	85	12	losses	loss	NOUN
easat-2167	85	13	,	,	PUNCT
easat-2167	85	14	and	and	CCONJ
easat-2167	85	15	enhanced	enhance	VERB
easat-2167	85	16	food	food	NOUN
easat-2167	85	17	security	security	NOUN
easat-2167	85	18	.	.	PUNCT
easat-2167	86	1	4	4	X
easat-2167	86	2	.	.	X
easat-2167	86	3	materials	material	NOUN
easat-2167	86	4	and	and	CCONJ
easat-2167	86	5	methods	method	NOUN
easat-2167	86	6	the	the	DET
easat-2167	86	7	dataset	dataset	ADJ
easat-2167	86	8	rice	rice	NOUN
easat-2167	86	9	leaf	leaf	NOUN
easat-2167	86	10	disease	disease	NOUN
easat-2167	86	11	with	with	ADP
easat-2167	86	12	segmentation	segmentation	NOUN
easat-2167	86	13	labels	label	NOUN
easat-2167	86	14	[	[	X
easat-2167	86	15	21	21	NUM
easat-2167	86	16	]	]	PUNCT
easat-2167	86	17	,	,	PUNCT
easat-2167	86	18	is	be	AUX
easat-2167	86	19	designed	design	VERB
easat-2167	86	20	for	for	ADP
easat-2167	86	21	the	the	DET
easat-2167	86	22	task	task	NOUN
easat-2167	86	23	of	of	ADP
easat-2167	86	24	image	image	NOUN
easat-2167	86	25	segmentation	segmentation	NOUN
easat-2167	86	26	,	,	PUNCT
easat-2167	86	27	specifically	specifically	ADV
easat-2167	86	28	targeting	target	VERB
easat-2167	86	29	the	the	DET
easat-2167	86	30	identification	identification	NOUN
easat-2167	86	31	and	and	CCONJ
easat-2167	86	32	segmentation	segmentation	NOUN
easat-2167	86	33	of	of	ADP
easat-2167	86	34	rice	rice	NOUN
easat-2167	86	35	leaf	leaf	NOUN
easat-2167	86	36	diseases	disease	NOUN
easat-2167	86	37	.	.	PUNCT
easat-2167	87	1	the	the	DET
easat-2167	87	2	dataset	dataset	NOUN
easat-2167	87	3	is	be	AUX
easat-2167	87	4	a	a	DET
easat-2167	87	5	curated	curate	VERB
easat-2167	87	6	collection	collection	NOUN
easat-2167	87	7	derived	derive	VERB
easat-2167	87	8	from	from	ADP
easat-2167	87	9	three	three	NUM
easat-2167	87	10	publicly	publicly	ADV
easat-2167	87	11	available	available	ADJ
easat-2167	87	12	datasets	dataset	NOUN
easat-2167	87	13	,	,	PUNCT
easat-2167	87	14	specifically	specifically	ADV
easat-2167	87	15	filtered	filter	VERB
easat-2167	87	16	and	and	CCONJ
easat-2167	87	17	annotated	annotate	VERB
easat-2167	87	18	to	to	PART
easat-2167	87	19	focus	focus	VERB
easat-2167	87	20	on	on	ADP
easat-2167	87	21	three	three	NUM
easat-2167	87	22	types	type	NOUN
easat-2167	87	23	of	of	ADP
easat-2167	87	24	rice	rice	NOUN
easat-2167	87	25	leaf	leaf	NOUN
easat-2167	87	26	diseases	disease	NOUN
easat-2167	87	27	:	:	PUNCT
easat-2167	87	28	rice	rice	NOUN
easat-2167	87	29	bacterial	bacterial	ADJ
easat-2167	87	30	blight	blight	NOUN
easat-2167	87	31	,	,	PUNCT
easat-2167	87	32	rice	rice	NOUN
easat-2167	87	33	blast	blast	NOUN
easat-2167	87	34	,	,	PUNCT
easat-2167	87	35	and	and	CCONJ
easat-2167	87	36	brown	brown	ADJ
easat-2167	87	37	spot	spot	NOUN
easat-2167	87	38	.	.	PUNCT
easat-2167	88	1	for	for	ADP
easat-2167	88	2	each	each	DET
easat-2167	88	3	disease	disease	NOUN
easat-2167	88	4	category	category	NOUN
easat-2167	88	5	,	,	PUNCT
easat-2167	88	6	the	the	DET
easat-2167	88	7	dataset	dataset	NOUN
easat-2167	88	8	includes	include	VERB
easat-2167	88	9	150	150	NUM
easat-2167	88	10	original	original	ADJ
easat-2167	88	11	images	image	NOUN
easat-2167	88	12	of	of	ADP
easat-2167	88	13	diseased	diseased	ADJ
easat-2167	88	14	rice	rice	NOUN
easat-2167	88	15	leaves	leave	NOUN
easat-2167	88	16	,	,	PUNCT
easat-2167	88	17	as	as	ADV
easat-2167	88	18	well	well	ADV
easat-2167	88	19	as	as	ADP
easat-2167	88	20	their	their	PRON
easat-2167	88	21	corresponding	correspond	VERB
easat-2167	88	22	segmentation	segmentation	NOUN
easat-2167	88	23	masks	mask	NOUN
easat-2167	88	24	that	that	PRON
easat-2167	88	25	provide	provide	VERB
easat-2167	88	26	semantic	semantic	ADJ
easat-2167	88	27	information	information	NOUN
easat-2167	88	28	essentially	essentially	ADV
easat-2167	88	29	labelling	label	VERB
easat-2167	88	30	the	the	DET
easat-2167	88	31	diseased	diseased	ADJ
easat-2167	88	32	areas	area	NOUN
easat-2167	88	33	in	in	ADP
easat-2167	88	34	the	the	DET
easat-2167	88	35	images	image	NOUN
easat-2167	88	36	.	.	PUNCT
easat-2167	89	1	the	the	DET
easat-2167	89	2	dataset	dataset	NOUN
easat-2167	89	3	consists	consist	VERB
easat-2167	89	4	of	of	ADP
easat-2167	89	5	two	two	NUM
easat-2167	89	6	main	main	ADJ
easat-2167	89	7	components	component	NOUN
easat-2167	89	8	:	:	PUNCT
easat-2167	89	9	4.1	4.1	NUM
easat-2167	89	10	.	.	PUNCT
easat-2167	90	1	contents	content	NOUN
easat-2167	90	2	images	image	NOUN
easat-2167	90	3	:	:	PUNCT
easat-2167	90	4	the	the	DET
easat-2167	90	5	dataset	dataset	NOUN
easat-2167	90	6	contains	contain	VERB
easat-2167	90	7	rgb	rgb	PROPN
easat-2167	90	8	images	image	NOUN
easat-2167	90	9	of	of	ADP
easat-2167	90	10	rice	rice	NOUN
easat-2167	90	11	leaves	leave	NOUN
easat-2167	90	12	captured	capture	VERB
easat-2167	90	13	in	in	ADP
easat-2167	90	14	different	different	ADJ
easat-2167	90	15	conditions	condition	NOUN
easat-2167	90	16	.	.	PUNCT
easat-2167	91	1	these	these	DET
easat-2167	91	2	images	image	NOUN
easat-2167	91	3	showcase	showcase	NOUN
easat-2167	91	4	leaves	leave	NOUN
easat-2167	91	5	affected	affect	VERB
easat-2167	91	6	by	by	ADP
easat-2167	91	7	various	various	ADJ
easat-2167	91	8	diseases	disease	NOUN
easat-2167	91	9	,	,	PUNCT
easat-2167	91	10	including	include	VERB
easat-2167	91	11	but	but	CCONJ
easat-2167	91	12	not	not	PART
easat-2167	91	13	limited	limit	VERB
easat-2167	91	14	to	to	ADP
easat-2167	91	15	bacterial	bacterial	ADJ
easat-2167	91	16	blight	blight	NOUN
easat-2167	91	17	,	,	PUNCT
easat-2167	91	18	brown	brown	ADJ
easat-2167	91	19	spot	spot	NOUN
easat-2167	91	20	,	,	PUNCT
easat-2167	91	21	and	and	CCONJ
easat-2167	91	22	blast	blast	NOUN
easat-2167	91	23	.	.	PUNCT
easat-2167	92	1	the	the	DET
easat-2167	92	2	images	image	NOUN
easat-2167	92	3	typically	typically	ADV
easat-2167	92	4	have	have	VERB
easat-2167	92	5	different	different	ADJ
easat-2167	92	6	resolutions	resolution	NOUN
easat-2167	92	7	,	,	PUNCT
easat-2167	92	8	though	though	SCONJ
easat-2167	92	9	they	they	PRON
easat-2167	92	10	may	may	AUX
easat-2167	92	11	be	be	AUX
easat-2167	92	12	normalized	normalize	VERB
easat-2167	92	13	or	or	CCONJ
easat-2167	92	14	resized	resize	VERB
easat-2167	92	15	for	for	ADP
easat-2167	92	16	consistent	consistent	ADJ
easat-2167	92	17	input	input	NOUN
easat-2167	92	18	into	into	ADP
easat-2167	92	19	machine	machine	NOUN
easat-2167	92	20	learning	learning	NOUN
easat-2167	92	21	models	model	NOUN
easat-2167	92	22	.	.	PUNCT
easat-2167	93	1	4.2	4.2	NUM
easat-2167	93	2	.	.	PUNCT
easat-2167	94	1	segmentation	segmentation	NOUN
easat-2167	94	2	masks	mask	NOUN
easat-2167	94	3	each	each	DET
easat-2167	94	4	image	image	NOUN
easat-2167	94	5	has	have	VERB
easat-2167	94	6	an	an	DET
easat-2167	94	7	associated	associated	ADJ
easat-2167	94	8	segmentation	segmentation	NOUN
easat-2167	94	9	mask	mask	NOUN
easat-2167	94	10	that	that	PRON
easat-2167	94	11	labels	label	VERB
easat-2167	94	12	the	the	DET
easat-2167	94	13	pixels	pixel	NOUN
easat-2167	94	14	in	in	ADP
easat-2167	94	15	the	the	DET
easat-2167	94	16	image	image	NOUN
easat-2167	94	17	as	as	ADP
easat-2167	94	18	either	either	DET
easat-2167	94	19	part	part	NOUN
easat-2167	94	20	of	of	ADP
easat-2167	94	21	a	a	DET
easat-2167	94	22	diseased	diseased	ADJ
easat-2167	94	23	area	area	NOUN
easat-2167	94	24	or	or	CCONJ
easat-2167	94	25	a	a	DET
easat-2167	94	26	healthy	healthy	ADJ
easat-2167	94	27	area	area	NOUN
easat-2167	94	28	.	.	PUNCT
easat-2167	95	1	the	the	DET
easat-2167	95	2	masks	mask	NOUN
easat-2167	95	3	are	be	AUX
easat-2167	95	4	usually	usually	ADV
easat-2167	95	5	in	in	ADP
easat-2167	95	6	grayscale	grayscale	NOUN
easat-2167	95	7	,	,	PUNCT
easat-2167	95	8	where	where	SCONJ
easat-2167	95	9	one	one	NUM
easat-2167	95	10	pixel	pixel	NOUN
easat-2167	95	11	value	value	NOUN
easat-2167	95	12	(	(	PUNCT
easat-2167	95	13	e.g.	e.g.	ADV
easat-2167	95	14	,	,	PUNCT
easat-2167	95	15	0	0	NUM
easat-2167	95	16	)	)	PUNCT
easat-2167	95	17	might	might	AUX
easat-2167	95	18	represent	represent	VERB
easat-2167	95	19	healthy	healthy	ADJ
easat-2167	95	20	leaf	leaf	NOUN
easat-2167	95	21	tissue	tissue	NOUN
easat-2167	95	22	,	,	PUNCT
easat-2167	95	23	and	and	CCONJ
easat-2167	95	24	another	another	DET
easat-2167	95	25	pixel	pixel	NOUN
easat-2167	95	26	value	value	NOUN
easat-2167	95	27	(	(	PUNCT
easat-2167	95	28	e.g.	e.g.	ADV
easat-2167	95	29	,	,	PUNCT
easat-2167	95	30	1	1	NUM
easat-2167	95	31	)	)	PUNCT
easat-2167	95	32	represents	represent	VERB
easat-2167	95	33	diseased	diseased	ADJ
easat-2167	95	34	tissue	tissue	NOUN
easat-2167	95	35	.	.	PUNCT
easat-2167	96	1	https://www.kaggle.com/datasets/slygirl/rice-leaf-disease-with-segmentation-labels	https://www.kaggle.com/datasets/slygirl/rice-leaf-disease-with-segmentation-label	VERB
easat-2167	96	2	797	797	NUM
easat-2167	96	3	edelweiss	edelweiss	PROPN
easat-2167	96	4	applied	apply	VERB
easat-2167	96	5	science	science	NOUN
easat-2167	96	6	and	and	CCONJ
easat-2167	96	7	technology	technology	NOUN
easat-2167	96	8	issn	issn	PROPN
easat-2167	96	9	:	:	PUNCT
easat-2167	96	10	2576	2576	NUM
easat-2167	96	11	-	-	SYM
easat-2167	96	12	8484	8484	NUM
easat-2167	96	13	vol	vol	NOUN
easat-2167	96	14	.	.	PROPN
easat-2167	96	15	8	8	NUM
easat-2167	96	16	,	,	PUNCT
easat-2167	96	17	no	no	INTJ
easat-2167	96	18	.	.	NOUN
easat-2167	97	1	6	6	NUM
easat-2167	97	2	:	:	PUNCT
easat-2167	97	3	793	793	NUM
easat-2167	97	4	-	-	NUM
easat-2167	97	5	813	813	NUM
easat-2167	97	6	,	,	PUNCT
easat-2167	97	7	2024	2024	NUM
easat-2167	97	8	doi	doi	NOUN
easat-2167	97	9	:	:	PUNCT
easat-2167	97	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	97	11	©	©	NOUN
easat-2167	97	12	2024	2024	NUM
easat-2167	97	13	by	by	ADP
easat-2167	97	14	the	the	DET
easat-2167	97	15	authors	author	NOUN
easat-2167	97	16	;	;	PUNCT
easat-2167	97	17	licensee	licensee	PROPN
easat-2167	97	18	learning	learning	PROPN
easat-2167	97	19	gate	gate	NOUN
easat-2167	97	20	figure.1	figure.1	PROPN
easat-2167	97	21	.	.	PUNCT
easat-2167	98	1	(	(	PUNCT
easat-2167	98	2	a)bacterial	a)bacterial	ADJ
easat-2167	98	3	blight	blight	NOUN
easat-2167	98	4	.	.	PUNCT
easat-2167	99	1	figure	figure	NOUN
easat-2167	99	2	1	1	NUM
easat-2167	99	3	.	.	PUNCT
easat-2167	100	1	(	(	PUNCT
easat-2167	100	2	b	b	X
easat-2167	100	3	)	)	PUNCT
easat-2167	100	4	brown	brown	ADJ
easat-2167	100	5	spot	spot	NOUN
easat-2167	100	6	.	.	PUNCT
easat-2167	101	1	figure	figure	NOUN
easat-2167	101	2	1	1	NUM
easat-2167	101	3	.	.	PUNCT
easat-2167	102	1	(	(	PUNCT
easat-2167	102	2	c	c	NOUN
easat-2167	102	3	)	)	PUNCT
easat-2167	102	4	blast	blast	NOUN
easat-2167	102	5	.	.	PUNCT
easat-2167	103	1	798	798	NUM
easat-2167	103	2	edelweiss	edelweiss	PROPN
easat-2167	103	3	applied	apply	VERB
easat-2167	103	4	science	science	NOUN
easat-2167	103	5	and	and	CCONJ
easat-2167	103	6	technology	technology	NOUN
easat-2167	103	7	issn	issn	PROPN
easat-2167	103	8	:	:	PUNCT
easat-2167	103	9	2576	2576	NUM
easat-2167	103	10	-	-	SYM
easat-2167	103	11	8484	8484	NUM
easat-2167	103	12	vol	vol	NOUN
easat-2167	103	13	.	.	PROPN
easat-2167	104	1	8	8	NUM
easat-2167	104	2	,	,	PUNCT
easat-2167	104	3	no	no	INTJ
easat-2167	104	4	.	.	NOUN
easat-2167	105	1	6	6	NUM
easat-2167	105	2	:	:	PUNCT
easat-2167	105	3	793	793	NUM
easat-2167	105	4	-	-	NUM
easat-2167	105	5	813	813	NUM
easat-2167	105	6	,	,	PUNCT
easat-2167	105	7	2024	2024	NUM
easat-2167	105	8	doi	doi	NOUN
easat-2167	105	9	:	:	PUNCT
easat-2167	105	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	105	11	©	©	NOUN
easat-2167	105	12	2024	2024	NUM
easat-2167	105	13	by	by	ADP
easat-2167	105	14	the	the	DET
easat-2167	105	15	authors	author	NOUN
easat-2167	105	16	;	;	PUNCT
easat-2167	105	17	licensee	licensee	PROPN
easat-2167	105	18	learning	learning	NOUN
easat-2167	105	19	gate	gate	NOUN
easat-2167	105	20	5	5	NUM
easat-2167	105	21	.	.	PUNCT
easat-2167	105	22	data	datum	NOUN
easat-2167	105	23	preparation	preparation	NOUN
easat-2167	105	24	in	in	ADP
easat-2167	105	25	this	this	DET
easat-2167	105	26	data	data	NOUN
easat-2167	105	27	preparation	preparation	NOUN
easat-2167	105	28	process	process	NOUN
easat-2167	105	29	,	,	PUNCT
easat-2167	105	30	the	the	DET
easat-2167	105	31	goal	goal	NOUN
easat-2167	105	32	was	be	AUX
easat-2167	105	33	to	to	PART
easat-2167	105	34	effectively	effectively	ADV
easat-2167	105	35	prepare	prepare	VERB
easat-2167	105	36	a	a	DET
easat-2167	105	37	small	small	ADJ
easat-2167	105	38	dataset	dataset	NOUN
easat-2167	105	39	of	of	ADP
easat-2167	105	40	rice	rice	NOUN
easat-2167	105	41	blast	blast	NOUN
easat-2167	105	42	images	image	NOUN
easat-2167	105	43	and	and	CCONJ
easat-2167	105	44	their	their	PRON
easat-2167	105	45	corresponding	correspond	VERB
easat-2167	105	46	masks	mask	NOUN
easat-2167	105	47	for	for	ADP
easat-2167	105	48	training	train	VERB
easat-2167	105	49	a	a	DET
easat-2167	105	50	deep	deep	ADJ
easat-2167	105	51	learning	learning	NOUN
easat-2167	105	52	model	model	NOUN
easat-2167	105	53	in	in	ADP
easat-2167	105	54	image	image	NOUN
easat-2167	105	55	segmentation	segmentation	NOUN
easat-2167	105	56	.	.	PUNCT
easat-2167	106	1	the	the	DET
easat-2167	106	2	steps	step	NOUN
easat-2167	106	3	involved	involve	VERB
easat-2167	106	4	were	be	AUX
easat-2167	106	5	crucial	crucial	ADJ
easat-2167	106	6	for	for	ADP
easat-2167	106	7	ensuring	ensure	VERB
easat-2167	106	8	that	that	SCONJ
easat-2167	106	9	the	the	DET
easat-2167	106	10	dataset	dataset	NOUN
easat-2167	106	11	was	be	AUX
easat-2167	106	12	ready	ready	ADJ
easat-2167	106	13	for	for	ADP
easat-2167	106	14	model	model	NOUN
easat-2167	106	15	training	training	NOUN
easat-2167	106	16	,	,	PUNCT
easat-2167	106	17	especially	especially	ADV
easat-2167	106	18	considering	consider	VERB
easat-2167	106	19	the	the	DET
easat-2167	106	20	limited	limited	ADJ
easat-2167	106	21	number	number	NOUN
easat-2167	106	22	of	of	ADP
easat-2167	106	23	images	image	NOUN
easat-2167	106	24	available	available	ADJ
easat-2167	106	25	.	.	PUNCT
easat-2167	107	1	5.1	5.1	NUM
easat-2167	107	2	.	.	PUNCT
easat-2167	107	3	loading	loading	NOUN
easat-2167	107	4	and	and	CCONJ
easat-2167	107	5	preprocessing	preprocesse	VERB
easat-2167	107	6	images	image	NOUN
easat-2167	107	7	and	and	CCONJ
easat-2167	107	8	masks	mask	NOUN
easat-2167	107	9	•	•	NOUN
easat-2167	107	10	image	image	NOUN
easat-2167	107	11	and	and	CCONJ
easat-2167	107	12	mask	mask	NOUN
easat-2167	107	13	loading	loading	NOUN
easat-2167	107	14	:	:	PUNCT
easat-2167	107	15	images	image	NOUN
easat-2167	107	16	and	and	CCONJ
easat-2167	107	17	masks	mask	NOUN
easat-2167	107	18	were	be	AUX
easat-2167	107	19	loaded	load	VERB
easat-2167	107	20	from	from	ADP
easat-2167	107	21	specified	specified	ADJ
easat-2167	107	22	directories	directory	NOUN
easat-2167	107	23	.	.	PUNCT
easat-2167	108	1	each	each	DET
easat-2167	108	2	image	image	NOUN
easat-2167	108	3	was	be	AUX
easat-2167	108	4	resized	resize	VERB
easat-2167	108	5	to	to	ADP
easat-2167	108	6	a	a	DET
easat-2167	108	7	target	target	NOUN
easat-2167	108	8	size	size	NOUN
easat-2167	108	9	of	of	ADP
easat-2167	108	10	128x128	128x128	NUM
easat-2167	108	11	pixels	pixel	NOUN
easat-2167	108	12	to	to	PART
easat-2167	108	13	standardize	standardize	VERB
easat-2167	108	14	the	the	DET
easat-2167	108	15	input	input	NOUN
easat-2167	108	16	dimensions	dimension	NOUN
easat-2167	108	17	for	for	ADP
easat-2167	108	18	the	the	DET
easat-2167	108	19	model	model	NOUN
easat-2167	108	20	.	.	PUNCT
easat-2167	109	1	masks	mask	NOUN
easat-2167	109	2	were	be	AUX
easat-2167	109	3	converted	convert	VERB
easat-2167	109	4	to	to	ADP
easat-2167	109	5	grayscale	grayscale	NOUN
easat-2167	109	6	and	and	CCONJ
easat-2167	109	7	reshaped	reshape	VERB
easat-2167	109	8	to	to	PART
easat-2167	109	9	match	match	VERB
easat-2167	109	10	the	the	DET
easat-2167	109	11	image	image	NOUN
easat-2167	109	12	dimensions	dimension	NOUN
easat-2167	109	13	.	.	PUNCT
easat-2167	110	1	•	•	NOUN
easat-2167	110	2	splitting	split	VERB
easat-2167	110	3	data	datum	NOUN
easat-2167	110	4	into	into	ADP
easat-2167	110	5	training	training	NOUN
easat-2167	110	6	and	and	CCONJ
easat-2167	110	7	validation	validation	NOUN
easat-2167	110	8	set	set	NOUN
easat-2167	110	9	:	:	PUNCT
easat-2167	110	10	the	the	DET
easat-2167	110	11	dataset	dataset	NOUN
easat-2167	110	12	was	be	AUX
easat-2167	110	13	split	split	VERB
easat-2167	110	14	into	into	ADP
easat-2167	110	15	training	training	NOUN
easat-2167	110	16	and	and	CCONJ
easat-2167	110	17	validation	validation	NOUN
easat-2167	110	18	sets	set	NOUN
easat-2167	110	19	,	,	PUNCT
easat-2167	110	20	with	with	ADP
easat-2167	110	21	20	20	NUM
easat-2167	110	22	%	%	NOUN
easat-2167	110	23	of	of	ADP
easat-2167	110	24	the	the	DET
easat-2167	110	25	data	datum	NOUN
easat-2167	110	26	allocated	allocate	VERB
easat-2167	110	27	to	to	ADP
easat-2167	110	28	validation	validation	NOUN
easat-2167	110	29	.	.	PUNCT
easat-2167	111	1	this	this	DET
easat-2167	111	2	split	split	NOUN
easat-2167	111	3	helps	help	VERB
easat-2167	111	4	in	in	ADP
easat-2167	111	5	monitoring	monitor	VERB
easat-2167	111	6	the	the	DET
easat-2167	111	7	model	model	NOUN
easat-2167	111	8	's	's	PART
easat-2167	111	9	performance	performance	NOUN
easat-2167	111	10	on	on	ADP
easat-2167	111	11	unseen	unseen	ADJ
easat-2167	111	12	data	datum	NOUN
easat-2167	111	13	during	during	ADP
easat-2167	111	14	training	training	NOUN
easat-2167	111	15	,	,	PUNCT
easat-2167	111	16	which	which	PRON
easat-2167	111	17	is	be	AUX
easat-2167	111	18	essential	essential	ADJ
easat-2167	111	19	for	for	ADP
easat-2167	111	20	assessing	assess	VERB
easat-2167	111	21	its	its	PRON
easat-2167	111	22	generalization	generalization	NOUN
easat-2167	111	23	ability	ability	NOUN
easat-2167	111	24	.	.	PUNCT
easat-2167	112	1	5.2	5.2	NUM
easat-2167	112	2	.	.	PUNCT
easat-2167	113	1	data	datum	NOUN
easat-2167	113	2	augmentation	augmentation	NOUN
easat-2167	113	3	to	to	PART
easat-2167	113	4	artificially	artificially	ADV
easat-2167	113	5	expand	expand	VERB
easat-2167	113	6	the	the	DET
easat-2167	113	7	training	training	NOUN
easat-2167	113	8	dataset	dataset	NOUN
easat-2167	113	9	and	and	CCONJ
easat-2167	113	10	make	make	VERB
easat-2167	113	11	the	the	DET
easat-2167	113	12	model	model	NOUN
easat-2167	113	13	more	more	ADV
easat-2167	113	14	robust	robust	ADJ
easat-2167	113	15	to	to	ADP
easat-2167	113	16	variations	variation	NOUN
easat-2167	113	17	,	,	PUNCT
easat-2167	113	18	several	several	ADJ
easat-2167	113	19	data	datum	NOUN
easat-2167	113	20	augmentation	augmentation	NOUN
easat-2167	113	21	techniques	technique	NOUN
easat-2167	113	22	were	be	AUX
easat-2167	113	23	applied	apply	VERB
easat-2167	113	24	.	.	PUNCT
easat-2167	114	1	the	the	DET
easat-2167	114	2	image	image	NOUN
easat-2167	114	3	augmentation	augmentation	NOUN
easat-2167	114	4	process	process	NOUN
easat-2167	114	5	involved	involve	VERB
easat-2167	114	6	several	several	ADJ
easat-2167	114	7	transformations	transformation	NOUN
easat-2167	114	8	to	to	PART
easat-2167	114	9	enhance	enhance	VERB
easat-2167	114	10	the	the	DET
easat-2167	114	11	dataset	dataset	NOUN
easat-2167	114	12	's	's	PART
easat-2167	114	13	variability	variability	NOUN
easat-2167	114	14	and	and	CCONJ
easat-2167	114	15	robustness	robustness	NOUN
easat-2167	114	16	.	.	PUNCT
easat-2167	115	1	images	image	NOUN
easat-2167	115	2	were	be	AUX
easat-2167	115	3	rotated	rotate	VERB
easat-2167	115	4	randomly	randomly	ADV
easat-2167	115	5	within	within	ADP
easat-2167	115	6	a	a	DET
easat-2167	115	7	10	10	NUM
easat-2167	115	8	-	-	PUNCT
easat-2167	115	9	degree	degree	NOUN
easat-2167	115	10	range	range	NOUN
easat-2167	115	11	to	to	PART
easat-2167	115	12	account	account	VERB
easat-2167	115	13	for	for	ADP
easat-2167	115	14	variations	variation	NOUN
easat-2167	115	15	in	in	ADP
easat-2167	115	16	orientation	orientation	NOUN
easat-2167	115	17	.	.	PUNCT
easat-2167	116	1	shifting	shift	VERB
easat-2167	116	2	transformations	transformation	NOUN
easat-2167	116	3	were	be	AUX
easat-2167	116	4	applied	apply	VERB
easat-2167	116	5	to	to	ADP
easat-2167	116	6	both	both	DET
easat-2167	116	7	the	the	DET
easat-2167	116	8	width	width	NOUN
easat-2167	116	9	and	and	CCONJ
easat-2167	116	10	height	height	NOUN
easat-2167	116	11	,	,	PUNCT
easat-2167	116	12	allowing	allow	VERB
easat-2167	116	13	for	for	ADP
easat-2167	116	14	displacements	displacement	NOUN
easat-2167	116	15	of	of	ADP
easat-2167	116	16	up	up	ADP
easat-2167	116	17	to	to	PART
easat-2167	116	18	10	10	NUM
easat-2167	116	19	%	%	NOUN
easat-2167	116	20	in	in	ADP
easat-2167	116	21	either	either	DET
easat-2167	116	22	direction	direction	NOUN
easat-2167	116	23	.	.	PUNCT
easat-2167	117	1	additionally	additionally	ADV
easat-2167	117	2	,	,	PUNCT
easat-2167	117	3	a	a	DET
easat-2167	117	4	shearing	shearing	NOUN
easat-2167	117	5	transformation	transformation	NOUN
easat-2167	117	6	within	within	ADP
easat-2167	117	7	a	a	DET
easat-2167	117	8	10	10	NUM
easat-2167	117	9	%	%	NOUN
easat-2167	117	10	range	range	NOUN
easat-2167	117	11	was	be	AUX
easat-2167	117	12	introduced	introduce	VERB
easat-2167	117	13	to	to	PART
easat-2167	117	14	simulate	simulate	VERB
easat-2167	117	15	distortions	distortion	NOUN
easat-2167	117	16	in	in	ADP
easat-2167	117	17	the	the	DET
easat-2167	117	18	images	image	NOUN
easat-2167	117	19	.	.	PUNCT
easat-2167	118	1	zooming	zooming	NOUN
easat-2167	118	2	was	be	AUX
easat-2167	118	3	also	also	ADV
easat-2167	118	4	performed	perform	VERB
easat-2167	118	5	,	,	PUNCT
easat-2167	118	6	with	with	ADP
easat-2167	118	7	images	image	NOUN
easat-2167	118	8	being	be	AUX
easat-2167	118	9	randomly	randomly	ADV
easat-2167	118	10	zoomed	zoom	VERB
easat-2167	118	11	in	in	ADP
easat-2167	118	12	or	or	CCONJ
easat-2167	118	13	out	out	ADV
easat-2167	118	14	by	by	ADP
easat-2167	118	15	up	up	ADP
easat-2167	118	16	to	to	PART
easat-2167	118	17	10	10	NUM
easat-2167	118	18	%	%	NOUN
easat-2167	118	19	.	.	PUNCT
easat-2167	119	1	to	to	PART
easat-2167	119	2	further	far	ADV
easat-2167	119	3	diversify	diversify	VERB
easat-2167	119	4	the	the	DET
easat-2167	119	5	dataset	dataset	NOUN
easat-2167	119	6	,	,	PUNCT
easat-2167	119	7	horizontal	horizontal	ADJ
easat-2167	119	8	flipping	flipping	NOUN
easat-2167	119	9	was	be	AUX
easat-2167	119	10	employed	employ	VERB
easat-2167	119	11	.	.	PUNCT
easat-2167	120	1	lastly	lastly	ADV
easat-2167	120	2	,	,	PUNCT
easat-2167	120	3	when	when	SCONJ
easat-2167	120	4	these	these	DET
easat-2167	120	5	transformations	transformation	NOUN
easat-2167	120	6	caused	cause	VERB
easat-2167	120	7	pixels	pixel	NOUN
easat-2167	120	8	to	to	PART
easat-2167	120	9	move	move	VERB
easat-2167	120	10	beyond	beyond	ADP
easat-2167	120	11	the	the	DET
easat-2167	120	12	image	image	NOUN
easat-2167	120	13	boundaries	boundary	NOUN
easat-2167	120	14	,	,	PUNCT
easat-2167	120	15	the	the	DET
easat-2167	120	16	missing	miss	VERB
easat-2167	120	17	areas	area	NOUN
easat-2167	120	18	were	be	AUX
easat-2167	120	19	filled	fill	VERB
easat-2167	120	20	using	use	VERB
easat-2167	120	21	the	the	DET
easat-2167	120	22	nearest	near	ADJ
easat-2167	120	23	fill	fill	NOUN
easat-2167	120	24	mode	mode	NOUN
easat-2167	120	25	,	,	PUNCT
easat-2167	120	26	ensuring	ensure	VERB
easat-2167	120	27	continuity	continuity	NOUN
easat-2167	120	28	in	in	ADP
easat-2167	120	29	the	the	DET
easat-2167	120	30	augmented	augment	VERB
easat-2167	120	31	images.4	images.4	PROPN
easat-2167	120	32	.	.	PROPN
easat-2167	120	33	batch	batch	NOUN
easat-2167	120	34	generation	generation	NOUN
easat-2167	120	35	for	for	ADP
easat-2167	120	36	training	training	NOUN
easat-2167	120	37	.	.	PUNCT
easat-2167	121	1	•	•	NOUN
easat-2167	121	2	batch	batch	NOUN
easat-2167	121	3	size	size	NOUN
easat-2167	121	4	:	:	PUNCT
easat-2167	121	5	the	the	DET
easat-2167	121	6	images	image	NOUN
easat-2167	121	7	and	and	CCONJ
easat-2167	121	8	masks	mask	NOUN
easat-2167	121	9	were	be	AUX
easat-2167	121	10	processed	process	VERB
easat-2167	121	11	in	in	ADP
easat-2167	121	12	batches	batch	NOUN
easat-2167	121	13	of	of	ADP
easat-2167	121	14	8	8	NUM
easat-2167	121	15	,	,	PUNCT
easat-2167	121	16	which	which	PRON
easat-2167	121	17	balances	balance	VERB
easat-2167	121	18	the	the	DET
easat-2167	121	19	memory	memory	NOUN
easat-2167	121	20	requirements	requirement	NOUN
easat-2167	121	21	and	and	CCONJ
easat-2167	121	22	the	the	DET
easat-2167	121	23	convergence	convergence	NOUN
easat-2167	121	24	speed	speed	NOUN
easat-2167	121	25	during	during	ADP
easat-2167	121	26	training	training	NOUN
easat-2167	121	27	.	.	PUNCT
easat-2167	122	1	•	•	NUM
easat-2167	122	2	generator	generator	NOUN
easat-2167	122	3	setup	setup	NOUN
easat-2167	122	4	:	:	PUNCT
easat-2167	122	5	two	two	NUM
easat-2167	122	6	separate	separate	ADJ
easat-2167	122	7	imagedatagenerator	imagedatagenerator	NOUN
easat-2167	122	8	objects	object	NOUN
easat-2167	122	9	were	be	AUX
easat-2167	122	10	used	use	VERB
easat-2167	122	11	,	,	PUNCT
easat-2167	122	12	one	one	NUM
easat-2167	122	13	for	for	ADP
easat-2167	122	14	images	image	NOUN
easat-2167	122	15	and	and	CCONJ
easat-2167	122	16	one	one	NUM
easat-2167	122	17	for	for	ADP
easat-2167	122	18	masks	mask	NOUN
easat-2167	122	19	,	,	PUNCT
easat-2167	122	20	both	both	PRON
easat-2167	122	21	applying	apply	VERB
easat-2167	122	22	the	the	DET
easat-2167	122	23	same	same	ADJ
easat-2167	122	24	augmentation	augmentation	NOUN
easat-2167	122	25	transformations	transformation	NOUN
easat-2167	122	26	to	to	PART
easat-2167	122	27	ensure	ensure	VERB
easat-2167	122	28	that	that	SCONJ
easat-2167	122	29	the	the	DET
easat-2167	122	30	corresponding	correspond	VERB
easat-2167	122	31	image	image	NOUN
easat-2167	122	32	and	and	CCONJ
easat-2167	122	33	mask	mask	NOUN
easat-2167	122	34	pairs	pair	NOUN
easat-2167	122	35	remain	remain	VERB
easat-2167	122	36	aligned	aligned	ADJ
easat-2167	122	37	.	.	PUNCT
easat-2167	123	1	the	the	DET
easat-2167	123	2	generators	generator	NOUN
easat-2167	123	3	were	be	AUX
easat-2167	123	4	then	then	ADV
easat-2167	123	5	zipped	zip	VERB
easat-2167	123	6	together	together	ADV
easat-2167	123	7	to	to	PART
easat-2167	123	8	create	create	VERB
easat-2167	123	9	batches	batch	NOUN
easat-2167	123	10	of	of	ADP
easat-2167	123	11	augmented	augment	VERB
easat-2167	123	12	images	image	NOUN
easat-2167	123	13	and	and	CCONJ
easat-2167	123	14	their	their	PRON
easat-2167	123	15	corresponding	correspond	VERB
easat-2167	123	16	masks	mask	NOUN
easat-2167	123	17	for	for	ADP
easat-2167	123	18	training	training	NOUN
easat-2167	123	19	.	.	PUNCT
easat-2167	124	1	validation	validation	NOUN
easat-2167	124	2	data	datum	NOUN
easat-2167	124	3	preparation	preparation	NOUN
easat-2167	124	4	for	for	ADP
easat-2167	124	5	validation	validation	NOUN
easat-2167	124	6	,	,	PUNCT
easat-2167	124	7	the	the	DET
easat-2167	124	8	images	image	NOUN
easat-2167	124	9	and	and	CCONJ
easat-2167	124	10	masks	mask	NOUN
easat-2167	124	11	were	be	AUX
easat-2167	124	12	not	not	PART
easat-2167	124	13	augmented	augment	VERB
easat-2167	124	14	.	.	PUNCT
easat-2167	125	1	this	this	PRON
easat-2167	125	2	ensures	ensure	VERB
easat-2167	125	3	that	that	SCONJ
easat-2167	125	4	the	the	DET
easat-2167	125	5	validation	validation	NOUN
easat-2167	125	6	set	set	NOUN
easat-2167	125	7	remains	remain	VERB
easat-2167	125	8	a	a	DET
easat-2167	125	9	reliable	reliable	ADJ
easat-2167	125	10	measure	measure	NOUN
easat-2167	125	11	of	of	ADP
easat-2167	125	12	how	how	SCONJ
easat-2167	125	13	well	well	ADV
easat-2167	125	14	the	the	DET
easat-2167	125	15	model	model	NOUN
easat-2167	125	16	is	be	AUX
easat-2167	125	17	likely	likely	ADJ
easat-2167	125	18	to	to	PART
easat-2167	125	19	perform	perform	VERB
easat-2167	125	20	on	on	ADP
easat-2167	125	21	unseen	unseen	ADJ
easat-2167	125	22	,	,	PUNCT
easat-2167	125	23	real	real	ADJ
easat-2167	125	24	-	-	PUNCT
easat-2167	125	25	world	world	NOUN
easat-2167	125	26	data	datum	NOUN
easat-2167	125	27	without	without	ADP
easat-2167	125	28	artificial	artificial	ADJ
easat-2167	125	29	alterations	alteration	NOUN
easat-2167	125	30	.	.	PUNCT
easat-2167	126	1	6	6	NUM
easat-2167	126	2	.	.	NUM
easat-2167	126	3	proposed	propose	VERB
easat-2167	126	4	architectures	architecture	NOUN
easat-2167	126	5	algorithm	algorithm	NOUN
easat-2167	126	6	:	:	PUNCT
easat-2167	126	7	u	u	NOUN
easat-2167	126	8	-	-	NOUN
easat-2167	126	9	net	net	ADJ
easat-2167	126	10	with	with	ADP
easat-2167	126	11	mobilenetv2	mobilenetv2	PROPN
easat-2167	126	12	encoder	encoder	NOUN
easat-2167	126	13	and	and	CCONJ
easat-2167	126	14	pix2pix	pix2pix	ADJ
easat-2167	126	15	decoder	decoder	NOUN
easat-2167	126	16	[	[	X
easat-2167	126	17	22	22	NUM
easat-2167	126	18	]	]	X
easat-2167	127	1	[	[	X
easat-2167	127	2	23	23	NUM
easat-2167	127	3	]	]	SYM
easat-2167	127	4	1	1	NUM
easat-2167	127	5	.	.	PUNCT
easat-2167	127	6	input	input	PROPN
easat-2167	127	7	initialization	initialization	NOUN
easat-2167	127	8	:	:	PUNCT
easat-2167	127	9	o	o	NOUN
easat-2167	127	10	define	define	VERB
easat-2167	127	11	the	the	DET
easat-2167	127	12	input	input	NOUN
easat-2167	127	13	size	size	NOUN
easat-2167	127	14	for	for	ADP
easat-2167	127	15	the	the	DET
easat-2167	127	16	model	model	NOUN
easat-2167	127	17	as	as	ADP
easat-2167	127	18	(	(	PUNCT
easat-2167	127	19	128	128	NUM
easat-2167	127	20	,	,	PUNCT
easat-2167	127	21	128	128	NUM
easat-2167	127	22	,	,	PUNCT
easat-2167	127	23	3	3	NUM
easat-2167	127	24	)	)	PUNCT
easat-2167	127	25	.	.	PUNCT
easat-2167	128	1	2	2	X
easat-2167	128	2	.	.	X
easat-2167	128	3	encoder	encoder	NOUN
easat-2167	128	4	(	(	PUNCT
easat-2167	128	5	feature	feature	NOUN
easat-2167	128	6	extraction	extraction	NOUN
easat-2167	128	7	):	):	PUNCT
easat-2167	128	8	o	o	NOUN
easat-2167	128	9	load	load	NOUN
easat-2167	128	10	the	the	DET
easat-2167	128	11	mobilenetv2	mobilenetv2	PROPN
easat-2167	128	12	model	model	NOUN
easat-2167	128	13	pre	pre	VERB
easat-2167	128	14	-	-	VERB
easat-2167	128	15	trained	train	VERB
easat-2167	128	16	on	on	ADP
easat-2167	128	17	imagenet	imagenet	NOUN
easat-2167	128	18	with	with	ADP
easat-2167	128	19	include_top	include_top	NOUN
easat-2167	128	20	=	=	ADJ
easat-2167	128	21	false	false	ADJ
easat-2167	128	22	to	to	PART
easat-2167	128	23	exclude	exclude	VERB
easat-2167	128	24	the	the	DET
easat-2167	128	25	final	final	ADJ
easat-2167	128	26	fully	fully	ADV
easat-2167	128	27	connected	connect	VERB
easat-2167	128	28	layer	layer	NOUN
easat-2167	128	29	.	.	PUNCT
easat-2167	129	1	o	o	NOUN
easat-2167	129	2	set	set	VERB
easat-2167	129	3	the	the	DET
easat-2167	129	4	input	input	NOUN
easat-2167	129	5	shape	shape	NOUN
easat-2167	129	6	to	to	PART
easat-2167	129	7	match	match	VERB
easat-2167	129	8	the	the	DET
easat-2167	129	9	specified	specify	VERB
easat-2167	129	10	input	input	NOUN
easat-2167	129	11	size	size	NOUN
easat-2167	129	12	.	.	PUNCT
easat-2167	130	1	o	o	PROPN
easat-2167	130	2	identify	identify	VERB
easat-2167	130	3	key	key	ADJ
easat-2167	130	4	layers	layer	NOUN
easat-2167	130	5	from	from	ADP
easat-2167	130	6	mobilenetv2	mobilenetv2	PROPN
easat-2167	130	7	to	to	PART
easat-2167	130	8	use	use	VERB
easat-2167	130	9	as	as	ADV
easat-2167	130	10	skip	skip	ADJ
easat-2167	130	11	connections	connection	NOUN
easat-2167	130	12	:	:	PUNCT
easat-2167	130	13	▪	▪	ADJ
easat-2167	130	14	block_1_expand_relu	block_1_expand_relu	NOUN
easat-2167	130	15	(	(	PUNCT
easat-2167	130	16	output	output	NOUN
easat-2167	130	17	size	size	NOUN
easat-2167	130	18	:	:	PUNCT
easat-2167	130	19	64x64	64x64	NUM
easat-2167	130	20	)	)	PUNCT
easat-2167	130	21	▪	▪	X
easat-2167	130	22	block_3_expand_relu	block_3_expand_relu	NOUN
easat-2167	130	23	(	(	PUNCT
easat-2167	130	24	output	output	NOUN
easat-2167	130	25	size	size	NOUN
easat-2167	130	26	:	:	PUNCT
easat-2167	130	27	32x32	32x32	NUM
easat-2167	130	28	)	)	PUNCT
easat-2167	130	29	▪	▪	NOUN
easat-2167	130	30	block_6_expand_relu	block_6_expand_relu	NOUN
easat-2167	130	31	(	(	PUNCT
easat-2167	130	32	output	output	NOUN
easat-2167	130	33	size	size	NOUN
easat-2167	130	34	:	:	PUNCT
easat-2167	130	35	16x16	16x16	NUM
easat-2167	130	36	)	)	PUNCT
easat-2167	130	37	799	799	NUM
easat-2167	130	38	edelweiss	edelweiss	PROPN
easat-2167	130	39	applied	apply	VERB
easat-2167	130	40	science	science	NOUN
easat-2167	130	41	and	and	CCONJ
easat-2167	130	42	technology	technology	NOUN
easat-2167	130	43	issn	issn	PROPN
easat-2167	130	44	:	:	PUNCT
easat-2167	130	45	2576	2576	NUM
easat-2167	130	46	-	-	SYM
easat-2167	130	47	8484	8484	NUM
easat-2167	130	48	vol	vol	NOUN
easat-2167	130	49	.	.	PROPN
easat-2167	130	50	8	8	NUM
easat-2167	130	51	,	,	PUNCT
easat-2167	130	52	no	no	INTJ
easat-2167	130	53	.	.	NOUN
easat-2167	131	1	6	6	NUM
easat-2167	131	2	:	:	PUNCT
easat-2167	131	3	793	793	NUM
easat-2167	131	4	-	-	NUM
easat-2167	131	5	813	813	NUM
easat-2167	131	6	,	,	PUNCT
easat-2167	131	7	2024	2024	NUM
easat-2167	131	8	doi	doi	NOUN
easat-2167	131	9	:	:	PUNCT
easat-2167	131	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	131	11	©	©	NOUN
easat-2167	131	12	2024	2024	NUM
easat-2167	131	13	by	by	ADP
easat-2167	131	14	the	the	DET
easat-2167	131	15	authors	author	NOUN
easat-2167	131	16	;	;	PUNCT
easat-2167	131	17	licensee	licensee	PROPN
easat-2167	131	18	learning	learning	NOUN
easat-2167	131	19	gate	gate	NOUN
easat-2167	131	20	▪	▪	ADV
easat-2167	131	21	block_13_expand_relu	block_13_expand_relu	PROPN
easat-2167	131	22	(	(	PUNCT
easat-2167	131	23	output	output	NOUN
easat-2167	131	24	size	size	NOUN
easat-2167	131	25	:	:	PUNCT
easat-2167	131	26	8x8	8x8	NUM
easat-2167	131	27	)	)	PUNCT
easat-2167	131	28	▪	▪	X
easat-2167	131	29	block_16_project	block_16_project	X
easat-2167	131	30	(	(	PUNCT
easat-2167	131	31	output	output	NOUN
easat-2167	131	32	size	size	NOUN
easat-2167	131	33	:	:	PUNCT
easat-2167	131	34	4x4	4x4	NUM
easat-2167	131	35	)	)	PUNCT
easat-2167	131	36	•	•	NOUN
easat-2167	131	37	extract	extract	VERB
easat-2167	131	38	the	the	DET
easat-2167	131	39	outputs	output	NOUN
easat-2167	131	40	of	of	ADP
easat-2167	131	41	these	these	DET
easat-2167	131	42	layers	layer	NOUN
easat-2167	131	43	to	to	PART
easat-2167	131	44	form	form	VERB
easat-2167	131	45	the	the	DET
easat-2167	131	46	encoder	encoder	NOUN
easat-2167	131	47	's	's	PART
easat-2167	131	48	outputs	output	NOUN
easat-2167	131	49	.	.	PUNCT
easat-2167	132	1	3	3	X
easat-2167	132	2	.	.	X
easat-2167	132	3	define	define	VERB
easat-2167	132	4	the	the	DET
easat-2167	132	5	upsampling	upsampling	NOUN
easat-2167	132	6	path	path	NOUN
easat-2167	132	7	(	(	PUNCT
easat-2167	132	8	decoder	decoder	NOUN
easat-2167	132	9	):	):	PUNCT
easat-2167	132	10	•	•	VERB
easat-2167	132	11	create	create	VERB
easat-2167	132	12	an	an	DET
easat-2167	132	13	upsampling	upsample	VERB
easat-2167	132	14	path	path	NOUN
easat-2167	132	15	using	use	VERB
easat-2167	132	16	the	the	DET
easat-2167	132	17	pix2pix	pix2pix	ADJ
easat-2167	132	18	upsampling	upsampling	NOUN
easat-2167	132	19	technique	technique	NOUN
easat-2167	132	20	:	:	PUNCT
easat-2167	132	21	▪	▪	NOUN
easat-2167	132	22	apply	apply	VERB
easat-2167	132	23	4	4	NUM
easat-2167	132	24	upsampling	upsample	VERB
easat-2167	132	25	layers	layer	NOUN
easat-2167	132	26	:	:	PUNCT
easat-2167	132	27	1	1	X
easat-2167	132	28	.	.	X
easat-2167	132	29	upsample	upsample	VERB
easat-2167	132	30	to	to	ADP
easat-2167	132	31	transition	transition	NOUN
easat-2167	132	32	from	from	ADP
easat-2167	132	33	4x4	4x4	NUM
easat-2167	132	34	to	to	ADP
easat-2167	132	35	8x8	8x8	NUM
easat-2167	132	36	.	.	PUNCT
easat-2167	133	1	2	2	NUM
easat-2167	133	2	.	.	X
easat-2167	133	3	upsample	upsample	VERB
easat-2167	133	4	to	to	ADP
easat-2167	133	5	transition	transition	NOUN
easat-2167	133	6	from	from	ADP
easat-2167	133	7	8x8	8x8	NUM
easat-2167	133	8	to	to	ADP
easat-2167	133	9	16x16	16x16	NUM
easat-2167	133	10	.	.	PUNCT
easat-2167	134	1	3	3	X
easat-2167	134	2	.	.	X
easat-2167	134	3	upsample	upsample	VERB
easat-2167	134	4	to	to	ADP
easat-2167	134	5	transition	transition	NOUN
easat-2167	134	6	from	from	ADP
easat-2167	134	7	16x16	16x16	NUM
easat-2167	134	8	to	to	ADP
easat-2167	134	9	32x32	32x32	NUM
easat-2167	134	10	.	.	NOUN
easat-2167	135	1	4	4	NUM
easat-2167	135	2	.	.	X
easat-2167	135	3	upsample	upsample	VERB
easat-2167	135	4	to	to	ADP
easat-2167	135	5	transition	transition	NOUN
easat-2167	135	6	from	from	ADP
easat-2167	135	7	32x32	32x32	NUM
easat-2167	135	8	to	to	ADP
easat-2167	135	9	64x64	64x64	NUM
easat-2167	135	10	.	.	PUNCT
easat-2167	136	1	•	•	NUM
easat-2167	136	2	each	each	DET
easat-2167	136	3	upsampling	upsample	VERB
easat-2167	136	4	layer	layer	NOUN
easat-2167	136	5	increases	increase	VERB
easat-2167	136	6	the	the	DET
easat-2167	136	7	spatial	spatial	ADJ
easat-2167	136	8	resolution	resolution	NOUN
easat-2167	136	9	while	while	SCONJ
easat-2167	136	10	reducing	reduce	VERB
easat-2167	136	11	the	the	DET
easat-2167	136	12	number	number	NOUN
easat-2167	136	13	of	of	ADP
easat-2167	136	14	feature	feature	NOUN
easat-2167	136	15	channels	channel	NOUN
easat-2167	136	16	.	.	PUNCT
easat-2167	137	1	4	4	X
easat-2167	137	2	.	.	X
easat-2167	137	3	forward	forward	ADV
easat-2167	137	4	pass	pass	VERB
easat-2167	137	5	through	through	ADP
easat-2167	137	6	the	the	DET
easat-2167	137	7	encoder	encoder	NOUN
easat-2167	137	8	:	:	PUNCT
easat-2167	137	9	•	•	ADP
easat-2167	137	10	pass	pass	VERB
easat-2167	137	11	the	the	DET
easat-2167	137	12	input	input	NOUN
easat-2167	137	13	through	through	ADP
easat-2167	137	14	the	the	DET
easat-2167	137	15	encoder	encoder	NOUN
easat-2167	137	16	to	to	PART
easat-2167	137	17	extract	extract	VERB
easat-2167	137	18	features	feature	NOUN
easat-2167	137	19	at	at	ADP
easat-2167	137	20	multiple	multiple	ADJ
easat-2167	137	21	scales	scale	NOUN
easat-2167	137	22	.	.	PUNCT
easat-2167	138	1	•	•	NOUN
easat-2167	138	2	save	save	VERB
easat-2167	138	3	the	the	DET
easat-2167	138	4	outputs	output	NOUN
easat-2167	138	5	from	from	ADP
easat-2167	138	6	the	the	DET
easat-2167	138	7	specified	specified	ADJ
easat-2167	138	8	layers	layer	NOUN
easat-2167	138	9	for	for	ADP
easat-2167	138	10	later	later	ADJ
easat-2167	138	11	use	use	NOUN
easat-2167	138	12	in	in	ADP
easat-2167	138	13	skip	skip	ADJ
easat-2167	138	14	connections	connection	NOUN
easat-2167	138	15	.	.	PUNCT
easat-2167	139	1	5	5	X
easat-2167	139	2	.	.	X
easat-2167	139	3	forward	forward	ADV
easat-2167	139	4	pass	pass	VERB
easat-2167	139	5	through	through	ADP
easat-2167	139	6	the	the	DET
easat-2167	139	7	decoder	decoder	NOUN
easat-2167	139	8	:	:	PUNCT
easat-2167	139	9	•	•	PUNCT
easat-2167	139	10	initialize	initialize	VERB
easat-2167	139	11	the	the	DET
easat-2167	139	12	decoder	decoder	NOUN
easat-2167	139	13	with	with	ADP
easat-2167	139	14	the	the	DET
easat-2167	139	15	last	last	ADJ
easat-2167	139	16	output	output	NOUN
easat-2167	139	17	from	from	ADP
easat-2167	139	18	the	the	DET
easat-2167	139	19	encoder	encoder	NOUN
easat-2167	139	20	(	(	PUNCT
easat-2167	139	21	smallest	small	ADJ
easat-2167	139	22	resolution	resolution	NOUN
easat-2167	139	23	)	)	PUNCT
easat-2167	139	24	.	.	PUNCT
easat-2167	140	1	•	•	NUM
easat-2167	140	2	sequentially	sequentially	ADV
easat-2167	140	3	pass	pass	VERB
easat-2167	140	4	through	through	ADP
easat-2167	140	5	each	each	DET
easat-2167	140	6	upsampling	upsampling	NOUN
easat-2167	140	7	layer	layer	NOUN
easat-2167	140	8	in	in	ADP
easat-2167	140	9	the	the	DET
easat-2167	140	10	decoder	decoder	NOUN
easat-2167	140	11	:	:	PUNCT
easat-2167	140	12	▪	▪	X
easat-2167	140	13	concatenate	concatenate	VERB
easat-2167	140	14	the	the	DET
easat-2167	140	15	upsampled	upsample	VERB
easat-2167	140	16	output	output	NOUN
easat-2167	140	17	with	with	ADP
easat-2167	140	18	the	the	DET
easat-2167	140	19	corresponding	correspond	VERB
easat-2167	140	20	encoder	encoder	NOUN
easat-2167	140	21	output	output	NOUN
easat-2167	140	22	from	from	ADP
easat-2167	140	23	the	the	DET
easat-2167	140	24	skip	skip	ADJ
easat-2167	140	25	connections	connection	NOUN
easat-2167	140	26	.	.	PUNCT
easat-2167	141	1	6	6	X
easat-2167	141	2	.	.	X
easat-2167	141	3	final	final	ADJ
easat-2167	141	4	output	output	NOUN
easat-2167	141	5	layer	layer	NOUN
easat-2167	141	6	:	:	PUNCT
easat-2167	141	7	•	•	X
easat-2167	141	8	apply	apply	VERB
easat-2167	141	9	a	a	DET
easat-2167	141	10	conv2dtranspose	conv2dtranspose	NOUN
easat-2167	141	11	layer	layer	NOUN
easat-2167	141	12	with	with	ADP
easat-2167	141	13	a	a	DET
easat-2167	141	14	stride	stride	NOUN
easat-2167	141	15	of	of	ADP
easat-2167	141	16	2	2	NUM
easat-2167	141	17	to	to	PART
easat-2167	141	18	double	double	VERB
easat-2167	141	19	the	the	DET
easat-2167	141	20	spatial	spatial	ADJ
easat-2167	141	21	resolution	resolution	NOUN
easat-2167	141	22	from	from	ADP
easat-2167	141	23	64x64	64x64	NUM
easat-2167	141	24	to	to	ADP
easat-2167	141	25	128x128	128x128	NUM
easat-2167	141	26	.	.	PUNCT
easat-2167	142	1	•	•	NUM
easat-2167	142	2	use	use	VERB
easat-2167	142	3	a	a	DET
easat-2167	142	4	sigmoid	sigmoid	NOUN
easat-2167	142	5	activation	activation	NOUN
easat-2167	142	6	function	function	NOUN
easat-2167	142	7	on	on	ADP
easat-2167	142	8	the	the	DET
easat-2167	142	9	final	final	ADJ
easat-2167	142	10	output	output	NOUN
easat-2167	142	11	to	to	PART
easat-2167	142	12	predict	predict	VERB
easat-2167	142	13	pixel	pixel	ADJ
easat-2167	142	14	-	-	ADJ
easat-2167	142	15	wise	wise	ADJ
easat-2167	142	16	probabilities	probability	NOUN
easat-2167	142	17	for	for	ADP
easat-2167	142	18	binary	binary	ADJ
easat-2167	142	19	segmentation	segmentation	NOUN
easat-2167	142	20	.	.	PUNCT
easat-2167	143	1	7	7	X
easat-2167	143	2	.	.	PUNCT
easat-2167	143	3	model	model	NOUN
easat-2167	143	4	compilation	compilation	NOUN
easat-2167	143	5	:	:	PUNCT
easat-2167	143	6	•	•	ADP
easat-2167	143	7	compile	compile	VERB
easat-2167	143	8	the	the	DET
easat-2167	143	9	model	model	NOUN
easat-2167	143	10	with	with	ADP
easat-2167	143	11	the	the	DET
easat-2167	143	12	adam	adam	PROPN
easat-2167	143	13	optimizer	optimizer	NOUN
easat-2167	143	14	.	.	PUNCT
easat-2167	144	1	•	•	NUM
easat-2167	144	2	use	use	NOUN
easat-2167	144	3	binary_crossentropy	binary_crossentropy	VERB
easat-2167	144	4	as	as	ADP
easat-2167	144	5	the	the	DET
easat-2167	144	6	loss	loss	NOUN
easat-2167	144	7	function	function	NOUN
easat-2167	144	8	,	,	PUNCT
easat-2167	144	9	appropriate	appropriate	ADJ
easat-2167	144	10	for	for	ADP
easat-2167	144	11	binary	binary	ADJ
easat-2167	144	12	segmentation	segmentation	NOUN
easat-2167	144	13	tasks	task	NOUN
easat-2167	144	14	.	.	PUNCT
easat-2167	145	1	•	•	NUM
easat-2167	145	2	include	include	VERB
easat-2167	145	3	accuracy	accuracy	NOUN
easat-2167	145	4	as	as	ADP
easat-2167	145	5	a	a	DET
easat-2167	145	6	metric	metric	NOUN
easat-2167	145	7	to	to	PART
easat-2167	145	8	monitor	monitor	VERB
easat-2167	145	9	during	during	ADP
easat-2167	145	10	training	training	NOUN
easat-2167	145	11	.	.	PUNCT
easat-2167	146	1	8	8	X
easat-2167	146	2	.	.	PUNCT
easat-2167	146	3	model	model	NOUN
easat-2167	146	4	summary	summary	NOUN
easat-2167	146	5	:	:	PUNCT
easat-2167	146	6	•	•	NOUN
easat-2167	146	7	print	print	VERB
easat-2167	146	8	a	a	DET
easat-2167	146	9	summary	summary	NOUN
easat-2167	146	10	of	of	ADP
easat-2167	146	11	the	the	DET
easat-2167	146	12	model	model	NOUN
easat-2167	146	13	architecture	architecture	NOUN
easat-2167	146	14	,	,	PUNCT
easat-2167	146	15	showing	show	VERB
easat-2167	146	16	the	the	DET
easat-2167	146	17	layers	layer	NOUN
easat-2167	146	18	and	and	CCONJ
easat-2167	146	19	their	their	PRON
easat-2167	146	20	output	output	NOUN
easat-2167	146	21	shapes	shape	NOUN
easat-2167	146	22	.	.	PUNCT
easat-2167	147	1	this	this	DET
easat-2167	147	2	algorithm	algorithm	NOUN
easat-2167	147	3	describes	describe	VERB
easat-2167	147	4	a	a	DET
easat-2167	147	5	u	u	ADJ
easat-2167	147	6	-	-	ADJ
easat-2167	147	7	net	net	ADJ
easat-2167	147	8	architecture	architecture	NOUN
easat-2167	147	9	for	for	ADP
easat-2167	147	10	image	image	NOUN
easat-2167	147	11	segmentation	segmentation	NOUN
easat-2167	147	12	that	that	PRON
easat-2167	147	13	efficiently	efficiently	ADV
easat-2167	147	14	combines	combine	VERB
easat-2167	147	15	the	the	DET
easat-2167	147	16	lightweight	lightweight	ADJ
easat-2167	147	17	mobilenetv2	mobilenetv2	NOUN
easat-2167	147	18	as	as	ADP
easat-2167	147	19	an	an	DET
easat-2167	147	20	encoder	encoder	NOUN
easat-2167	147	21	with	with	ADP
easat-2167	147	22	a	a	DET
easat-2167	147	23	pix2pix	pix2pix	VERB
easat-2167	147	24	-	-	PUNCT
easat-2167	147	25	inspired	inspire	VERB
easat-2167	147	26	upsampling	upsampling	NOUN
easat-2167	147	27	decoder	decoder	NOUN
easat-2167	147	28	.	.	PUNCT
easat-2167	148	1	the	the	DET
easat-2167	148	2	model	model	NOUN
easat-2167	148	3	is	be	AUX
easat-2167	148	4	optimized	optimize	VERB
easat-2167	148	5	for	for	ADP
easat-2167	148	6	binary	binary	ADJ
easat-2167	148	7	segmentation	segmentation	NOUN
easat-2167	148	8	tasks	task	NOUN
easat-2167	148	9	and	and	CCONJ
easat-2167	148	10	is	be	AUX
easat-2167	148	11	designed	design	VERB
easat-2167	148	12	to	to	PART
easat-2167	148	13	handle	handle	VERB
easat-2167	148	14	inputs	input	NOUN
easat-2167	148	15	with	with	ADP
easat-2167	148	16	a	a	DET
easat-2167	148	17	resolution	resolution	NOUN
easat-2167	148	18	of	of	ADP
easat-2167	148	19	128x128	128x128	NUM
easat-2167	148	20	pixels	pixel	NOUN
easat-2167	148	21	.	.	PUNCT
easat-2167	149	1	the	the	DET
easat-2167	149	2	final	final	ADJ
easat-2167	149	3	output	output	NOUN
easat-2167	149	4	is	be	AUX
easat-2167	149	5	a	a	DET
easat-2167	149	6	segmentation	segmentation	NOUN
easat-2167	149	7	mask	mask	NOUN
easat-2167	149	8	with	with	ADP
easat-2167	149	9	the	the	DET
easat-2167	149	10	same	same	ADJ
easat-2167	149	11	spatial	spatial	ADJ
easat-2167	149	12	resolution	resolution	NOUN
easat-2167	149	13	as	as	ADP
easat-2167	149	14	the	the	DET
easat-2167	149	15	input	input	NOUN
easat-2167	149	16	image	image	NOUN
easat-2167	149	17	.	.	PUNCT
easat-2167	150	1	this	this	DET
easat-2167	150	2	model	model	NOUN
easat-2167	150	3	leverages	leverage	VERB
easat-2167	150	4	the	the	DET
easat-2167	150	5	efficiency	efficiency	NOUN
easat-2167	150	6	of	of	ADP
easat-2167	150	7	mobilenetv2	mobilenetv2	PROPN
easat-2167	150	8	for	for	ADP
easat-2167	150	9	feature	feature	NOUN
easat-2167	150	10	extraction	extraction	NOUN
easat-2167	150	11	while	while	SCONJ
easat-2167	150	12	employing	employ	VERB
easat-2167	150	13	a	a	DET
easat-2167	150	14	u	u	ADJ
easat-2167	150	15	-	-	ADJ
easat-2167	150	16	net	net	ADJ
easat-2167	150	17	architecture	architecture	NOUN
easat-2167	150	18	with	with	ADP
easat-2167	150	19	skip	skip	ADJ
easat-2167	150	20	connections	connection	NOUN
easat-2167	150	21	to	to	PART
easat-2167	150	22	retain	retain	VERB
easat-2167	150	23	spatial	spatial	ADJ
easat-2167	150	24	details	detail	NOUN
easat-2167	150	25	during	during	ADP
easat-2167	150	26	upsampling	upsample	VERB
easat-2167	150	27	.	.	PUNCT
easat-2167	151	1	by	by	ADP
easat-2167	151	2	combining	combine	VERB
easat-2167	151	3	the	the	DET
easat-2167	151	4	strengths	strength	NOUN
easat-2167	151	5	of	of	ADP
easat-2167	151	6	both	both	DET
easat-2167	151	7	architectures	architecture	NOUN
easat-2167	151	8	,	,	PUNCT
easat-2167	151	9	it	it	PRON
easat-2167	151	10	creates	create	VERB
easat-2167	151	11	a	a	DET
easat-2167	151	12	powerful	powerful	ADJ
easat-2167	151	13	tool	tool	NOUN
easat-2167	151	14	for	for	ADP
easat-2167	151	15	segmenting	segment	VERB
easat-2167	151	16	images	image	NOUN
easat-2167	151	17	,	,	PUNCT
easat-2167	151	18	particularly	particularly	ADV
easat-2167	151	19	useful	useful	ADJ
easat-2167	151	20	in	in	ADP
easat-2167	151	21	agricultural	agricultural	ADJ
easat-2167	151	22	applications	application	NOUN
easat-2167	151	23	like	like	ADP
easat-2167	151	24	identifying	identify	VERB
easat-2167	151	25	diseased	diseased	ADJ
easat-2167	151	26	areas	area	NOUN
easat-2167	151	27	in	in	ADP
easat-2167	151	28	rice	rice	NOUN
easat-2167	151	29	leaves	leave	NOUN
easat-2167	151	30	.	.	PUNCT
easat-2167	152	1	this	this	DET
easat-2167	152	2	segmentation	segmentation	NOUN
easat-2167	152	3	model	model	NOUN
easat-2167	152	4	can	can	AUX
easat-2167	152	5	help	help	VERB
easat-2167	152	6	in	in	ADP
easat-2167	152	7	early	early	ADJ
easat-2167	152	8	disease	disease	NOUN
easat-2167	152	9	detection	detection	NOUN
easat-2167	152	10	,	,	PUNCT
easat-2167	152	11	potentially	potentially	ADV
easat-2167	152	12	saving	save	VERB
easat-2167	152	13	crops	crop	NOUN
easat-2167	152	14	from	from	ADP
easat-2167	152	15	severe	severe	ADJ
easat-2167	152	16	damage	damage	NOUN
easat-2167	152	17	and	and	CCONJ
easat-2167	152	18	aiding	aid	VERB
easat-2167	152	19	in	in	ADP
easat-2167	152	20	precision	precision	NOUN
easat-2167	152	21	agriculture	agriculture	NOUN
easat-2167	152	22	.	.	PUNCT
easat-2167	153	1	800	800	NUM
easat-2167	153	2	edelweiss	edelweiss	PROPN
easat-2167	153	3	applied	apply	VERB
easat-2167	153	4	science	science	NOUN
easat-2167	153	5	and	and	CCONJ
easat-2167	153	6	technology	technology	NOUN
easat-2167	153	7	issn	issn	PROPN
easat-2167	153	8	:	:	PUNCT
easat-2167	153	9	2576	2576	NUM
easat-2167	153	10	-	-	SYM
easat-2167	153	11	8484	8484	NUM
easat-2167	153	12	vol	vol	NOUN
easat-2167	153	13	.	.	PROPN
easat-2167	153	14	8	8	NUM
easat-2167	153	15	,	,	PUNCT
easat-2167	153	16	no	no	INTJ
easat-2167	153	17	.	.	NOUN
easat-2167	154	1	6	6	NUM
easat-2167	154	2	:	:	PUNCT
easat-2167	154	3	793	793	NUM
easat-2167	154	4	-	-	NUM
easat-2167	154	5	813	813	NUM
easat-2167	154	6	,	,	PUNCT
easat-2167	154	7	2024	2024	NUM
easat-2167	154	8	doi	doi	NOUN
easat-2167	154	9	:	:	PUNCT
easat-2167	154	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	154	11	©	©	NOUN
easat-2167	154	12	2024	2024	NUM
easat-2167	154	13	by	by	ADP
easat-2167	154	14	the	the	DET
easat-2167	154	15	authors	author	NOUN
easat-2167	154	16	;	;	PUNCT
easat-2167	154	17	licensee	licensee	PROPN
easat-2167	154	18	learning	learning	NOUN
easat-2167	154	19	gate	gate	NOUN
easat-2167	154	20	(	(	PUNCT
easat-2167	154	21	a	a	NOUN
easat-2167	154	22	)	)	PUNCT
easat-2167	154	23	(	(	PUNCT
easat-2167	154	24	b	b	X
easat-2167	154	25	)	)	PUNCT
easat-2167	154	26	figure	figure	NOUN
easat-2167	154	27	2	2	NUM
easat-2167	154	28	.	.	PUNCT
easat-2167	155	1	(	(	PUNCT
easat-2167	155	2	a	a	X
easat-2167	155	3	)	)	PUNCT
easat-2167	155	4	u	u	NOUN
easat-2167	155	5	-	-	NOUN
easat-2167	155	6	net	net	ADJ
easat-2167	155	7	with	with	ADP
easat-2167	155	8	mobilenetv2	mobilenetv2	PROPN
easat-2167	155	9	encoder	encoder	NOUN
easat-2167	155	10	and	and	CCONJ
easat-2167	155	11	pix2pix	pix2pix	ADJ
easat-2167	155	12	decoder	decoder	NOUN
easat-2167	155	13	(	(	PUNCT
easat-2167	155	14	b	b	NOUN
easat-2167	155	15	)	)	PUNCT
easat-2167	155	16	simple	simple	ADJ
easat-2167	155	17	deeplabv3	deeplabv3	NOUN
easat-2167	155	18	+	+	CCONJ
easat-2167	155	19	architecture	architecture	NOUN
easat-2167	155	20	using	use	VERB
easat-2167	155	21	efficientnetb4	efficientnetb4	PROPN
easat-2167	155	22	.	.	PUNCT
easat-2167	156	1	6.1	6.1	NUM
easat-2167	156	2	.	.	PUNCT
easat-2167	156	3	algorithm	algorithm	NOUN
easat-2167	156	4	:	:	PUNCT
easat-2167	156	5	simple	simple	ADJ
easat-2167	156	6	deeplabv3	deeplabv3	NOUN
easat-2167	156	7	+	+	CCONJ
easat-2167	156	8	architecture	architecture	NOUN
easat-2167	156	9	using	use	VERB
easat-2167	156	10	efficientnetb4	efficientnetb4	ADP
easat-2167	156	11	the	the	DET
easat-2167	156	12	function	function	NOUN
easat-2167	156	13	simple_deeplabv3plus(input_shape=(128	simple_deeplabv3plus(input_shape=(128	PROPN
easat-2167	156	14	,	,	PUNCT
easat-2167	156	15	128	128	NUM
easat-2167	156	16	,	,	PUNCT
easat-2167	156	17	3	3	NUM
easat-2167	156	18	)	)	PUNCT
easat-2167	156	19	,	,	PUNCT
easat-2167	156	20	classes=1	classes=1	PROPN
easat-2167	156	21	)	)	PUNCT
easat-2167	156	22	defines	define	VERB
easat-2167	156	23	a	a	DET
easat-2167	156	24	segmentation	segmentation	NOUN
easat-2167	156	25	model	model	NOUN
easat-2167	156	26	inspired	inspire	VERB
easat-2167	156	27	by	by	ADP
easat-2167	156	28	the	the	DET
easat-2167	156	29	deeplabv3	deeplabv3	NOUN
easat-2167	156	30	+	+	CCONJ
easat-2167	156	31	architecture	architecture	NOUN
easat-2167	156	32	.	.	PUNCT
easat-2167	157	1	this	this	DET
easat-2167	157	2	model	model	NOUN
easat-2167	157	3	combines	combine	VERB
easat-2167	157	4	the	the	DET
easat-2167	157	5	efficientnetb4	efficientnetb4	ADJ
easat-2167	157	6	architecture	architecture	NOUN
easat-2167	157	7	as	as	ADP
easat-2167	157	8	a	a	DET
easat-2167	157	9	feature	feature	NOUN
easat-2167	157	10	extractor	extractor	NOUN
easat-2167	157	11	with	with	ADP
easat-2167	157	12	an	an	DET
easat-2167	157	13	upsampling	upsampling	ADJ
easat-2167	157	14	path	path	NOUN
easat-2167	157	15	designed	design	VERB
easat-2167	157	16	to	to	PART
easat-2167	157	17	produce	produce	VERB
easat-2167	157	18	high	high	ADJ
easat-2167	157	19	-	-	PUNCT
easat-2167	157	20	resolution	resolution	NOUN
easat-2167	157	21	segmentation	segmentation	NOUN
easat-2167	157	22	maps	map	NOUN
easat-2167	157	23	.	.	PUNCT
easat-2167	158	1	below	below	ADV
easat-2167	158	2	is	be	AUX
easat-2167	158	3	a	a	DET
easat-2167	158	4	801	801	NUM
easat-2167	158	5	edelweiss	edelweiss	PROPN
easat-2167	158	6	applied	apply	VERB
easat-2167	158	7	science	science	NOUN
easat-2167	158	8	and	and	CCONJ
easat-2167	158	9	technology	technology	NOUN
easat-2167	158	10	issn	issn	PROPN
easat-2167	158	11	:	:	PUNCT
easat-2167	158	12	2576	2576	NUM
easat-2167	158	13	-	-	SYM
easat-2167	158	14	8484	8484	NUM
easat-2167	158	15	vol	vol	NOUN
easat-2167	158	16	.	.	PROPN
easat-2167	158	17	8	8	NUM
easat-2167	158	18	,	,	PUNCT
easat-2167	158	19	no	no	INTJ
easat-2167	158	20	.	.	NOUN
easat-2167	159	1	6	6	NUM
easat-2167	159	2	:	:	PUNCT
easat-2167	159	3	793	793	NUM
easat-2167	159	4	-	-	NUM
easat-2167	159	5	813	813	NUM
easat-2167	159	6	,	,	PUNCT
easat-2167	159	7	2024	2024	NUM
easat-2167	159	8	doi	doi	NOUN
easat-2167	159	9	:	:	PUNCT
easat-2167	159	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	159	11	©	©	NOUN
easat-2167	159	12	2024	2024	NUM
easat-2167	159	13	by	by	ADP
easat-2167	159	14	the	the	DET
easat-2167	159	15	authors	author	NOUN
easat-2167	159	16	;	;	PUNCT
easat-2167	159	17	licensee	licensee	PROPN
easat-2167	159	18	learning	learning	NOUN
easat-2167	159	19	gate	gate	NOUN
easat-2167	159	20	detailed	detailed	ADJ
easat-2167	159	21	explanation	explanation	NOUN
easat-2167	159	22	of	of	ADP
easat-2167	159	23	how	how	SCONJ
easat-2167	159	24	this	this	DET
easat-2167	159	25	model	model	NOUN
easat-2167	159	26	works	work	VERB
easat-2167	159	27	,	,	PUNCT
easat-2167	159	28	along	along	ADP
easat-2167	159	29	with	with	ADP
easat-2167	159	30	the	the	DET
easat-2167	159	31	technical	technical	ADJ
easat-2167	159	32	details	detail	NOUN
easat-2167	159	33	and	and	CCONJ
easat-2167	159	34	its	its	PRON
easat-2167	159	35	potential	potential	ADJ
easat-2167	159	36	applications.[24][25	applications.[24][25	NOUN
easat-2167	159	37	]	]	X
easat-2167	159	38	6.1.1	6.1.1	X
easat-2167	159	39	.	.	PUNCT
easat-2167	160	1	efficientnetb4	efficientnetb4	PROPN
easat-2167	160	2	as	as	ADP
easat-2167	160	3	the	the	DET
easat-2167	160	4	base	base	NOUN
easat-2167	160	5	model	model	NOUN
easat-2167	160	6	(	(	PUNCT
easat-2167	160	7	encoder	encoder	NOUN
easat-2167	160	8	):	):	PUNCT
easat-2167	160	9	•	•	ADP
easat-2167	160	10	efficientnetb4	efficientnetb4	ADJ
easat-2167	160	11	is	be	AUX
easat-2167	160	12	part	part	NOUN
easat-2167	160	13	of	of	ADP
easat-2167	160	14	the	the	DET
easat-2167	160	15	efficientnet	efficientnet	ADJ
easat-2167	160	16	family	family	NOUN
easat-2167	160	17	of	of	ADP
easat-2167	160	18	models	model	NOUN
easat-2167	160	19	,	,	PUNCT
easat-2167	160	20	which	which	PRON
easat-2167	160	21	are	be	AUX
easat-2167	160	22	designed	design	VERB
easat-2167	160	23	to	to	PART
easat-2167	160	24	achieve	achieve	VERB
easat-2167	160	25	high	high	ADJ
easat-2167	160	26	accuracy	accuracy	NOUN
easat-2167	160	27	while	while	SCONJ
easat-2167	160	28	being	be	AUX
easat-2167	160	29	computationally	computationally	ADV
easat-2167	160	30	efficient	efficient	ADJ
easat-2167	160	31	.	.	PUNCT
easat-2167	161	1	these	these	DET
easat-2167	161	2	models	model	NOUN
easat-2167	161	3	scale	scale	VERB
easat-2167	161	4	width	width	NOUN
easat-2167	161	5	,	,	PUNCT
easat-2167	161	6	depth	depth	NOUN
easat-2167	161	7	,	,	PUNCT
easat-2167	161	8	and	and	CCONJ
easat-2167	161	9	resolution	resolution	NOUN
easat-2167	161	10	systematically	systematically	ADV
easat-2167	161	11	to	to	PART
easat-2167	161	12	optimize	optimize	VERB
easat-2167	161	13	performance	performance	NOUN
easat-2167	161	14	.	.	PUNCT
easat-2167	162	1	•	•	NUM
easat-2167	162	2	in	in	ADP
easat-2167	162	3	this	this	DET
easat-2167	162	4	function	function	NOUN
easat-2167	162	5	,	,	PUNCT
easat-2167	162	6	efficientnetb4	efficientnetb4	PROPN
easat-2167	162	7	serves	serve	VERB
easat-2167	162	8	as	as	ADP
easat-2167	162	9	the	the	DET
easat-2167	162	10	encoder	encoder	NOUN
easat-2167	162	11	,	,	PUNCT
easat-2167	162	12	extracting	extract	VERB
easat-2167	162	13	features	feature	NOUN
easat-2167	162	14	from	from	ADP
easat-2167	162	15	the	the	DET
easat-2167	162	16	input	input	NOUN
easat-2167	162	17	image	image	NOUN
easat-2167	162	18	at	at	ADP
easat-2167	162	19	different	different	ADJ
easat-2167	162	20	levels	level	NOUN
easat-2167	162	21	of	of	ADP
easat-2167	162	22	abstraction	abstraction	NOUN
easat-2167	162	23	.	.	PUNCT
easat-2167	163	1	this	this	DET
easat-2167	163	2	model	model	NOUN
easat-2167	163	3	is	be	AUX
easat-2167	163	4	pre	pre	ADJ
easat-2167	163	5	-	-	VERB
easat-2167	163	6	trained	train	VERB
easat-2167	163	7	on	on	ADP
easat-2167	163	8	the	the	DET
easat-2167	163	9	imagenet	imagenet	NOUN
easat-2167	163	10	dataset	dataset	NOUN
easat-2167	163	11	,	,	PUNCT
easat-2167	163	12	providing	provide	VERB
easat-2167	163	13	a	a	DET
easat-2167	163	14	strong	strong	ADJ
easat-2167	163	15	starting	starting	NOUN
easat-2167	163	16	point	point	NOUN
easat-2167	163	17	for	for	ADP
easat-2167	163	18	feature	feature	NOUN
easat-2167	163	19	extraction	extraction	NOUN
easat-2167	163	20	.	.	PUNCT
easat-2167	164	1	base_model	base_model	NOUN
easat-2167	164	2	=	=	SYM
easat-2167	164	3	efficientnetb4(input_shape	efficientnetb4(input_shape	NOUN
easat-2167	164	4	=	=	SYM
easat-2167	164	5	input_shape	input_shape	NOUN
easat-2167	164	6	,	,	PUNCT
easat-2167	164	7	include_top	include_top	NOUN
easat-2167	164	8	=	=	SYM
easat-2167	164	9	false	false	ADJ
easat-2167	164	10	,	,	PUNCT
easat-2167	164	11	weights='imagenet	weights='imagenet	NOUN
easat-2167	164	12	'	'	PUNCT
easat-2167	164	13	)	)	PUNCT
easat-2167	164	14	•	•	ADP
easat-2167	164	15	input_shape	input_shape	NOUN
easat-2167	164	16	=	=	PRON
easat-2167	164	17	input_shape	input_shape	PRON
easat-2167	164	18	:	:	PUNCT
easat-2167	164	19	the	the	DET
easat-2167	164	20	input	input	NOUN
easat-2167	164	21	size	size	NOUN
easat-2167	164	22	is	be	AUX
easat-2167	164	23	specified	specify	VERB
easat-2167	164	24	as	as	ADP
easat-2167	164	25	128x128	128x128	NUM
easat-2167	164	26	pixels	pixel	NOUN
easat-2167	164	27	with	with	ADP
easat-2167	164	28	3	3	NUM
easat-2167	164	29	color	color	NOUN
easat-2167	164	30	channels	channel	NOUN
easat-2167	164	31	(	(	PUNCT
easat-2167	164	32	rgb	rgb	PROPN
easat-2167	164	33	)	)	PUNCT
easat-2167	164	34	.	.	PUNCT
easat-2167	165	1	•	•	NUM
easat-2167	165	2	include_top	include_top	ADV
easat-2167	165	3	=	=	ADJ
easat-2167	165	4	false	false	ADJ
easat-2167	165	5	:	:	PUNCT
easat-2167	165	6	this	this	PRON
easat-2167	165	7	excludes	exclude	VERB
easat-2167	165	8	the	the	DET
easat-2167	165	9	fully	fully	ADV
easat-2167	165	10	connected	connected	ADJ
easat-2167	165	11	layers	layer	NOUN
easat-2167	165	12	,	,	PUNCT
easat-2167	165	13	as	as	SCONJ
easat-2167	165	14	they	they	PRON
easat-2167	165	15	are	be	AUX
easat-2167	165	16	not	not	PART
easat-2167	165	17	needed	need	VERB
easat-2167	165	18	for	for	ADP
easat-2167	165	19	the	the	DET
easat-2167	165	20	segmentation	segmentation	NOUN
easat-2167	165	21	task	task	NOUN
easat-2167	165	22	.	.	PROPN
easat-2167	166	1	•	•	NUM
easat-2167	166	2	weights='imagenet	weights='imagenet	NOUN
easat-2167	166	3	'	'	PUNCT
easat-2167	166	4	:	:	PUNCT
easat-2167	166	5	pre	pre	ADJ
easat-2167	166	6	-	-	ADJ
easat-2167	166	7	trained	trained	ADJ
easat-2167	166	8	weights	weight	NOUN
easat-2167	166	9	are	be	AUX
easat-2167	166	10	loaded	load	VERB
easat-2167	166	11	to	to	PART
easat-2167	166	12	leverage	leverage	VERB
easat-2167	166	13	the	the	DET
easat-2167	166	14	learned	learn	VERB
easat-2167	166	15	features	feature	NOUN
easat-2167	166	16	from	from	ADP
easat-2167	166	17	the	the	DET
easat-2167	166	18	imagenet	imagenet	NOUN
easat-2167	166	19	dataset	dataset	NOUN
easat-2167	166	20	.	.	PUNCT
easat-2167	167	1	•	•	NUM
easat-2167	167	2	output	output	NOUN
easat-2167	167	3	shape	shape	NOUN
easat-2167	167	4	:	:	PUNCT
easat-2167	167	5	the	the	DET
easat-2167	167	6	output	output	NOUN
easat-2167	167	7	from	from	ADP
easat-2167	167	8	efficientnetb4	efficientnetb4	PROPN
easat-2167	167	9	will	will	AUX
easat-2167	167	10	be	be	AUX
easat-2167	167	11	a	a	DET
easat-2167	167	12	feature	feature	NOUN
easat-2167	167	13	map	map	NOUN
easat-2167	167	14	of	of	ADP
easat-2167	167	15	size	size	NOUN
easat-2167	167	16	(	(	PUNCT
easat-2167	167	17	none	none	NOUN
easat-2167	167	18	,	,	PUNCT
easat-2167	167	19	4	4	NUM
easat-2167	167	20	,	,	PUNCT
easat-2167	167	21	4	4	NUM
easat-2167	167	22	,	,	PUNCT
easat-2167	167	23	1792	1792	NUM
easat-2167	167	24	)	)	PUNCT
easat-2167	167	25	,	,	PUNCT
easat-2167	167	26	where	where	SCONJ
easat-2167	167	27	1792	1792	NUM
easat-2167	167	28	is	be	AUX
easat-2167	167	29	the	the	DET
easat-2167	167	30	number	number	NOUN
easat-2167	167	31	of	of	ADP
easat-2167	167	32	feature	feature	NOUN
easat-2167	167	33	channels	channel	NOUN
easat-2167	167	34	.	.	PUNCT
easat-2167	168	1	6.1.2	6.1.2	NUM
easat-2167	168	2	.	.	PUNCT
easat-2167	168	3	upsampling	upsample	VERB
easat-2167	168	4	path	path	NOUN
easat-2167	168	5	(	(	PUNCT
easat-2167	168	6	decoder	decoder	NOUN
easat-2167	168	7	):	):	PUNCT
easat-2167	168	8	•	•	ADP
easat-2167	168	9	the	the	DET
easat-2167	168	10	upsampling	upsampling	NOUN
easat-2167	168	11	path	path	NOUN
easat-2167	168	12	is	be	AUX
easat-2167	168	13	designed	design	VERB
easat-2167	168	14	to	to	PART
easat-2167	168	15	progressively	progressively	ADV
easat-2167	168	16	restore	restore	VERB
easat-2167	168	17	the	the	DET
easat-2167	168	18	spatial	spatial	ADJ
easat-2167	168	19	dimensions	dimension	NOUN
easat-2167	168	20	of	of	ADP
easat-2167	168	21	the	the	DET
easat-2167	168	22	feature	feature	NOUN
easat-2167	168	23	maps	map	NOUN
easat-2167	168	24	back	back	ADV
easat-2167	168	25	to	to	ADP
easat-2167	168	26	the	the	DET
easat-2167	168	27	original	original	ADJ
easat-2167	168	28	input	input	NOUN
easat-2167	168	29	size	size	NOUN
easat-2167	168	30	(	(	PUNCT
easat-2167	168	31	128x128	128x128	NUM
easat-2167	168	32	pixels	pixel	NOUN
easat-2167	168	33	)	)	PUNCT
easat-2167	168	34	while	while	SCONJ
easat-2167	168	35	reducing	reduce	VERB
easat-2167	168	36	the	the	DET
easat-2167	168	37	number	number	NOUN
easat-2167	168	38	of	of	ADP
easat-2167	168	39	channels	channel	NOUN
easat-2167	168	40	.	.	PUNCT
easat-2167	169	1	•	•	NUM
easat-2167	169	2	upsampling	upsample	VERB
easat-2167	169	3	layers	layer	NOUN
easat-2167	170	1	:	:	PUNCT
easat-2167	170	2	the	the	DET
easat-2167	170	3	model	model	NOUN
easat-2167	170	4	uses	use	VERB
easat-2167	170	5	upsampling2d	upsampling2d	PROPN
easat-2167	170	6	layers	layer	NOUN
easat-2167	170	7	to	to	PART
easat-2167	170	8	increase	increase	VERB
easat-2167	170	9	the	the	DET
easat-2167	170	10	spatial	spatial	ADJ
easat-2167	170	11	resolution	resolution	NOUN
easat-2167	170	12	of	of	ADP
easat-2167	170	13	the	the	DET
easat-2167	170	14	feature	feature	NOUN
easat-2167	170	15	maps	map	NOUN
easat-2167	170	16	.	.	PUNCT
easat-2167	171	1	each	each	DET
easat-2167	171	2	upsampling	upsample	VERB
easat-2167	171	3	step	step	NOUN
easat-2167	171	4	doubles	double	VERB
easat-2167	171	5	the	the	DET
easat-2167	171	6	resolution	resolution	NOUN
easat-2167	171	7	.	.	PUNCT
easat-2167	172	1	•	•	NUM
easat-2167	172	2	4x4	4x4	NUM
easat-2167	172	3	to	to	ADP
easat-2167	172	4	8x8	8x8	NUM
easat-2167	172	5	upsampling	upsample	VERB
easat-2167	172	6	:	:	PUNCT
easat-2167	172	7	•	•	NUM
easat-2167	172	8	upsampling2d((2	upsampling2d((2	PROPN
easat-2167	172	9	,	,	PUNCT
easat-2167	172	10	2	2	NUM
easat-2167	172	11	)	)	PUNCT
easat-2167	172	12	)	)	PUNCT
easat-2167	172	13	increases	increase	VERB
easat-2167	172	14	the	the	DET
easat-2167	172	15	resolution	resolution	NOUN
easat-2167	172	16	from	from	ADP
easat-2167	172	17	4x4	4x4	NUM
easat-2167	172	18	to	to	ADP
easat-2167	172	19	8x8	8x8	NUM
easat-2167	172	20	.	.	PUNCT
easat-2167	172	21	•	•	NUM
easat-2167	172	22	conv2d(512	conv2d(512	PROPN
easat-2167	172	23	,	,	PUNCT
easat-2167	172	24	(	(	PUNCT
easat-2167	172	25	3	3	NUM
easat-2167	172	26	,	,	PUNCT
easat-2167	172	27	3	3	NUM
easat-2167	172	28	)	)	PUNCT
easat-2167	172	29	,	,	PUNCT
easat-2167	172	30	padding='same	padding='same	PROPN
easat-2167	172	31	'	'	PART
easat-2167	172	32	,	,	PUNCT
easat-2167	172	33	activation='relu	activation='relu	NOUN
easat-2167	172	34	'	'	PUNCT
easat-2167	172	35	)	)	PUNCT
easat-2167	172	36	applies	apply	VERB
easat-2167	172	37	a	a	DET
easat-2167	172	38	convolution	convolution	NOUN
easat-2167	172	39	with	with	ADP
easat-2167	172	40	512	512	NUM
easat-2167	172	41	filters	filter	NOUN
easat-2167	172	42	,	,	PUNCT
easat-2167	172	43	preserving	preserve	VERB
easat-2167	172	44	the	the	DET
easat-2167	172	45	resolution	resolution	NOUN
easat-2167	172	46	but	but	CCONJ
easat-2167	172	47	refining	refine	VERB
easat-2167	172	48	the	the	DET
easat-2167	172	49	features	feature	NOUN
easat-2167	172	50	.	.	PUNCT
easat-2167	173	1	•	•	NUM
easat-2167	173	2	batchnormalization	batchnormalization	NOUN
easat-2167	173	3	(	(	PUNCT
easat-2167	173	4	)	)	PUNCT
easat-2167	173	5	normalizes	normalize	VERB
easat-2167	173	6	the	the	DET
easat-2167	173	7	outputs	output	NOUN
easat-2167	173	8	to	to	PART
easat-2167	173	9	stabilize	stabilize	VERB
easat-2167	173	10	and	and	CCONJ
easat-2167	173	11	accelerate	accelerate	VERB
easat-2167	173	12	the	the	DET
easat-2167	173	13	training	training	NOUN
easat-2167	173	14	process	process	NOUN
easat-2167	173	15	.	.	PUNCT
easat-2167	174	1	•	•	NUM
easat-2167	174	2	8x8	8x8	NUM
easat-2167	174	3	to	to	ADP
easat-2167	174	4	16x16	16x16	NUM
easat-2167	174	5	upsampling	upsample	VERB
easat-2167	174	6	:	:	PUNCT
easat-2167	174	7	•	•	ADP
easat-2167	174	8	resolution	resolution	NOUN
easat-2167	174	9	is	be	AUX
easat-2167	174	10	increased	increase	VERB
easat-2167	174	11	to	to	ADP
easat-2167	174	12	16x16	16x16	NUM
easat-2167	174	13	,	,	PUNCT
easat-2167	174	14	and	and	CCONJ
easat-2167	174	15	the	the	DET
easat-2167	174	16	number	number	NOUN
easat-2167	174	17	of	of	ADP
easat-2167	174	18	filters	filter	NOUN
easat-2167	174	19	is	be	AUX
easat-2167	174	20	reduced	reduce	VERB
easat-2167	174	21	to	to	ADP
easat-2167	174	22	256	256	NUM
easat-2167	174	23	.	.	NOUN
easat-2167	175	1	•	•	NUM
easat-2167	175	2	16x16	16x16	NUM
easat-2167	175	3	to	to	ADP
easat-2167	175	4	32x32	32x32	NUM
easat-2167	175	5	upsampling	upsample	VERB
easat-2167	175	6	:	:	PUNCT
easat-2167	175	7	•	•	ADP
easat-2167	175	8	resolution	resolution	NOUN
easat-2167	175	9	is	be	AUX
easat-2167	175	10	increased	increase	VERB
easat-2167	175	11	to	to	ADP
easat-2167	175	12	32x32	32x32	NUM
easat-2167	175	13	,	,	PUNCT
easat-2167	175	14	with	with	ADP
easat-2167	175	15	128	128	NUM
easat-2167	175	16	filters	filter	NOUN
easat-2167	175	17	.	.	PUNCT
easat-2167	176	1	•	•	NUM
easat-2167	176	2	32x32	32x32	NUM
easat-2167	176	3	to	to	ADP
easat-2167	176	4	64x64	64x64	NUM
easat-2167	176	5	upsampling	upsample	VERB
easat-2167	176	6	:	:	PUNCT
easat-2167	176	7	•	•	NOUN
easat-2167	176	8	64x64	64x64	NUM
easat-2167	176	9	to	to	ADP
easat-2167	176	10	128x128	128x128	PROPN
easat-2167	176	11	upsampling	upsample	VERB
easat-2167	176	12	:	:	PUNCT
easat-2167	176	13	•	•	NOUN
easat-2167	176	14	finally	finally	ADV
easat-2167	176	15	,	,	PUNCT
easat-2167	176	16	the	the	DET
easat-2167	176	17	resolution	resolution	NOUN
easat-2167	176	18	is	be	AUX
easat-2167	176	19	restored	restore	VERB
easat-2167	176	20	to	to	ADP
easat-2167	176	21	the	the	DET
easat-2167	176	22	original	original	ADJ
easat-2167	176	23	input	input	NOUN
easat-2167	176	24	size	size	NOUN
easat-2167	176	25	of	of	ADP
easat-2167	176	26	128x128	128x128	NUM
easat-2167	176	27	,	,	PUNCT
easat-2167	176	28	with	with	ADP
easat-2167	176	29	32	32	NUM
easat-2167	176	30	filters	filter	NOUN
easat-2167	176	31	.	.	PUNCT
easat-2167	177	1	6.1.3	6.1.3	NUM
easat-2167	177	2	.	.	PUNCT
easat-2167	177	3	final	final	ADJ
easat-2167	177	4	convolutional	convolutional	ADJ
easat-2167	177	5	layer	layer	NOUN
easat-2167	177	6	•	•	ADP
easat-2167	177	7	after	after	ADP
easat-2167	177	8	upsampling	upsample	VERB
easat-2167	177	9	,	,	PUNCT
easat-2167	177	10	a	a	DET
easat-2167	177	11	final	final	ADJ
easat-2167	177	12	convolutional	convolutional	ADJ
easat-2167	177	13	layer	layer	NOUN
easat-2167	177	14	is	be	AUX
easat-2167	177	15	used	use	VERB
easat-2167	177	16	to	to	PART
easat-2167	177	17	produce	produce	VERB
easat-2167	177	18	the	the	DET
easat-2167	177	19	desired	desire	VERB
easat-2167	177	20	number	number	NOUN
easat-2167	177	21	of	of	ADP
easat-2167	177	22	output	output	NOUN
easat-2167	177	23	channels	channel	NOUN
easat-2167	177	24	.	.	PUNCT
easat-2167	178	1	for	for	ADP
easat-2167	178	2	segmentation	segmentation	NOUN
easat-2167	178	3	tasks	task	NOUN
easat-2167	178	4	,	,	PUNCT
easat-2167	178	5	this	this	PRON
easat-2167	178	6	often	often	ADV
easat-2167	178	7	corresponds	correspond	VERB
easat-2167	178	8	to	to	ADP
easat-2167	178	9	the	the	DET
easat-2167	178	10	number	number	NOUN
easat-2167	178	11	of	of	ADP
easat-2167	178	12	classes	class	NOUN
easat-2167	178	13	.	.	PUNCT
easat-2167	179	1	•	•	NUM
easat-2167	179	2	conv2d(classes	conv2d(classe	NOUN
easat-2167	179	3	,	,	PUNCT
easat-2167	179	4	(	(	PUNCT
easat-2167	179	5	1	1	NUM
easat-2167	179	6	,	,	PUNCT
easat-2167	179	7	1	1	NUM
easat-2167	179	8	)	)	PUNCT
easat-2167	179	9	,	,	PUNCT
easat-2167	179	10	padding='same	padding='same	PROPN
easat-2167	179	11	'	'	PUNCT
easat-2167	179	12	)	)	PUNCT
easat-2167	179	13	applies	apply	VERB
easat-2167	179	14	a	a	DET
easat-2167	179	15	1x1	1x1	NUM
easat-2167	179	16	convolution	convolution	NOUN
easat-2167	179	17	to	to	PART
easat-2167	179	18	reduce	reduce	VERB
easat-2167	179	19	the	the	DET
easat-2167	179	20	802	802	NUM
easat-2167	179	21	edelweiss	edelweiss	PROPN
easat-2167	179	22	applied	apply	VERB
easat-2167	179	23	science	science	NOUN
easat-2167	179	24	and	and	CCONJ
easat-2167	179	25	technology	technology	NOUN
easat-2167	179	26	issn	issn	PROPN
easat-2167	179	27	:	:	PUNCT
easat-2167	179	28	2576	2576	NUM
easat-2167	179	29	-	-	SYM
easat-2167	179	30	8484	8484	NUM
easat-2167	179	31	vol	vol	NOUN
easat-2167	179	32	.	.	PROPN
easat-2167	179	33	8	8	NUM
easat-2167	179	34	,	,	PUNCT
easat-2167	179	35	no	no	INTJ
easat-2167	179	36	.	.	NOUN
easat-2167	180	1	6	6	NUM
easat-2167	180	2	:	:	PUNCT
easat-2167	180	3	793	793	NUM
easat-2167	180	4	-	-	NUM
easat-2167	180	5	813	813	NUM
easat-2167	180	6	,	,	PUNCT
easat-2167	180	7	2024	2024	NUM
easat-2167	180	8	doi	doi	NOUN
easat-2167	180	9	:	:	PUNCT
easat-2167	180	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	180	11	©	©	NOUN
easat-2167	180	12	2024	2024	NUM
easat-2167	180	13	by	by	ADP
easat-2167	180	14	the	the	DET
easat-2167	180	15	authors	author	NOUN
easat-2167	180	16	;	;	PUNCT
easat-2167	180	17	licensee	licensee	NOUN
easat-2167	180	18	learning	learn	VERB
easat-2167	180	19	gate	gate	NOUN
easat-2167	180	20	number	number	NOUN
easat-2167	180	21	of	of	ADP
easat-2167	180	22	channels	channel	NOUN
easat-2167	180	23	to	to	ADP
easat-2167	180	24	the	the	DET
easat-2167	180	25	number	number	NOUN
easat-2167	180	26	of	of	ADP
easat-2167	180	27	classes	class	NOUN
easat-2167	180	28	.	.	PUNCT
easat-2167	181	1	in	in	ADP
easat-2167	181	2	this	this	DET
easat-2167	181	3	case	case	NOUN
easat-2167	181	4	,	,	PUNCT
easat-2167	181	5	classes=1	classes=1	PROPN
easat-2167	181	6	,	,	PUNCT
easat-2167	181	7	so	so	ADV
easat-2167	181	8	the	the	DET
easat-2167	181	9	output	output	NOUN
easat-2167	181	10	is	be	AUX
easat-2167	181	11	a	a	DET
easat-2167	181	12	single	single	ADJ
easat-2167	181	13	-	-	PUNCT
easat-2167	181	14	channel	channel	NOUN
easat-2167	181	15	feature	feature	NOUN
easat-2167	181	16	map	map	NOUN
easat-2167	181	17	.	.	PUNCT
easat-2167	182	1	•	•	NOUN
easat-2167	182	2	sigmoid	sigmoid	NOUN
easat-2167	182	3	activation	activation	NOUN
easat-2167	182	4	:	:	PUNCT
easat-2167	182	5	o	o	X
easat-2167	182	6	the	the	DET
easat-2167	182	7	sigmoid	sigmoid	NOUN
easat-2167	182	8	activation	activation	NOUN
easat-2167	182	9	function	function	NOUN
easat-2167	182	10	is	be	AUX
easat-2167	182	11	used	use	VERB
easat-2167	182	12	to	to	PART
easat-2167	182	13	produce	produce	VERB
easat-2167	182	14	output	output	NOUN
easat-2167	182	15	values	value	NOUN
easat-2167	182	16	between	between	ADP
easat-2167	182	17	0	0	NUM
easat-2167	182	18	and	and	CCONJ
easat-2167	182	19	1	1	NUM
easat-2167	182	20	,	,	PUNCT
easat-2167	182	21	which	which	PRON
easat-2167	182	22	is	be	AUX
easat-2167	182	23	suitable	suitable	ADJ
easat-2167	182	24	for	for	ADP
easat-2167	182	25	binary	binary	ADJ
easat-2167	182	26	segmentation	segmentation	NOUN
easat-2167	182	27	tasks	task	NOUN
easat-2167	182	28	.	.	PUNCT
easat-2167	183	1	6.1.4	6.1.4	X
easat-2167	183	2	.	.	PUNCT
easat-2167	183	3	model	model	NOUN
easat-2167	183	4	compilation	compilation	NOUN
easat-2167	183	5	•	•	ADP
easat-2167	183	6	the	the	DET
easat-2167	183	7	model	model	NOUN
easat-2167	183	8	is	be	AUX
easat-2167	183	9	compiled	compile	VERB
easat-2167	183	10	by	by	ADP
easat-2167	183	11	connecting	connect	VERB
easat-2167	183	12	the	the	DET
easat-2167	183	13	input	input	NOUN
easat-2167	183	14	of	of	ADP
easat-2167	183	15	the	the	DET
easat-2167	183	16	efficientnetb4	efficientnetb4	PROPN
easat-2167	183	17	encoder	encoder	NOUN
easat-2167	183	18	to	to	ADP
easat-2167	183	19	the	the	DET
easat-2167	183	20	output	output	NOUN
easat-2167	183	21	of	of	ADP
easat-2167	183	22	the	the	DET
easat-2167	183	23	upsampling	upsampling	NOUN
easat-2167	183	24	decoder	decoder	NOUN
easat-2167	183	25	.	.	PUNCT
easat-2167	184	1	•	•	NOUN
easat-2167	184	2	this	this	DET
easat-2167	184	3	returns	return	VERB
easat-2167	184	4	a	a	DET
easat-2167	184	5	keras	keras	PROPN
easat-2167	184	6	model	model	NOUN
easat-2167	184	7	that	that	PRON
easat-2167	184	8	takes	take	VERB
easat-2167	184	9	an	an	DET
easat-2167	184	10	input	input	NOUN
easat-2167	184	11	image	image	NOUN
easat-2167	184	12	of	of	ADP
easat-2167	184	13	size	size	NOUN
easat-2167	184	14	128x128	128x128	PROPN
easat-2167	184	15	and	and	CCONJ
easat-2167	184	16	produces	produce	VERB
easat-2167	184	17	a	a	DET
easat-2167	184	18	segmentation	segmentation	NOUN
easat-2167	184	19	map	map	NOUN
easat-2167	184	20	of	of	ADP
easat-2167	184	21	the	the	DET
easat-2167	184	22	same	same	ADJ
easat-2167	184	23	size	size	NOUN
easat-2167	184	24	.	.	PUNCT
easat-2167	185	1	6.2	6.2	NUM
easat-2167	185	2	.	.	PUNCT
easat-2167	186	1	technical	technical	ADJ
easat-2167	186	2	summary	summary	NOUN
easat-2167	186	3	and	and	CCONJ
easat-2167	186	4	use	use	VERB
easat-2167	186	5	cases	case	NOUN
easat-2167	186	6	1	1	NUM
easat-2167	186	7	.	.	PUNCT
easat-2167	187	1	efficientnetb4	efficientnetb4	PROPN
easat-2167	187	2	:	:	PUNCT
easat-2167	187	3	efficientnetb4	efficientnetb4	PROPN
easat-2167	187	4	is	be	AUX
easat-2167	187	5	chosen	choose	VERB
easat-2167	187	6	as	as	ADP
easat-2167	187	7	the	the	DET
easat-2167	187	8	backbone	backbone	NOUN
easat-2167	187	9	for	for	ADP
easat-2167	187	10	its	its	PRON
easat-2167	187	11	balance	balance	NOUN
easat-2167	187	12	between	between	ADP
easat-2167	187	13	performance	performance	NOUN
easat-2167	187	14	and	and	CCONJ
easat-2167	187	15	computational	computational	ADJ
easat-2167	187	16	efficiency	efficiency	NOUN
easat-2167	187	17	.	.	PUNCT
easat-2167	188	1	it	it	PRON
easat-2167	188	2	extracts	extract	VERB
easat-2167	188	3	rich	rich	ADJ
easat-2167	188	4	features	feature	NOUN
easat-2167	188	5	from	from	ADP
easat-2167	188	6	input	input	NOUN
easat-2167	188	7	images	image	NOUN
easat-2167	188	8	while	while	SCONJ
easat-2167	188	9	maintaining	maintain	VERB
easat-2167	188	10	a	a	DET
easat-2167	188	11	relatively	relatively	ADV
easat-2167	188	12	small	small	ADJ
easat-2167	188	13	model	model	NOUN
easat-2167	188	14	size	size	NOUN
easat-2167	188	15	.	.	PUNCT
easat-2167	189	1	2	2	X
easat-2167	189	2	.	.	X
easat-2167	189	3	upsampling	upsample	VERB
easat-2167	189	4	path	path	NOUN
easat-2167	189	5	:	:	PUNCT
easat-2167	189	6	the	the	DET
easat-2167	189	7	upsampling	upsample	VERB
easat-2167	189	8	path	path	NOUN
easat-2167	189	9	progressively	progressively	ADV
easat-2167	189	10	restores	restore	VERB
easat-2167	189	11	the	the	DET
easat-2167	189	12	spatial	spatial	ADJ
easat-2167	189	13	resolution	resolution	NOUN
easat-2167	189	14	of	of	ADP
easat-2167	189	15	the	the	DET
easat-2167	189	16	feature	feature	NOUN
easat-2167	189	17	maps	map	NOUN
easat-2167	189	18	,	,	PUNCT
easat-2167	189	19	using	use	VERB
easat-2167	189	20	conv2d	conv2d	PART
easat-2167	189	21	layers	layer	NOUN
easat-2167	189	22	to	to	PART
easat-2167	189	23	refine	refine	VERB
easat-2167	189	24	the	the	DET
easat-2167	189	25	features	feature	NOUN
easat-2167	189	26	at	at	ADP
easat-2167	189	27	each	each	DET
easat-2167	189	28	step	step	NOUN
easat-2167	189	29	.	.	PUNCT
easat-2167	190	1	batchnormalization	batchnormalization	NOUN
easat-2167	190	2	is	be	AUX
easat-2167	190	3	applied	apply	VERB
easat-2167	190	4	after	after	SCONJ
easat-2167	190	5	each	each	DET
easat-2167	190	6	convolution	convolution	NOUN
easat-2167	190	7	to	to	PART
easat-2167	190	8	improve	improve	VERB
easat-2167	190	9	training	train	VERB
easat-2167	190	10	stability	stability	NOUN
easat-2167	190	11	and	and	CCONJ
easat-2167	190	12	model	model	NOUN
easat-2167	190	13	performance	performance	NOUN
easat-2167	190	14	.	.	PUNCT
easat-2167	191	1	3	3	X
easat-2167	191	2	.	.	X
easat-2167	191	3	final	final	ADJ
easat-2167	191	4	output	output	NOUN
easat-2167	191	5	:	:	PUNCT
easat-2167	191	6	the	the	DET
easat-2167	191	7	final	final	ADJ
easat-2167	191	8	convolution	convolution	NOUN
easat-2167	191	9	layer	layer	NOUN
easat-2167	191	10	reduces	reduce	VERB
easat-2167	191	11	the	the	DET
easat-2167	191	12	feature	feature	NOUN
easat-2167	191	13	maps	map	NOUN
easat-2167	191	14	to	to	ADP
easat-2167	191	15	a	a	DET
easat-2167	191	16	single	single	ADJ
easat-2167	191	17	channel	channel	NOUN
easat-2167	191	18	,	,	PUNCT
easat-2167	191	19	and	and	CCONJ
easat-2167	191	20	the	the	DET
easat-2167	191	21	sigmoid	sigmoid	NOUN
easat-2167	191	22	activation	activation	NOUN
easat-2167	191	23	function	function	NOUN
easat-2167	191	24	produces	produce	VERB
easat-2167	191	25	a	a	DET
easat-2167	191	26	probability	probability	NOUN
easat-2167	191	27	map	map	NOUN
easat-2167	191	28	,	,	PUNCT
easat-2167	191	29	indicating	indicate	VERB
easat-2167	191	30	the	the	DET
easat-2167	191	31	presence	presence	NOUN
easat-2167	191	32	or	or	CCONJ
easat-2167	191	33	absence	absence	NOUN
easat-2167	191	34	of	of	ADP
easat-2167	191	35	a	a	DET
easat-2167	191	36	specific	specific	ADJ
easat-2167	191	37	class	class	NOUN
easat-2167	191	38	(	(	PUNCT
easat-2167	191	39	e.g.	e.g.	ADV
easat-2167	191	40	,	,	PUNCT
easat-2167	191	41	diseased	diseased	ADJ
easat-2167	191	42	region	region	NOUN
easat-2167	191	43	)	)	PUNCT
easat-2167	191	44	in	in	ADP
easat-2167	191	45	each	each	DET
easat-2167	191	46	pixel	pixel	NOUN
easat-2167	191	47	.	.	PUNCT
easat-2167	192	1	7	7	X
easat-2167	192	2	.	.	NUM
easat-2167	192	3	proposed	propose	VERB
easat-2167	192	4	efficient	efficient	ADJ
easat-2167	192	5	architecture	architecture	NOUN
easat-2167	192	6	-	-	PUNCT
easat-2167	192	7	u	u	NOUN
easat-2167	192	8	-	-	NOUN
easat-2167	192	9	net	net	NOUN
easat-2167	192	10	with	with	ADP
easat-2167	192	11	efficientnetb7	efficientnetb7	NOUN
easat-2167	192	12	and	and	CCONJ
easat-2167	192	13	attention	attention	NOUN
easat-2167	192	14	gates	gate	VERB
easat-2167	192	15	the	the	DET
easat-2167	192	16	proposed	propose	VERB
easat-2167	192	17	model	model	NOUN
easat-2167	192	18	,	,	PUNCT
easat-2167	192	19	unetefficientnetb7withattention	unetefficientnetb7withattention	NOUN
easat-2167	192	20	,	,	PUNCT
easat-2167	192	21	combines	combine	VERB
easat-2167	192	22	several	several	ADJ
easat-2167	192	23	advanced	advanced	ADJ
easat-2167	192	24	deep	deep	ADJ
easat-2167	192	25	learning	learning	NOUN
easat-2167	192	26	techniques	technique	NOUN
easat-2167	192	27	to	to	PART
easat-2167	192	28	create	create	VERB
easat-2167	192	29	a	a	DET
easat-2167	192	30	powerful	powerful	ADJ
easat-2167	192	31	architecture	architecture	NOUN
easat-2167	192	32	for	for	ADP
easat-2167	192	33	image	image	NOUN
easat-2167	192	34	segmentation	segmentation	NOUN
easat-2167	192	35	tasks	task	NOUN
easat-2167	192	36	.	.	PUNCT
easat-2167	193	1	below	below	ADV
easat-2167	193	2	is	be	AUX
easat-2167	193	3	a	a	DET
easat-2167	193	4	comprehensive	comprehensive	ADJ
easat-2167	193	5	breakdown	breakdown	NOUN
easat-2167	193	6	of	of	ADP
easat-2167	193	7	the	the	DET
easat-2167	193	8	model	model	NOUN
easat-2167	193	9	's	's	PART
easat-2167	193	10	components	component	NOUN
easat-2167	193	11	,	,	PUNCT
easat-2167	193	12	how	how	SCONJ
easat-2167	193	13	they	they	PRON
easat-2167	193	14	work	work	VERB
easat-2167	193	15	together	together	ADV
easat-2167	193	16	,	,	PUNCT
easat-2167	193	17	and	and	CCONJ
easat-2167	193	18	the	the	DET
easat-2167	193	19	significance	significance	NOUN
easat-2167	193	20	of	of	ADP
easat-2167	193	21	each	each	DET
easat-2167	193	22	element.[26][27	element.[26][27	NOUN
easat-2167	193	23	]	]	PUNCT
easat-2167	193	24	.	.	PUNCT
easat-2167	194	1	7.1	7.1	NUM
easat-2167	194	2	.	.	PUNCT
easat-2167	195	1	efficientnetb7	efficientnetb7	NOUN
easat-2167	195	2	as	as	SCONJ
easat-2167	195	3	the	the	DET
easat-2167	195	4	backbone	backbone	NOUN
easat-2167	195	5	(	(	PUNCT
easat-2167	195	6	encoder	encoder	NOUN
easat-2167	195	7	)	)	PUNCT
easat-2167	195	8	the	the	DET
easat-2167	195	9	model	model	NOUN
easat-2167	195	10	utilizes	utilize	VERB
easat-2167	195	11	efficientnetb7	efficientnetb7	NOUN
easat-2167	195	12	as	as	ADP
easat-2167	195	13	the	the	DET
easat-2167	195	14	backbone	backbone	NOUN
easat-2167	195	15	for	for	ADP
easat-2167	195	16	feature	feature	NOUN
easat-2167	195	17	extraction	extraction	NOUN
easat-2167	195	18	.	.	PUNCT
easat-2167	196	1	efficientnetb7	efficientnetb7	NOUN
easat-2167	196	2	is	be	AUX
easat-2167	196	3	part	part	NOUN
easat-2167	196	4	of	of	ADP
easat-2167	196	5	the	the	DET
easat-2167	196	6	efficientnet	efficientnet	NOUN
easat-2167	196	7	family	family	NOUN
easat-2167	196	8	,	,	PUNCT
easat-2167	196	9	which	which	PRON
easat-2167	196	10	is	be	AUX
easat-2167	196	11	known	know	VERB
easat-2167	196	12	for	for	ADP
easat-2167	196	13	achieving	achieve	VERB
easat-2167	196	14	state	state	NOUN
easat-2167	196	15	-	-	PUNCT
easat-2167	196	16	of	of	ADP
easat-2167	196	17	-	-	PUNCT
easat-2167	196	18	the	the	DET
easat-2167	196	19	-	-	PUNCT
easat-2167	196	20	art	art	NOUN
easat-2167	196	21	accuracy	accuracy	NOUN
easat-2167	196	22	with	with	ADP
easat-2167	196	23	fewer	few	ADJ
easat-2167	196	24	parameters	parameter	NOUN
easat-2167	196	25	by	by	ADP
easat-2167	196	26	scaling	scale	VERB
easat-2167	196	27	up	up	ADP
easat-2167	196	28	the	the	DET
easat-2167	196	29	network	network	NOUN
easat-2167	196	30	width	width	NOUN
easat-2167	196	31	,	,	PUNCT
easat-2167	196	32	depth	depth	NOUN
easat-2167	196	33	,	,	PUNCT
easat-2167	196	34	and	and	CCONJ
easat-2167	196	35	resolution	resolution	NOUN
easat-2167	196	36	in	in	ADP
easat-2167	196	37	a	a	DET
easat-2167	196	38	balanced	balanced	ADJ
easat-2167	196	39	manner	manner	NOUN
easat-2167	196	40	.	.	PUNCT
easat-2167	197	1	•	•	NUM
easat-2167	197	2	input_shape	input_shape	NOUN
easat-2167	197	3	=	=	PRON
easat-2167	197	4	input_shape	input_shape	NOUN
easat-2167	197	5	:	:	PUNCT
easat-2167	197	6	specifies	specify	VERB
easat-2167	197	7	the	the	DET
easat-2167	197	8	input	input	NOUN
easat-2167	197	9	size	size	NOUN
easat-2167	197	10	(	(	PUNCT
easat-2167	197	11	128x128x3	128x128x3	NUM
easat-2167	197	12	)	)	PUNCT
easat-2167	197	13	,	,	PUNCT
easat-2167	197	14	where	where	SCONJ
easat-2167	197	15	3	3	NUM
easat-2167	197	16	corresponds	correspond	VERB
easat-2167	197	17	to	to	ADP
easat-2167	197	18	the	the	DET
easat-2167	197	19	rgb	rgb	PROPN
easat-2167	197	20	channels	channel	NOUN
easat-2167	197	21	.	.	PUNCT
easat-2167	198	1	•	•	NUM
easat-2167	198	2	include_top	include_top	ADV
easat-2167	198	3	=	=	SYM
easat-2167	198	4	false	false	ADJ
easat-2167	198	5	:	:	PUNCT
easat-2167	198	6	excludes	exclude	VERB
easat-2167	198	7	the	the	DET
easat-2167	198	8	fully	fully	ADV
easat-2167	198	9	connected	connected	ADJ
easat-2167	198	10	layers	layer	NOUN
easat-2167	198	11	at	at	ADP
easat-2167	198	12	the	the	DET
easat-2167	198	13	top	top	NOUN
easat-2167	198	14	of	of	ADP
easat-2167	198	15	the	the	DET
easat-2167	198	16	network	network	NOUN
easat-2167	198	17	,	,	PUNCT
easat-2167	198	18	as	as	SCONJ
easat-2167	198	19	these	these	PRON
easat-2167	198	20	are	be	AUX
easat-2167	198	21	not	not	PART
easat-2167	198	22	needed	need	VERB
easat-2167	198	23	for	for	ADP
easat-2167	198	24	segmentation	segmentation	NOUN
easat-2167	198	25	.	.	PUNCT
easat-2167	199	1	•	•	NUM
easat-2167	199	2	weights='imagenet	weights='imagenet	NOUN
easat-2167	199	3	'	'	PUNCT
easat-2167	199	4	:	:	PUNCT
easat-2167	199	5	uses	use	VERB
easat-2167	199	6	pre	pre	ADJ
easat-2167	199	7	-	-	VERB
easat-2167	199	8	trained	train	VERB
easat-2167	199	9	weights	weight	NOUN
easat-2167	199	10	from	from	ADP
easat-2167	199	11	imagenet	imagenet	NOUN
easat-2167	199	12	,	,	PUNCT
easat-2167	199	13	which	which	PRON
easat-2167	199	14	helps	help	VERB
easat-2167	199	15	in	in	ADP
easat-2167	199	16	leveraging	leverage	VERB
easat-2167	199	17	features	feature	NOUN
easat-2167	199	18	learned	learn	VERB
easat-2167	199	19	from	from	ADP
easat-2167	199	20	a	a	DET
easat-2167	199	21	large	large	ADJ
easat-2167	199	22	and	and	CCONJ
easat-2167	199	23	diverse	diverse	ADJ
easat-2167	199	24	dataset	dataset	NOUN
easat-2167	199	25	.	.	PUNCT
easat-2167	200	1	7.2	7.2	NUM
easat-2167	200	2	.	.	PUNCT
easat-2167	201	1	skip	skip	ADJ
easat-2167	201	2	connections	connection	NOUN
easat-2167	201	3	skip	skip	ADJ
easat-2167	201	4	connections	connection	NOUN
easat-2167	201	5	are	be	AUX
easat-2167	201	6	extracted	extract	VERB
easat-2167	201	7	from	from	ADP
easat-2167	201	8	specific	specific	ADJ
easat-2167	201	9	layers	layer	NOUN
easat-2167	201	10	of	of	ADP
easat-2167	201	11	efficientnetb7	efficientnetb7	NOUN
easat-2167	201	12	.	.	PUNCT
easat-2167	202	1	these	these	DET
easat-2167	202	2	connections	connection	NOUN
easat-2167	202	3	allow	allow	VERB
easat-2167	202	4	the	the	DET
easat-2167	202	5	model	model	NOUN
easat-2167	202	6	to	to	PART
easat-2167	202	7	retain	retain	VERB
easat-2167	202	8	and	and	CCONJ
easat-2167	202	9	reuse	reuse	VERB
easat-2167	202	10	high	high	ADJ
easat-2167	202	11	-	-	PUNCT
easat-2167	202	12	resolution	resolution	NOUN
easat-2167	202	13	spatial	spatial	ADJ
easat-2167	202	14	information	information	NOUN
easat-2167	202	15	,	,	PUNCT
easat-2167	202	16	which	which	PRON
easat-2167	202	17	is	be	AUX
easat-2167	202	18	crucial	crucial	ADJ
easat-2167	202	19	for	for	ADP
easat-2167	202	20	accurate	accurate	ADJ
easat-2167	202	21	segmentation	segmentation	NOUN
easat-2167	202	22	.	.	PUNCT
easat-2167	203	1	•	•	ADP
easat-2167	203	2	these	these	DET
easat-2167	203	3	layers	layer	NOUN
easat-2167	203	4	are	be	AUX
easat-2167	203	5	selected	select	VERB
easat-2167	203	6	because	because	SCONJ
easat-2167	203	7	they	they	PRON
easat-2167	203	8	represent	represent	VERB
easat-2167	203	9	different	different	ADJ
easat-2167	203	10	stages	stage	NOUN
easat-2167	203	11	of	of	ADP
easat-2167	203	12	down	down	ADJ
easat-2167	203	13	sampling	sample	VERB
easat-2167	203	14	and	and	CCONJ
easat-2167	203	15	feature	feature	NOUN
easat-2167	203	16	extraction	extraction	NOUN
easat-2167	203	17	,	,	PUNCT
easat-2167	203	18	each	each	PRON
easat-2167	203	19	providing	provide	VERB
easat-2167	203	20	a	a	DET
easat-2167	203	21	different	different	ADJ
easat-2167	203	22	level	level	NOUN
easat-2167	203	23	of	of	ADP
easat-2167	203	24	detail	detail	NOUN
easat-2167	203	25	.	.	PUNCT
easat-2167	204	1	7.3	7.3	NUM
easat-2167	204	2	.	.	PUNCT
easat-2167	205	1	decoder	decoder	NOUN
easat-2167	205	2	path	path	NOUN
easat-2167	205	3	with	with	ADP
easat-2167	205	4	attention	attention	NOUN
easat-2167	205	5	gates	gate	NOUN
easat-2167	205	6	803	803	NUM
easat-2167	205	7	edelweiss	edelweiss	PROPN
easat-2167	205	8	applied	apply	VERB
easat-2167	205	9	science	science	NOUN
easat-2167	205	10	and	and	CCONJ
easat-2167	205	11	technology	technology	NOUN
easat-2167	205	12	issn	issn	PROPN
easat-2167	205	13	:	:	PUNCT
easat-2167	205	14	2576	2576	NUM
easat-2167	205	15	-	-	SYM
easat-2167	205	16	8484	8484	NUM
easat-2167	205	17	vol	vol	NOUN
easat-2167	205	18	.	.	PROPN
easat-2167	206	1	8	8	NUM
easat-2167	206	2	,	,	PUNCT
easat-2167	206	3	no	no	INTJ
easat-2167	206	4	.	.	NOUN
easat-2167	207	1	6	6	NUM
easat-2167	207	2	:	:	PUNCT
easat-2167	207	3	793	793	NUM
easat-2167	207	4	-	-	NUM
easat-2167	207	5	813	813	NUM
easat-2167	207	6	,	,	PUNCT
easat-2167	207	7	2024	2024	NUM
easat-2167	207	8	doi	doi	NOUN
easat-2167	207	9	:	:	PUNCT
easat-2167	207	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	207	11	©	©	NOUN
easat-2167	207	12	2024	2024	NUM
easat-2167	207	13	by	by	ADP
easat-2167	207	14	the	the	DET
easat-2167	207	15	authors	author	NOUN
easat-2167	207	16	;	;	PUNCT
easat-2167	207	17	licensee	licensee	PROPN
easat-2167	207	18	learning	learning	NOUN
easat-2167	207	19	gate	gate	VERB
easat-2167	207	20	the	the	DET
easat-2167	207	21	decoder	decoder	NOUN
easat-2167	207	22	path	path	NOUN
easat-2167	207	23	is	be	AUX
easat-2167	207	24	designed	design	VERB
easat-2167	207	25	to	to	PART
easat-2167	207	26	up	up	ADP
easat-2167	207	27	sample	sample	VERB
easat-2167	207	28	the	the	DET
easat-2167	207	29	feature	feature	NOUN
easat-2167	207	30	maps	map	NOUN
easat-2167	207	31	back	back	ADV
easat-2167	207	32	to	to	ADP
easat-2167	207	33	the	the	DET
easat-2167	207	34	original	original	ADJ
easat-2167	207	35	image	image	NOUN
easat-2167	207	36	size	size	NOUN
easat-2167	207	37	while	while	SCONJ
easat-2167	207	38	integrating	integrate	VERB
easat-2167	207	39	the	the	DET
easat-2167	207	40	skip	skip	ADJ
easat-2167	207	41	connections	connection	NOUN
easat-2167	207	42	.	.	PUNCT
easat-2167	208	1	attention	attention	NOUN
easat-2167	208	2	gates	gate	NOUN
easat-2167	208	3	are	be	AUX
easat-2167	208	4	introduced	introduce	VERB
easat-2167	208	5	at	at	ADP
easat-2167	208	6	each	each	DET
easat-2167	208	7	skip	skip	ADJ
easat-2167	208	8	connection	connection	NOUN
easat-2167	208	9	to	to	PART
easat-2167	208	10	focus	focus	VERB
easat-2167	208	11	on	on	ADP
easat-2167	208	12	the	the	DET
easat-2167	208	13	most	most	ADV
easat-2167	208	14	relevant	relevant	ADJ
easat-2167	208	15	features	feature	NOUN
easat-2167	208	16	.	.	PUNCT
easat-2167	209	1	7.3.1	7.3.1	X
easat-2167	209	2	.	.	PUNCT
easat-2167	209	3	attention	attention	NOUN
easat-2167	209	4	gates	gate	NOUN
easat-2167	209	5	attention	attention	NOUN
easat-2167	209	6	gates	gate	NOUN
easat-2167	209	7	are	be	AUX
easat-2167	209	8	used	use	VERB
easat-2167	209	9	to	to	PART
easat-2167	209	10	refine	refine	VERB
easat-2167	209	11	the	the	DET
easat-2167	209	12	skip	skip	ADJ
easat-2167	209	13	connections	connection	NOUN
easat-2167	209	14	by	by	ADP
easat-2167	209	15	focusing	focus	VERB
easat-2167	209	16	on	on	ADP
easat-2167	209	17	the	the	DET
easat-2167	209	18	most	most	ADV
easat-2167	209	19	relevant	relevant	ADJ
easat-2167	209	20	spatial	spatial	ADJ
easat-2167	209	21	features	feature	NOUN
easat-2167	209	22	before	before	ADP
easat-2167	209	23	concatenating	concatenate	VERB
easat-2167	209	24	them	they	PRON
easat-2167	209	25	with	with	ADP
easat-2167	209	26	the	the	DET
easat-2167	209	27	decoder	decoder	NOUN
easat-2167	209	28	features	feature	NOUN
easat-2167	209	29	.	.	PUNCT
easat-2167	210	1	this	this	PRON
easat-2167	210	2	helps	help	VERB
easat-2167	210	3	in	in	ADP
easat-2167	210	4	improving	improve	VERB
easat-2167	210	5	the	the	DET
easat-2167	210	6	segmentation	segmentation	NOUN
easat-2167	210	7	accuracy	accuracy	NOUN
easat-2167	210	8	by	by	ADP
easat-2167	210	9	highlighting	highlight	VERB
easat-2167	210	10	important	important	ADJ
easat-2167	210	11	regions	region	NOUN
easat-2167	210	12	in	in	ADP
easat-2167	210	13	the	the	DET
easat-2167	210	14	feature	feature	NOUN
easat-2167	210	15	maps	map	NOUN
easat-2167	210	16	.	.	PUNCT
easat-2167	211	1	•	•	NUM
easat-2167	211	2	theta_x	theta_x	NOUN
easat-2167	211	3	and	and	CCONJ
easat-2167	211	4	phi_g	phi_g	NOUN
easat-2167	211	5	:	:	PUNCT
easat-2167	211	6	these	these	PRON
easat-2167	211	7	are	be	AUX
easat-2167	211	8	feature	feature	NOUN
easat-2167	211	9	transformations	transformation	NOUN
easat-2167	211	10	of	of	ADP
easat-2167	211	11	the	the	DET
easat-2167	211	12	input	input	NOUN
easat-2167	211	13	x	x	X
easat-2167	211	14	(	(	PUNCT
easat-2167	211	15	skip	skip	ADJ
easat-2167	211	16	connection	connection	NOUN
easat-2167	211	17	)	)	PUNCT
easat-2167	211	18	and	and	CCONJ
easat-2167	211	19	gating	gate	VERB
easat-2167	211	20	signal	signal	NOUN
easat-2167	211	21	g	g	PROPN
easat-2167	211	22	(	(	PUNCT
easat-2167	211	23	decoder	decoder	NOUN
easat-2167	211	24	feature	feature	NOUN
easat-2167	211	25	)	)	PUNCT
easat-2167	211	26	respectively	respectively	ADV
easat-2167	211	27	,	,	PUNCT
easat-2167	211	28	which	which	PRON
easat-2167	211	29	are	be	AUX
easat-2167	211	30	then	then	ADV
easat-2167	211	31	added	add	VERB
easat-2167	211	32	together	together	ADV
easat-2167	211	33	.	.	PUNCT
easat-2167	212	1	•	•	NOUN
easat-2167	212	2	upsampled_theta_x	upsampled_theta_x	ADJ
easat-2167	212	3	:	:	PUNCT
easat-2167	212	4	upsamples	upsample	NOUN
easat-2167	212	5	the	the	DET
easat-2167	212	6	transformed	transform	VERB
easat-2167	212	7	skip	skip	ADJ
easat-2167	212	8	connection	connection	NOUN
easat-2167	212	9	to	to	PART
easat-2167	212	10	match	match	VERB
easat-2167	212	11	the	the	DET
easat-2167	212	12	size	size	NOUN
easat-2167	212	13	of	of	ADP
easat-2167	212	14	the	the	DET
easat-2167	212	15	gating	gate	VERB
easat-2167	212	16	signal	signal	NOUN
easat-2167	212	17	.	.	PUNCT
easat-2167	213	1	•	•	NUM
easat-2167	213	2	psi	psi	NOUN
easat-2167	213	3	:	:	PUNCT
easat-2167	213	4	a	a	DET
easat-2167	213	5	sigmoid	sigmoid	NOUN
easat-2167	213	6	activation	activation	NOUN
easat-2167	213	7	is	be	AUX
easat-2167	213	8	applied	apply	VERB
easat-2167	213	9	to	to	PART
easat-2167	213	10	create	create	VERB
easat-2167	213	11	an	an	DET
easat-2167	213	12	attention	attention	NOUN
easat-2167	213	13	map	map	NOUN
easat-2167	213	14	,	,	PUNCT
easat-2167	213	15	which	which	PRON
easat-2167	213	16	is	be	AUX
easat-2167	213	17	then	then	ADV
easat-2167	213	18	used	use	VERB
easat-2167	213	19	to	to	PART
easat-2167	213	20	weight	weight	VERB
easat-2167	213	21	the	the	DET
easat-2167	213	22	original	original	ADJ
easat-2167	213	23	skip	skip	ADJ
easat-2167	213	24	connection	connection	NOUN
easat-2167	213	25	features	feature	NOUN
easat-2167	213	26	.	.	PUNCT
easat-2167	214	1	•	•	NUM
easat-2167	214	2	final	final	ADJ
easat-2167	214	3	multiplication	multiplication	NOUN
easat-2167	214	4	:	:	PUNCT
easat-2167	214	5	the	the	DET
easat-2167	214	6	original	original	ADJ
easat-2167	214	7	skip	skip	ADJ
easat-2167	214	8	connection	connection	NOUN
easat-2167	214	9	features	feature	NOUN
easat-2167	214	10	are	be	AUX
easat-2167	214	11	multiplied	multiply	VERB
easat-2167	214	12	by	by	ADP
easat-2167	214	13	the	the	DET
easat-2167	214	14	attention	attention	NOUN
easat-2167	214	15	map	map	NOUN
easat-2167	214	16	,	,	PUNCT
easat-2167	214	17	effectively	effectively	ADV
easat-2167	214	18	focusing	focus	VERB
easat-2167	214	19	on	on	ADP
easat-2167	214	20	the	the	DET
easat-2167	214	21	most	most	ADV
easat-2167	214	22	relevant	relevant	ADJ
easat-2167	214	23	features	feature	NOUN
easat-2167	214	24	.	.	PUNCT
easat-2167	215	1	7.3.2	7.3.2	X
easat-2167	215	2	.	.	PUNCT
easat-2167	215	3	decoder	decoder	NOUN
easat-2167	215	4	operations	operation	NOUN
easat-2167	215	5	the	the	DET
easat-2167	215	6	decoder	decoder	NOUN
easat-2167	215	7	upsamples	upsample	NOUN
easat-2167	215	8	the	the	DET
easat-2167	215	9	features	feature	NOUN
easat-2167	215	10	step	step	NOUN
easat-2167	215	11	-	-	PUNCT
easat-2167	215	12	by	by	ADP
easat-2167	215	13	-	-	PUNCT
easat-2167	215	14	step	step	NOUN
easat-2167	215	15	,	,	PUNCT
easat-2167	215	16	concatenating	concatenate	VERB
easat-2167	215	17	the	the	DET
easat-2167	215	18	attention	attention	NOUN
easat-2167	215	19	-	-	PUNCT
easat-2167	215	20	refined	refine	VERB
easat-2167	215	21	skip	skip	ADJ
easat-2167	215	22	connections	connection	NOUN
easat-2167	215	23	at	at	ADP
easat-2167	215	24	each	each	DET
easat-2167	215	25	stage	stage	NOUN
easat-2167	215	26	to	to	PART
easat-2167	215	27	recover	recover	VERB
easat-2167	215	28	spatial	spatial	ADJ
easat-2167	215	29	resolution	resolution	NOUN
easat-2167	215	30	.	.	PUNCT
easat-2167	216	1	•	•	NUM
easat-2167	216	2	conv2d	conv2d	ADJ
easat-2167	216	3	layers	layer	NOUN
easat-2167	216	4	:	:	PUNCT
easat-2167	216	5	these	these	DET
easat-2167	216	6	layers	layer	NOUN
easat-2167	216	7	refine	refine	VERB
easat-2167	216	8	the	the	DET
easat-2167	216	9	upsampled	upsample	VERB
easat-2167	216	10	features	feature	NOUN
easat-2167	216	11	by	by	ADP
easat-2167	216	12	applying	apply	VERB
easat-2167	216	13	convolutions	convolution	NOUN
easat-2167	216	14	.	.	PUNCT
easat-2167	217	1	•	•	NUM
easat-2167	217	2	batchnormalization	batchnormalization	NOUN
easat-2167	217	3	:	:	PUNCT
easat-2167	217	4	stabilizes	stabilize	VERB
easat-2167	217	5	and	and	CCONJ
easat-2167	217	6	accelerates	accelerate	VERB
easat-2167	217	7	training	training	NOUN
easat-2167	217	8	by	by	ADP
easat-2167	217	9	normalizing	normalize	VERB
easat-2167	217	10	the	the	DET
easat-2167	217	11	output	output	NOUN
easat-2167	217	12	of	of	ADP
easat-2167	217	13	each	each	DET
easat-2167	217	14	convolutional	convolutional	ADJ
easat-2167	217	15	layer	layer	NOUN
easat-2167	217	16	.	.	PUNCT
easat-2167	218	1	•	•	NUM
easat-2167	218	2	upsampling2d	upsampling2d	PROPN
easat-2167	218	3	:	:	PUNCT
easat-2167	218	4	doubles	double	VERB
easat-2167	218	5	the	the	DET
easat-2167	218	6	spatial	spatial	ADJ
easat-2167	218	7	dimensions	dimension	NOUN
easat-2167	218	8	of	of	ADP
easat-2167	218	9	the	the	DET
easat-2167	218	10	feature	feature	NOUN
easat-2167	218	11	maps	map	NOUN
easat-2167	218	12	.	.	PUNCT
easat-2167	219	1	this	this	DET
easat-2167	219	2	process	process	NOUN
easat-2167	219	3	is	be	AUX
easat-2167	219	4	repeated	repeat	VERB
easat-2167	219	5	for	for	ADP
easat-2167	219	6	each	each	DET
easat-2167	219	7	skip	skip	ADJ
easat-2167	219	8	connection	connection	NOUN
easat-2167	219	9	until	until	SCONJ
easat-2167	219	10	the	the	DET
easat-2167	219	11	feature	feature	NOUN
easat-2167	219	12	map	map	NOUN
easat-2167	219	13	is	be	AUX
easat-2167	219	14	upsampled	upsample	VERB
easat-2167	219	15	back	back	ADV
easat-2167	219	16	to	to	ADP
easat-2167	219	17	the	the	DET
easat-2167	219	18	original	original	ADJ
easat-2167	219	19	input	input	NOUN
easat-2167	219	20	size	size	NOUN
easat-2167	219	21	.	.	PUNCT
easat-2167	220	1	7.4	7.4	NUM
easat-2167	220	2	.	.	PUNCT
easat-2167	221	1	final	final	ADJ
easat-2167	221	2	output	output	NOUN
easat-2167	221	3	layer	layer	NOUN
easat-2167	221	4	the	the	DET
easat-2167	221	5	final	final	ADJ
easat-2167	221	6	layer	layer	NOUN
easat-2167	221	7	uses	use	VERB
easat-2167	221	8	a	a	DET
easat-2167	221	9	1x1	1x1	NUM
easat-2167	221	10	convolution	convolution	NOUN
easat-2167	221	11	to	to	PART
easat-2167	221	12	reduce	reduce	VERB
easat-2167	221	13	the	the	DET
easat-2167	221	14	feature	feature	NOUN
easat-2167	221	15	map	map	NOUN
easat-2167	221	16	to	to	ADP
easat-2167	221	17	the	the	DET
easat-2167	221	18	desired	desire	VERB
easat-2167	221	19	number	number	NOUN
easat-2167	221	20	of	of	ADP
easat-2167	221	21	output	output	NOUN
easat-2167	221	22	classes	class	NOUN
easat-2167	221	23	and	and	CCONJ
easat-2167	221	24	applies	apply	VERB
easat-2167	221	25	a	a	DET
easat-2167	221	26	sigmoid	sigmoid	NOUN
easat-2167	221	27	activation	activation	NOUN
easat-2167	221	28	for	for	ADP
easat-2167	221	29	binary	binary	ADJ
easat-2167	221	30	segmentation	segmentation	NOUN
easat-2167	221	31	.	.	PUNCT
easat-2167	222	1	•	•	NUM
easat-2167	222	2	conv2d(num_classes	conv2d(num_classe	NOUN
easat-2167	222	3	,	,	PUNCT
easat-2167	222	4	(	(	PUNCT
easat-2167	222	5	1	1	NUM
easat-2167	222	6	,	,	PUNCT
easat-2167	222	7	1	1	NUM
easat-2167	222	8	)	)	PUNCT
easat-2167	222	9	):	):	PUNCT
easat-2167	222	10	reduces	reduce	VERB
easat-2167	222	11	the	the	DET
easat-2167	222	12	channel	channel	NOUN
easat-2167	222	13	dimension	dimension	NOUN
easat-2167	222	14	to	to	ADP
easat-2167	222	15	num_classes	num_classe	NOUN
easat-2167	222	16	(	(	PUNCT
easat-2167	222	17	typically	typically	ADV
easat-2167	222	18	1	1	NUM
easat-2167	222	19	for	for	ADP
easat-2167	222	20	binary	binary	ADJ
easat-2167	222	21	segmentation	segmentation	NOUN
easat-2167	222	22	)	)	PUNCT
easat-2167	222	23	.	.	PUNCT
easat-2167	223	1	•	•	NUM
easat-2167	223	2	activation('sigmoid	activation('sigmoid	PROPN
easat-2167	223	3	'	'	PUNCT
easat-2167	223	4	):	):	PUNCT
easat-2167	223	5	outputs	output	VERB
easat-2167	223	6	a	a	DET
easat-2167	223	7	probability	probability	NOUN
easat-2167	223	8	map	map	NOUN
easat-2167	223	9	indicating	indicate	VERB
easat-2167	223	10	the	the	DET
easat-2167	223	11	likelihood	likelihood	NOUN
easat-2167	223	12	of	of	ADP
easat-2167	223	13	each	each	DET
easat-2167	223	14	pixel	pixel	NOUN
easat-2167	223	15	belonging	belong	VERB
easat-2167	223	16	to	to	ADP
easat-2167	223	17	the	the	DET
easat-2167	223	18	target	target	NOUN
easat-2167	223	19	class	class	NOUN
easat-2167	223	20	.	.	PUNCT
easat-2167	224	1	7.5	7.5	NUM
easat-2167	224	2	.	.	PUNCT
easat-2167	224	3	model	model	NOUN
easat-2167	224	4	compilation	compilation	NOUN
easat-2167	224	5	the	the	DET
easat-2167	224	6	model	model	NOUN
easat-2167	224	7	is	be	AUX
easat-2167	224	8	then	then	ADV
easat-2167	224	9	compiled	compile	VERB
easat-2167	224	10	into	into	ADP
easat-2167	224	11	a	a	DET
easat-2167	224	12	keras	keras	PROPN
easat-2167	224	13	model	model	NOUN
easat-2167	224	14	with	with	ADP
easat-2167	224	15	inputs	input	NOUN
easat-2167	224	16	from	from	ADP
easat-2167	224	17	the	the	DET
easat-2167	224	18	efficientnetb7	efficientnetb7	NOUN
easat-2167	224	19	encoder	encoder	NOUN
easat-2167	224	20	and	and	CCONJ
easat-2167	224	21	outputs	output	NOUN
easat-2167	224	22	from	from	ADP
easat-2167	224	23	the	the	DET
easat-2167	224	24	final	final	ADJ
easat-2167	224	25	upsamling	upsamle	VERB
easat-2167	224	26	layer	layer	NOUN
easat-2167	224	27	.	.	PUNCT
easat-2167	225	1	804	804	NUM
easat-2167	225	2	edelweiss	edelweiss	PROPN
easat-2167	225	3	applied	apply	VERB
easat-2167	225	4	science	science	NOUN
easat-2167	225	5	and	and	CCONJ
easat-2167	225	6	technology	technology	NOUN
easat-2167	225	7	issn	issn	PROPN
easat-2167	225	8	:	:	PUNCT
easat-2167	225	9	2576	2576	NUM
easat-2167	225	10	-	-	SYM
easat-2167	225	11	8484	8484	NUM
easat-2167	225	12	vol	vol	NOUN
easat-2167	225	13	.	.	PROPN
easat-2167	225	14	8	8	NUM
easat-2167	225	15	,	,	PUNCT
easat-2167	225	16	no	no	INTJ
easat-2167	225	17	.	.	NOUN
easat-2167	226	1	6	6	NUM
easat-2167	226	2	:	:	PUNCT
easat-2167	226	3	793	793	NUM
easat-2167	226	4	-	-	NUM
easat-2167	226	5	813	813	NUM
easat-2167	226	6	,	,	PUNCT
easat-2167	226	7	2024	2024	NUM
easat-2167	226	8	doi	doi	NOUN
easat-2167	226	9	:	:	PUNCT
easat-2167	226	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	226	11	©	©	NOUN
easat-2167	226	12	2024	2024	NUM
easat-2167	226	13	by	by	ADP
easat-2167	226	14	the	the	DET
easat-2167	226	15	authors	author	NOUN
easat-2167	226	16	;	;	PUNCT
easat-2167	226	17	licensee	licensee	PROPN
easat-2167	226	18	learning	learn	VERB
easat-2167	226	19	gate	gate	PROPN
easat-2167	226	20	figure	figure	NOUN
easat-2167	226	21	3	3	NUM
easat-2167	226	22	.	.	PUNCT
easat-2167	226	23	u	u	NOUN
easat-2167	226	24	-	-	NOUN
easat-2167	226	25	net	net	ADJ
easat-2167	226	26	with	with	ADP
easat-2167	226	27	efficientnetb7	efficientnetb7	NOUN
easat-2167	226	28	and	and	CCONJ
easat-2167	226	29	attention	attention	NOUN
easat-2167	226	30	gates	gate	NOUN
easat-2167	226	31	.	.	PUNCT
easat-2167	227	1	805	805	NUM
easat-2167	227	2	edelweiss	edelweiss	PROPN
easat-2167	227	3	applied	apply	VERB
easat-2167	227	4	science	science	NOUN
easat-2167	227	5	and	and	CCONJ
easat-2167	227	6	technology	technology	NOUN
easat-2167	227	7	issn	issn	PROPN
easat-2167	227	8	:	:	PUNCT
easat-2167	227	9	2576	2576	NUM
easat-2167	227	10	-	-	SYM
easat-2167	227	11	8484	8484	NUM
easat-2167	227	12	vol	vol	NOUN
easat-2167	227	13	.	.	PROPN
easat-2167	227	14	8	8	NUM
easat-2167	227	15	,	,	PUNCT
easat-2167	227	16	no	no	INTJ
easat-2167	227	17	.	.	NOUN
easat-2167	228	1	6	6	NUM
easat-2167	228	2	:	:	PUNCT
easat-2167	228	3	793	793	NUM
easat-2167	228	4	-	-	NUM
easat-2167	228	5	813	813	NUM
easat-2167	228	6	,	,	PUNCT
easat-2167	228	7	2024	2024	NUM
easat-2167	228	8	doi	doi	NOUN
easat-2167	228	9	:	:	PUNCT
easat-2167	228	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	228	11	©	©	NOUN
easat-2167	228	12	2024	2024	NUM
easat-2167	228	13	by	by	ADP
easat-2167	228	14	the	the	DET
easat-2167	228	15	authors	author	NOUN
easat-2167	228	16	;	;	PUNCT
easat-2167	228	17	licensee	licensee	PROPN
easat-2167	228	18	learning	learning	NOUN
easat-2167	228	19	gate	gate	NOUN
easat-2167	228	20	this	this	DET
easat-2167	228	21	model	model	NOUN
easat-2167	228	22	represents	represent	VERB
easat-2167	228	23	a	a	DET
easat-2167	228	24	significant	significant	ADJ
easat-2167	228	25	advancement	advancement	NOUN
easat-2167	228	26	in	in	ADP
easat-2167	228	27	the	the	DET
easat-2167	228	28	field	field	NOUN
easat-2167	228	29	of	of	ADP
easat-2167	228	30	image	image	NOUN
easat-2167	228	31	segmentation	segmentation	NOUN
easat-2167	228	32	,	,	PUNCT
easat-2167	228	33	particularly	particularly	ADV
easat-2167	228	34	for	for	ADP
easat-2167	228	35	applications	application	NOUN
easat-2167	228	36	like	like	ADP
easat-2167	228	37	medical	medical	ADJ
easat-2167	228	38	imaging	imaging	NOUN
easat-2167	228	39	,	,	PUNCT
easat-2167	228	40	remote	remote	ADJ
easat-2167	228	41	sensing	sensing	NOUN
easat-2167	228	42	,	,	PUNCT
easat-2167	228	43	and	and	CCONJ
easat-2167	228	44	precision	precision	NOUN
easat-2167	228	45	agriculture	agriculture	NOUN
easat-2167	228	46	.	.	PUNCT
easat-2167	229	1	here	here	ADV
easat-2167	229	2	’s	’	VERB
easat-2167	229	3	why	why	SCONJ
easat-2167	229	4	:	:	PUNCT
easat-2167	229	5	1	1	X
easat-2167	229	6	.	.	X
easat-2167	229	7	use	use	NOUN
easat-2167	229	8	of	of	ADP
easat-2167	229	9	efficientnetb7	efficientnetb7	NOUN
easat-2167	229	10	:	:	PUNCT
easat-2167	229	11	leveraging	leverage	VERB
easat-2167	229	12	efficientnetb7	efficientnetb7	NOUN
easat-2167	229	13	as	as	SCONJ
easat-2167	229	14	the	the	DET
easat-2167	229	15	backbone	backbone	NOUN
easat-2167	229	16	provides	provide	VERB
easat-2167	229	17	a	a	DET
easat-2167	229	18	powerful	powerful	ADJ
easat-2167	229	19	and	and	CCONJ
easat-2167	229	20	efficient	efficient	ADJ
easat-2167	229	21	feature	feature	NOUN
easat-2167	229	22	extractor	extractor	NOUN
easat-2167	229	23	,	,	PUNCT
easat-2167	229	24	offering	offer	VERB
easat-2167	229	25	state	state	NOUN
easat-2167	229	26	-	-	PUNCT
easat-2167	229	27	of	of	ADP
easat-2167	229	28	-	-	PUNCT
easat-2167	229	29	the	the	DET
easat-2167	229	30	-	-	PUNCT
easat-2167	229	31	art	art	NOUN
easat-2167	229	32	performance	performance	NOUN
easat-2167	229	33	with	with	ADP
easat-2167	229	34	fewer	few	ADJ
easat-2167	229	35	parameters	parameter	NOUN
easat-2167	229	36	.	.	PUNCT
easat-2167	230	1	2	2	X
easat-2167	230	2	.	.	X
easat-2167	230	3	attention	attention	NOUN
easat-2167	230	4	gates	gate	NOUN
easat-2167	230	5	:	:	PUNCT
easat-2167	230	6	the	the	DET
easat-2167	230	7	integration	integration	NOUN
easat-2167	230	8	of	of	ADP
easat-2167	230	9	attention	attention	NOUN
easat-2167	230	10	gates	gate	NOUN
easat-2167	230	11	at	at	ADP
easat-2167	230	12	each	each	DET
easat-2167	230	13	skip	skip	ADJ
easat-2167	230	14	connection	connection	NOUN
easat-2167	230	15	is	be	AUX
easat-2167	230	16	a	a	DET
easat-2167	230	17	novel	novel	ADJ
easat-2167	230	18	approach	approach	NOUN
easat-2167	230	19	to	to	PART
easat-2167	230	20	focus	focus	VERB
easat-2167	230	21	on	on	ADP
easat-2167	230	22	the	the	DET
easat-2167	230	23	most	most	ADV
easat-2167	230	24	relevant	relevant	ADJ
easat-2167	230	25	features	feature	NOUN
easat-2167	230	26	,	,	PUNCT
easat-2167	230	27	improving	improve	VERB
easat-2167	230	28	the	the	DET
easat-2167	230	29	model	model	NOUN
easat-2167	230	30	's	's	PART
easat-2167	230	31	ability	ability	NOUN
easat-2167	230	32	to	to	PART
easat-2167	230	33	differentiate	differentiate	VERB
easat-2167	230	34	between	between	ADP
easat-2167	230	35	important	important	ADJ
easat-2167	230	36	and	and	CCONJ
easat-2167	230	37	less	less	ADV
easat-2167	230	38	important	important	ADJ
easat-2167	230	39	information	information	NOUN
easat-2167	230	40	in	in	ADP
easat-2167	230	41	the	the	DET
easat-2167	230	42	feature	feature	NOUN
easat-2167	230	43	maps	map	NOUN
easat-2167	230	44	.	.	PUNCT
easat-2167	231	1	3	3	X
easat-2167	231	2	.	.	X
easat-2167	231	3	advanced	advanced	ADJ
easat-2167	231	4	decoder	decoder	NOUN
easat-2167	231	5	:	:	PUNCT
easat-2167	231	6	the	the	DET
easat-2167	231	7	combination	combination	NOUN
easat-2167	231	8	of	of	ADP
easat-2167	231	9	upsampling	upsample	VERB
easat-2167	231	10	layers	layer	NOUN
easat-2167	231	11	,	,	PUNCT
easat-2167	231	12	convolutional	convolutional	ADJ
easat-2167	231	13	layers	layer	NOUN
easat-2167	231	14	,	,	PUNCT
easat-2167	231	15	and	and	CCONJ
easat-2167	231	16	attention	attention	NOUN
easat-2167	231	17	gates	gate	NOUN
easat-2167	231	18	in	in	ADP
easat-2167	231	19	the	the	DET
easat-2167	231	20	decoder	decoder	NOUN
easat-2167	231	21	path	path	NOUN
easat-2167	231	22	ensures	ensure	VERB
easat-2167	231	23	that	that	SCONJ
easat-2167	231	24	the	the	DET
easat-2167	231	25	model	model	NOUN
easat-2167	231	26	can	can	AUX
easat-2167	231	27	accurately	accurately	ADV
easat-2167	231	28	reconstruct	reconstruct	VERB
easat-2167	231	29	high	high	ADJ
easat-2167	231	30	-	-	PUNCT
easat-2167	231	31	resolution	resolution	NOUN
easat-2167	231	32	segmentation	segmentation	NOUN
easat-2167	231	33	maps	map	NOUN
easat-2167	231	34	.	.	PUNCT
easat-2167	232	1	4	4	X
easat-2167	232	2	.	.	X
easat-2167	232	3	generalization	generalization	NOUN
easat-2167	232	4	:	:	PUNCT
easat-2167	232	5	the	the	DET
easat-2167	232	6	use	use	NOUN
easat-2167	232	7	of	of	ADP
easat-2167	232	8	efficientnetb7	efficientnetb7	NOUN
easat-2167	232	9	with	with	ADP
easat-2167	232	10	attention	attention	NOUN
easat-2167	232	11	mechanisms	mechanism	NOUN
easat-2167	232	12	can	can	AUX
easat-2167	232	13	generalize	generalize	VERB
easat-2167	232	14	well	well	ADV
easat-2167	232	15	to	to	ADP
easat-2167	232	16	various	various	ADJ
easat-2167	232	17	segmentation	segmentation	NOUN
easat-2167	232	18	tasks	task	NOUN
easat-2167	232	19	,	,	PUNCT
easat-2167	232	20	making	make	VERB
easat-2167	232	21	it	it	PRON
easat-2167	232	22	a	a	DET
easat-2167	232	23	versatile	versatile	ADJ
easat-2167	232	24	tool	tool	NOUN
easat-2167	232	25	for	for	ADP
easat-2167	232	26	different	different	ADJ
easat-2167	232	27	applications	application	NOUN
easat-2167	232	28	.	.	PUNCT
easat-2167	233	1	5	5	X
easat-2167	233	2	.	.	X
easat-2167	233	3	state	state	NOUN
easat-2167	233	4	-	-	PUNCT
easat-2167	233	5	of	of	ADP
easat-2167	233	6	-	-	PUNCT
easat-2167	233	7	the	the	DET
easat-2167	233	8	-	-	PUNCT
easat-2167	233	9	art	art	NOUN
easat-2167	233	10	performance	performance	NOUN
easat-2167	233	11	:	:	PUNCT
easat-2167	233	12	given	give	VERB
easat-2167	233	13	the	the	DET
easat-2167	233	14	architecture	architecture	NOUN
easat-2167	233	15	’s	’s	PART
easat-2167	233	16	components	component	NOUN
easat-2167	233	17	and	and	CCONJ
easat-2167	233	18	their	their	PRON
easat-2167	233	19	combination	combination	NOUN
easat-2167	233	20	,	,	PUNCT
easat-2167	233	21	this	this	DET
easat-2167	233	22	model	model	NOUN
easat-2167	233	23	is	be	AUX
easat-2167	233	24	likely	likely	ADJ
easat-2167	233	25	to	to	PART
easat-2167	233	26	achieve	achieve	VERB
easat-2167	233	27	high	high	ADJ
easat-2167	233	28	accuracy	accuracy	NOUN
easat-2167	233	29	in	in	ADP
easat-2167	233	30	segmentation	segmentation	NOUN
easat-2167	233	31	tasks	task	NOUN
easat-2167	233	32	while	while	SCONJ
easat-2167	233	33	being	be	AUX
easat-2167	233	34	computationally	computationally	ADV
easat-2167	233	35	efficient	efficient	ADJ
easat-2167	233	36	,	,	PUNCT
easat-2167	233	37	which	which	PRON
easat-2167	233	38	is	be	AUX
easat-2167	233	39	crucial	crucial	ADJ
easat-2167	233	40	for	for	ADP
easat-2167	233	41	real	real	ADJ
easat-2167	233	42	-	-	PUNCT
easat-2167	233	43	time	time	NOUN
easat-2167	233	44	applications	application	NOUN
easat-2167	233	45	.	.	PUNCT
easat-2167	234	1	8	8	X
easat-2167	234	2	.	.	X
easat-2167	234	3	results	result	NOUN
easat-2167	234	4	&	&	CCONJ
easat-2167	234	5	discussions	discussion	NOUN
easat-2167	234	6	8.1	8.1	NUM
easat-2167	234	7	.	.	PUNCT
easat-2167	235	1	model	model	PROPN
easat-2167	235	2	-	-	PUNCT
easat-2167	235	3	i	i	PRON
easat-2167	235	4	u	u	ADJ
easat-2167	235	5	-	-	ADJ
easat-2167	235	6	net	net	ADJ
easat-2167	235	7	model	model	NOUN
easat-2167	235	8	with	with	ADP
easat-2167	235	9	mobilenetv2	mobilenetv2	PROPN
easat-2167	235	10	encoder	encoder	NOUN
easat-2167	235	11	based	base	VERB
easat-2167	235	12	on	on	ADP
easat-2167	235	13	the	the	DET
easat-2167	235	14	training	training	NOUN
easat-2167	235	15	results	result	NOUN
easat-2167	235	16	of	of	ADP
easat-2167	235	17	the	the	DET
easat-2167	235	18	mobilenetv2	mobilenetv2	PROPN
easat-2167	235	19	model	model	NOUN
easat-2167	235	20	,	,	PUNCT
easat-2167	235	21	here	here	ADV
easat-2167	235	22	's	be	AUX
easat-2167	235	23	a	a	DET
easat-2167	235	24	summary	summary	NOUN
easat-2167	235	25	of	of	ADP
easat-2167	235	26	the	the	DET
easat-2167	235	27	model	model	NOUN
easat-2167	235	28	's	's	PART
easat-2167	235	29	performance	performance	NOUN
easat-2167	235	30	over	over	ADP
easat-2167	235	31	100	100	NUM
easat-2167	235	32	epochs	epoch	NOUN
easat-2167	235	33	on	on	ADP
easat-2167	235	34	blast	blast	NOUN
easat-2167	235	35	rice	rice	NOUN
easat-2167	235	36	diseases	disease	NOUN
easat-2167	235	37	images	image	NOUN
easat-2167	235	38	that	that	PRON
easat-2167	235	39	include	include	VERB
easat-2167	235	40	150	150	NUM
easat-2167	235	41	images	image	NOUN
easat-2167	235	42	of	of	ADP
easat-2167	235	43	true	true	ADJ
easat-2167	235	44	mask	mask	NOUN
easat-2167	235	45	and	and	CCONJ
easat-2167	235	46	actual	actual	ADJ
easat-2167	235	47	image	image	NOUN
easat-2167	235	48	:	:	PUNCT
easat-2167	235	49	8.1.1	8.1.1	X
easat-2167	235	50	.	.	PUNCT
easat-2167	235	51	training	training	NOUN
easat-2167	235	52	and	and	CCONJ
easat-2167	235	53	validation	validation	NOUN
easat-2167	235	54	performance	performance	NOUN
easat-2167	235	55	•	•	ADP
easat-2167	235	56	initial	initial	ADJ
easat-2167	235	57	performance	performance	NOUN
easat-2167	235	58	:	:	PUNCT
easat-2167	235	59	•	•	ADP
easat-2167	235	60	the	the	DET
easat-2167	235	61	model	model	NOUN
easat-2167	235	62	began	begin	VERB
easat-2167	235	63	with	with	ADP
easat-2167	235	64	a	a	DET
easat-2167	235	65	relatively	relatively	ADV
easat-2167	235	66	strong	strong	ADJ
easat-2167	235	67	starting	starting	NOUN
easat-2167	235	68	point	point	NOUN
easat-2167	235	69	in	in	ADP
easat-2167	235	70	terms	term	NOUN
easat-2167	235	71	of	of	ADP
easat-2167	235	72	accuracy	accuracy	NOUN
easat-2167	235	73	and	and	CCONJ
easat-2167	235	74	loss	loss	NOUN
easat-2167	235	75	.	.	PUNCT
easat-2167	236	1	the	the	DET
easat-2167	236	2	first	first	ADJ
easat-2167	236	3	epoch	epoch	NOUN
easat-2167	236	4	achieved	achieve	VERB
easat-2167	236	5	an	an	DET
easat-2167	236	6	accuracy	accuracy	NOUN
easat-2167	236	7	of	of	ADP
easat-2167	236	8	79.8	79.8	NUM
easat-2167	236	9	%	%	NOUN
easat-2167	236	10	and	and	CCONJ
easat-2167	236	11	a	a	DET
easat-2167	236	12	loss	loss	NOUN
easat-2167	236	13	of	of	ADP
easat-2167	236	14	0.1487	0.1487	NUM
easat-2167	236	15	,	,	PUNCT
easat-2167	236	16	with	with	ADP
easat-2167	236	17	validation	validation	NOUN
easat-2167	236	18	accuracy	accuracy	NOUN
easat-2167	236	19	at	at	ADP
easat-2167	236	20	77.7	77.7	NUM
easat-2167	236	21	%	%	NOUN
easat-2167	236	22	and	and	CCONJ
easat-2167	236	23	validation	validation	NOUN
easat-2167	236	24	loss	loss	NOUN
easat-2167	236	25	at	at	ADP
easat-2167	236	26	0.1955	0.1955	NUM
easat-2167	236	27	.	.	PUNCT
easat-2167	237	1	•	•	NUM
easat-2167	237	2	fluctuations	fluctuation	NOUN
easat-2167	237	3	:	:	PUNCT
easat-2167	237	4	•	•	NOUN
easat-2167	237	5	throughout	throughout	ADP
easat-2167	237	6	the	the	DET
easat-2167	237	7	training	training	NOUN
easat-2167	237	8	process	process	NOUN
easat-2167	237	9	,	,	PUNCT
easat-2167	237	10	both	both	DET
easat-2167	237	11	training	training	NOUN
easat-2167	237	12	and	and	CCONJ
easat-2167	237	13	validation	validation	NOUN
easat-2167	237	14	accuracy	accuracy	NOUN
easat-2167	237	15	exhibited	exhibit	VERB
easat-2167	237	16	fluctuations	fluctuation	NOUN
easat-2167	237	17	.	.	PUNCT
easat-2167	238	1	for	for	ADP
easat-2167	238	2	instance	instance	NOUN
easat-2167	238	3	,	,	PUNCT
easat-2167	238	4	training	training	NOUN
easat-2167	238	5	accuracy	accuracy	NOUN
easat-2167	238	6	ranged	range	VERB
easat-2167	238	7	between	between	ADP
easat-2167	238	8	70	70	NUM
easat-2167	238	9	%	%	NOUN
easat-2167	238	10	and	and	CCONJ
easat-2167	238	11	80	80	NUM
easat-2167	238	12	%	%	NOUN
easat-2167	238	13	,	,	PUNCT
easat-2167	238	14	while	while	SCONJ
easat-2167	238	15	validation	validation	NOUN
easat-2167	238	16	accuracy	accuracy	NOUN
easat-2167	238	17	varied	varied	ADJ
easat-2167	238	18	similarly	similarly	ADV
easat-2167	238	19	,	,	PUNCT
easat-2167	238	20	occasionally	occasionally	ADV
easat-2167	238	21	dipping	dip	VERB
easat-2167	238	22	into	into	ADP
easat-2167	238	23	the	the	DET
easat-2167	238	24	70	70	NUM
easat-2167	238	25	%	%	NOUN
easat-2167	238	26	range	range	NOUN
easat-2167	238	27	before	before	ADP
easat-2167	238	28	recovering	recover	VERB
easat-2167	238	29	.	.	PUNCT
easat-2167	239	1	•	•	NUM
easat-2167	239	2	loss	loss	NOUN
easat-2167	239	3	values	value	NOUN
easat-2167	239	4	also	also	ADV
easat-2167	239	5	fluctuated	fluctuate	VERB
easat-2167	239	6	,	,	PUNCT
easat-2167	239	7	with	with	ADP
easat-2167	239	8	periods	period	NOUN
easat-2167	239	9	where	where	SCONJ
easat-2167	239	10	the	the	DET
easat-2167	239	11	training	training	NOUN
easat-2167	239	12	loss	loss	NOUN
easat-2167	239	13	decreased	decrease	VERB
easat-2167	239	14	,	,	PUNCT
easat-2167	239	15	followed	follow	VERB
easat-2167	239	16	by	by	ADP
easat-2167	239	17	increases	increase	NOUN
easat-2167	239	18	,	,	PUNCT
easat-2167	239	19	indicating	indicate	VERB
easat-2167	239	20	potential	potential	ADJ
easat-2167	239	21	challenges	challenge	NOUN
easat-2167	239	22	in	in	ADP
easat-2167	239	23	learning	learn	VERB
easat-2167	239	24	stability	stability	NOUN
easat-2167	239	25	.	.	PUNCT
easat-2167	240	1	•	•	NUM
easat-2167	240	2	midway	midway	NOUN
easat-2167	240	3	improvements	improvement	NOUN
easat-2167	240	4	:	:	PUNCT
easat-2167	240	5	•	•	NOUN
easat-2167	240	6	around	around	ADP
easat-2167	240	7	epoch	epoch	NOUN
easat-2167	240	8	14	14	NUM
easat-2167	240	9	,	,	PUNCT
easat-2167	240	10	the	the	DET
easat-2167	240	11	model	model	NOUN
easat-2167	240	12	reached	reach	VERB
easat-2167	240	13	a	a	DET
easat-2167	240	14	notable	notable	ADJ
easat-2167	240	15	improvement	improvement	NOUN
easat-2167	240	16	,	,	PUNCT
easat-2167	240	17	with	with	ADP
easat-2167	240	18	validation	validation	NOUN
easat-2167	240	19	accuracy	accuracy	NOUN
easat-2167	240	20	peaking	peaking	NOUN
easat-2167	240	21	at	at	ADP
easat-2167	240	22	81.5	81.5	NUM
easat-2167	240	23	%	%	NOUN
easat-2167	240	24	and	and	CCONJ
easat-2167	240	25	validation	validation	NOUN
easat-2167	240	26	loss	loss	NOUN
easat-2167	240	27	dropping	drop	VERB
easat-2167	240	28	to	to	ADP
easat-2167	240	29	0.1722	0.1722	NUM
easat-2167	240	30	,	,	PUNCT
easat-2167	240	31	indicating	indicate	VERB
easat-2167	240	32	better	well	ADJ
easat-2167	240	33	generalization	generalization	NOUN
easat-2167	240	34	at	at	ADP
easat-2167	240	35	that	that	DET
easat-2167	240	36	point	point	NOUN
easat-2167	240	37	.	.	PUNCT
easat-2167	241	1	•	•	NOUN
easat-2167	241	2	overfitting	overfitte	VERB
easat-2167	241	3	signs	sign	NOUN
easat-2167	241	4	:	:	PUNCT
easat-2167	241	5	•	•	ADV
easat-2167	241	6	from	from	ADP
easat-2167	241	7	the	the	DET
easat-2167	241	8	later	later	ADJ
easat-2167	241	9	epochs	epoch	NOUN
easat-2167	241	10	(	(	PUNCT
easat-2167	241	11	around	around	ADP
easat-2167	241	12	epoch	epoch	NOUN
easat-2167	241	13	40	40	NUM
easat-2167	241	14	onwards	onward	NOUN
easat-2167	241	15	)	)	PUNCT
easat-2167	241	16	,	,	PUNCT
easat-2167	241	17	there	there	PRON
easat-2167	241	18	are	be	VERB
easat-2167	241	19	signs	sign	NOUN
easat-2167	241	20	of	of	ADP
easat-2167	241	21	overfitting	overfitte	VERB
easat-2167	241	22	.	.	PUNCT
easat-2167	242	1	the	the	DET
easat-2167	242	2	validation	validation	NOUN
easat-2167	242	3	loss	loss	NOUN
easat-2167	242	4	occasionally	occasionally	ADV
easat-2167	242	5	increased	increase	VERB
easat-2167	242	6	while	while	SCONJ
easat-2167	242	7	training	training	NOUN
easat-2167	242	8	accuracy	accuracy	NOUN
easat-2167	242	9	remained	remain	VERB
easat-2167	242	10	relatively	relatively	ADV
easat-2167	242	11	high	high	ADJ
easat-2167	242	12	,	,	PUNCT
easat-2167	242	13	suggesting	suggest	VERB
easat-2167	242	14	that	that	SCONJ
easat-2167	242	15	the	the	DET
easat-2167	242	16	model	model	NOUN
easat-2167	242	17	was	be	AUX
easat-2167	242	18	becoming	become	VERB
easat-2167	242	19	too	too	ADV
easat-2167	242	20	specialized	specialized	ADJ
easat-2167	242	21	to	to	ADP
easat-2167	242	22	the	the	DET
easat-2167	242	23	training	training	NOUN
easat-2167	242	24	data	datum	NOUN
easat-2167	242	25	.	.	PUNCT
easat-2167	243	1	•	•	NUM
easat-2167	243	2	particularly	particularly	ADV
easat-2167	243	3	,	,	PUNCT
easat-2167	243	4	validation	validation	NOUN
easat-2167	243	5	loss	loss	NOUN
easat-2167	243	6	values	value	NOUN
easat-2167	243	7	increased	increase	VERB
easat-2167	243	8	significantly	significantly	ADV
easat-2167	243	9	in	in	ADP
easat-2167	243	10	some	some	DET
easat-2167	243	11	epochs	epoch	NOUN
easat-2167	243	12	,	,	PUNCT
easat-2167	243	13	for	for	ADP
easat-2167	243	14	example	example	NOUN
easat-2167	243	15	,	,	PUNCT
easat-2167	243	16	during	during	ADP
easat-2167	243	17	epoch	epoch	NOUN
easat-2167	243	18	38	38	NUM
easat-2167	243	19	(	(	PUNCT
easat-2167	243	20	loss	loss	NOUN
easat-2167	243	21	=	=	SYM
easat-2167	243	22	0.4128	0.4128	NUM
easat-2167	243	23	)	)	PUNCT
easat-2167	243	24	,	,	PUNCT
easat-2167	243	25	showing	show	VERB
easat-2167	243	26	poor	poor	ADJ
easat-2167	243	27	generalization	generalization	NOUN
easat-2167	243	28	in	in	ADP
easat-2167	243	29	those	those	DET
easat-2167	243	30	phases	phase	NOUN
easat-2167	243	31	.	.	PUNCT
easat-2167	244	1	•	•	NUM
easat-2167	244	2	final	final	ADJ
easat-2167	244	3	performance	performance	NOUN
easat-2167	244	4	:	:	PUNCT
easat-2167	244	5	•	•	NOUN
easat-2167	244	6	by	by	ADP
easat-2167	244	7	the	the	DET
easat-2167	244	8	end	end	NOUN
easat-2167	244	9	of	of	ADP
easat-2167	244	10	the	the	DET
easat-2167	244	11	training	training	NOUN
easat-2167	244	12	,	,	PUNCT
easat-2167	244	13	the	the	DET
easat-2167	244	14	model	model	NOUN
easat-2167	244	15	managed	manage	VERB
easat-2167	244	16	to	to	PART
easat-2167	244	17	achieve	achieve	VERB
easat-2167	244	18	a	a	DET
easat-2167	244	19	decent	decent	ADJ
easat-2167	244	20	performance	performance	NOUN
easat-2167	244	21	with	with	ADP
easat-2167	244	22	validation	validation	NOUN
easat-2167	244	23	accuracy	accuracy	NOUN
easat-2167	244	24	reaching	reach	VERB
easat-2167	244	25	as	as	ADV
easat-2167	244	26	high	high	ADJ
easat-2167	244	27	as	as	ADP
easat-2167	244	28	88.25	88.25	NUM
easat-2167	244	29	%	%	NOUN
easat-2167	244	30	in	in	ADP
easat-2167	244	31	epoch	epoch	PROPN
easat-2167	244	32	43	43	NUM
easat-2167	244	33	.	.	PUNCT
easat-2167	245	1	however	however	ADV
easat-2167	245	2	,	,	PUNCT
easat-2167	245	3	there	there	PRON
easat-2167	245	4	were	be	VERB
easat-2167	245	5	considerable	considerable	ADJ
easat-2167	245	6	variations	variation	NOUN
easat-2167	245	7	in	in	ADP
easat-2167	245	8	validation	validation	NOUN
easat-2167	245	9	loss	loss	NOUN
easat-2167	245	10	,	,	PUNCT
easat-2167	245	11	suggesting	suggest	VERB
easat-2167	245	12	that	that	SCONJ
easat-2167	245	13	while	while	SCONJ
easat-2167	245	14	the	the	DET
easat-2167	245	15	model	model	NOUN
easat-2167	245	16	could	could	AUX
easat-2167	245	17	reach	reach	VERB
easat-2167	245	18	high	high	ADJ
easat-2167	245	19	accuracy	accuracy	NOUN
easat-2167	245	20	,	,	PUNCT
easat-2167	245	21	its	its	PRON
easat-2167	245	22	predictive	predictive	ADJ
easat-2167	245	23	stability	stability	NOUN
easat-2167	245	24	was	be	AUX
easat-2167	245	25	inconsistent	inconsistent	ADJ
easat-2167	245	26	.	.	PUNCT
easat-2167	246	1	806	806	NUM
easat-2167	246	2	edelweiss	edelweiss	PROPN
easat-2167	246	3	applied	apply	VERB
easat-2167	246	4	science	science	NOUN
easat-2167	246	5	and	and	CCONJ
easat-2167	246	6	technology	technology	NOUN
easat-2167	246	7	issn	issn	PROPN
easat-2167	246	8	:	:	PUNCT
easat-2167	246	9	2576	2576	NUM
easat-2167	246	10	-	-	SYM
easat-2167	246	11	8484	8484	NUM
easat-2167	246	12	vol	vol	NOUN
easat-2167	246	13	.	.	PROPN
easat-2167	246	14	8	8	NUM
easat-2167	246	15	,	,	PUNCT
easat-2167	246	16	no	no	INTJ
easat-2167	246	17	.	.	NOUN
easat-2167	247	1	6	6	NUM
easat-2167	247	2	:	:	PUNCT
easat-2167	247	3	793	793	NUM
easat-2167	247	4	-	-	NUM
easat-2167	247	5	813	813	NUM
easat-2167	247	6	,	,	PUNCT
easat-2167	247	7	2024	2024	NUM
easat-2167	247	8	doi	doi	NOUN
easat-2167	247	9	:	:	PUNCT
easat-2167	247	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	247	11	©	©	NOUN
easat-2167	247	12	2024	2024	NUM
easat-2167	247	13	by	by	ADP
easat-2167	247	14	the	the	DET
easat-2167	247	15	authors	author	NOUN
easat-2167	247	16	;	;	PUNCT
easat-2167	247	17	licensee	licensee	PROPN
easat-2167	247	18	learning	learn	VERB
easat-2167	247	19	gate	gate	NOUN
easat-2167	247	20	figure	figure	NOUN
easat-2167	247	21	4	4	NUM
easat-2167	247	22	.	.	PUNCT
easat-2167	247	23	blast	blast	NOUN
easat-2167	247	24	(	(	PUNCT
easat-2167	247	25	original	original	ADJ
easat-2167	247	26	image	image	NOUN
easat-2167	247	27	,	,	PUNCT
easat-2167	247	28	predicted	predict	VERB
easat-2167	247	29	mask	mask	NOUN
easat-2167	247	30	,	,	PUNCT
easat-2167	247	31	true	true	ADJ
easat-2167	247	32	mask	mask	NOUN
easat-2167	247	33	)	)	PUNCT
easat-2167	247	34	model	model	NOUN
easat-2167	247	35	-	-	PUNCT
easat-2167	247	36	i.	i.	NOUN
easat-2167	247	37	•	•	ADP
easat-2167	247	38	final	final	ADJ
easat-2167	247	39	training	training	NOUN
easat-2167	247	40	accuracy	accuracy	NOUN
easat-2167	247	41	:	:	PUNCT
easat-2167	247	42	the	the	DET
easat-2167	247	43	model	model	NOUN
easat-2167	247	44	achieved	achieve	VERB
easat-2167	247	45	a	a	DET
easat-2167	247	46	final	final	ADJ
easat-2167	247	47	accuracy	accuracy	NOUN
easat-2167	247	48	of	of	ADP
easat-2167	247	49	78.95	78.95	NUM
easat-2167	247	50	%	%	NOUN
easat-2167	247	51	on	on	ADP
easat-2167	247	52	the	the	DET
easat-2167	247	53	training	training	NOUN
easat-2167	247	54	data	datum	NOUN
easat-2167	247	55	,	,	PUNCT
easat-2167	247	56	with	with	ADP
easat-2167	247	57	a	a	DET
easat-2167	247	58	corresponding	corresponding	ADJ
easat-2167	247	59	loss	loss	NOUN
easat-2167	247	60	of	of	ADP
easat-2167	247	61	0.1909	0.1909	NUM
easat-2167	247	62	.	.	PUNCT
easat-2167	248	1	•	•	NUM
easat-2167	248	2	final	final	ADJ
easat-2167	248	3	validation	validation	NOUN
easat-2167	248	4	performance	performance	NOUN
easat-2167	248	5	:	:	PUNCT
easat-2167	248	6	•	•	NUM
easat-2167	248	7	validation	validation	NOUN
easat-2167	248	8	loss	loss	NOUN
easat-2167	248	9	:	:	PUNCT
easat-2167	248	10	the	the	DET
easat-2167	248	11	model	model	NOUN
easat-2167	248	12	's	's	PART
easat-2167	248	13	overall	overall	ADJ
easat-2167	248	14	validation	validation	NOUN
easat-2167	248	15	loss	loss	NOUN
easat-2167	248	16	was	be	AUX
easat-2167	248	17	0.2052	0.2052	NUM
easat-2167	248	18	,	,	PUNCT
easat-2167	248	19	which	which	PRON
easat-2167	248	20	indicates	indicate	VERB
easat-2167	248	21	the	the	DET
easat-2167	248	22	average	average	ADJ
easat-2167	248	23	difference	difference	NOUN
easat-2167	248	24	between	between	ADP
easat-2167	248	25	the	the	DET
easat-2167	248	26	predicted	predict	VERB
easat-2167	248	27	and	and	CCONJ
easat-2167	248	28	actual	actual	ADJ
easat-2167	248	29	outputs	output	NOUN
easat-2167	248	30	on	on	ADP
easat-2167	248	31	the	the	DET
easat-2167	248	32	validation	validation	NOUN
easat-2167	248	33	set	set	NOUN
easat-2167	248	34	.	.	PUNCT
easat-2167	249	1	•	•	NUM
easat-2167	249	2	validation	validation	NOUN
easat-2167	249	3	accuracy	accuracy	NOUN
easat-2167	249	4	:	:	PUNCT
easat-2167	249	5	the	the	DET
easat-2167	249	6	model	model	NOUN
easat-2167	249	7	's	's	PART
easat-2167	249	8	overall	overall	ADJ
easat-2167	249	9	validation	validation	NOUN
easat-2167	249	10	accuracy	accuracy	NOUN
easat-2167	249	11	was	be	AUX
easat-2167	249	12	75.95	75.95	NUM
easat-2167	249	13	%	%	NOUN
easat-2167	249	14	,	,	PUNCT
easat-2167	249	15	reflecting	reflect	VERB
easat-2167	249	16	its	its	PRON
easat-2167	249	17	ability	ability	NOUN
easat-2167	249	18	to	to	PART
easat-2167	249	19	generalize	generalize	VERB
easat-2167	249	20	to	to	ADP
easat-2167	249	21	unseen	unseen	ADJ
easat-2167	249	22	data	datum	NOUN
easat-2167	249	23	.	.	PUNCT
easat-2167	250	1	the	the	DET
easat-2167	250	2	mobilenetv2	mobilenetv2	PROPN
easat-2167	250	3	model	model	NOUN
easat-2167	250	4	demonstrated	demonstrate	VERB
easat-2167	250	5	a	a	DET
easat-2167	250	6	decent	decent	ADJ
easat-2167	250	7	overall	overall	ADJ
easat-2167	250	8	performance	performance	NOUN
easat-2167	250	9	with	with	ADP
easat-2167	250	10	a	a	DET
easat-2167	250	11	training	training	NOUN
easat-2167	250	12	accuracy	accuracy	NOUN
easat-2167	250	13	of	of	ADP
easat-2167	250	14	78.95	78.95	NUM
easat-2167	250	15	%	%	NOUN
easat-2167	250	16	and	and	CCONJ
easat-2167	250	17	a	a	DET
easat-2167	250	18	validation	validation	NOUN
easat-2167	250	19	accuracy	accuracy	NOUN
easat-2167	250	20	of	of	ADP
easat-2167	250	21	75.95	75.95	NUM
easat-2167	250	22	%	%	NOUN
easat-2167	250	23	.	.	PUNCT
easat-2167	251	1	the	the	DET
easat-2167	251	2	slightly	slightly	ADV
easat-2167	251	3	higher	high	ADJ
easat-2167	251	4	training	training	NOUN
easat-2167	251	5	accuracy	accuracy	NOUN
easat-2167	251	6	compared	compare	VERB
easat-2167	251	7	to	to	ADP
easat-2167	251	8	validation	validation	NOUN
easat-2167	251	9	accuracy	accuracy	NOUN
easat-2167	251	10	suggests	suggest	VERB
easat-2167	251	11	some	some	DET
easat-2167	251	12	overfitting	overfitting	NOUN
easat-2167	251	13	,	,	PUNCT
easat-2167	251	14	but	but	CCONJ
easat-2167	251	15	the	the	DET
easat-2167	251	16	model	model	NOUN
easat-2167	251	17	performed	perform	VERB
easat-2167	251	18	reasonably	reasonably	ADV
easat-2167	251	19	well	well	ADV
easat-2167	251	20	on	on	ADP
easat-2167	251	21	the	the	DET
easat-2167	251	22	validation	validation	NOUN
easat-2167	251	23	set	set	NOUN
easat-2167	251	24	,	,	PUNCT
easat-2167	251	25	indicating	indicate	VERB
easat-2167	251	26	good	good	ADJ
easat-2167	251	27	generalization	generalization	NOUN
easat-2167	251	28	with	with	ADP
easat-2167	251	29	some	some	DET
easat-2167	251	30	room	room	NOUN
easat-2167	251	31	for	for	ADP
easat-2167	251	32	improvement	improvement	NOUN
easat-2167	251	33	.	.	PUNCT
easat-2167	252	1	figure	figure	NOUN
easat-2167	252	2	5	5	NUM
easat-2167	252	3	.	.	PUNCT
easat-2167	252	4	loss	loss	NOUN
easat-2167	252	5	and	and	CCONJ
easat-2167	252	6	accuracy	accuracy	NOUN
easat-2167	252	7	graph	graph	NOUN
easat-2167	252	8	.	.	PUNCT
easat-2167	253	1	807	807	NUM
easat-2167	253	2	edelweiss	edelweiss	PROPN
easat-2167	253	3	applied	apply	VERB
easat-2167	253	4	science	science	NOUN
easat-2167	253	5	and	and	CCONJ
easat-2167	253	6	technology	technology	NOUN
easat-2167	253	7	issn	issn	PROPN
easat-2167	253	8	:	:	PUNCT
easat-2167	253	9	2576	2576	NUM
easat-2167	253	10	-	-	SYM
easat-2167	253	11	8484	8484	NUM
easat-2167	253	12	vol	vol	NOUN
easat-2167	253	13	.	.	PROPN
easat-2167	254	1	8	8	NUM
easat-2167	254	2	,	,	PUNCT
easat-2167	254	3	no	no	INTJ
easat-2167	254	4	.	.	NOUN
easat-2167	255	1	6	6	NUM
easat-2167	255	2	:	:	PUNCT
easat-2167	255	3	793	793	NUM
easat-2167	255	4	-	-	NUM
easat-2167	255	5	813	813	NUM
easat-2167	255	6	,	,	PUNCT
easat-2167	255	7	2024	2024	NUM
easat-2167	255	8	doi	doi	NOUN
easat-2167	255	9	:	:	PUNCT
easat-2167	255	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	255	11	©	©	NOUN
easat-2167	255	12	2024	2024	NUM
easat-2167	255	13	by	by	ADP
easat-2167	255	14	the	the	DET
easat-2167	255	15	authors	author	NOUN
easat-2167	255	16	;	;	PUNCT
easat-2167	255	17	licensee	licensee	PROPN
easat-2167	255	18	learning	learning	NOUN
easat-2167	255	19	gate	gate	PROPN
easat-2167	255	20	8.2	8.2	NUM
easat-2167	255	21	.	.	PUNCT
easat-2167	256	1	model	model	PROPN
easat-2167	256	2	-	-	PUNCT
easat-2167	256	3	ii	ii	PROPN
easat-2167	256	4	of	of	ADP
easat-2167	256	5	the	the	DET
easat-2167	256	6	simple	simple	ADJ
easat-2167	256	7	deeplabv3	deeplabv3	NOUN
easat-2167	256	8	+	+	CCONJ
easat-2167	256	9	architecture	architecture	NOUN
easat-2167	256	10	using	use	VERB
easat-2167	256	11	efficientnetb4	efficientnetb4	PROPN
easat-2167	256	12	8.2.1	8.2.1	NUM
easat-2167	256	13	.	.	PUNCT
easat-2167	256	14	initial	initial	ADJ
easat-2167	256	15	observations	observation	NOUN
easat-2167	256	16	1	1	NUM
easat-2167	256	17	.	.	PUNCT
easat-2167	256	18	epoch	epoch	PROPN
easat-2167	256	19	1	1	NUM
easat-2167	256	20	:	:	PUNCT
easat-2167	256	21	the	the	DET
easat-2167	256	22	model	model	NOUN
easat-2167	256	23	starts	start	VERB
easat-2167	256	24	with	with	ADP
easat-2167	256	25	an	an	DET
easat-2167	256	26	accuracy	accuracy	NOUN
easat-2167	256	27	of	of	ADP
easat-2167	256	28	48.09	48.09	NUM
easat-2167	256	29	%	%	NOUN
easat-2167	256	30	on	on	ADP
easat-2167	256	31	the	the	DET
easat-2167	256	32	training	training	NOUN
easat-2167	256	33	set	set	NOUN
easat-2167	256	34	and	and	CCONJ
easat-2167	256	35	61.79	61.79	NUM
easat-2167	256	36	%	%	NOUN
easat-2167	256	37	on	on	ADP
easat-2167	256	38	the	the	DET
easat-2167	256	39	validation	validation	NOUN
easat-2167	256	40	set	set	NOUN
easat-2167	256	41	,	,	PUNCT
easat-2167	256	42	with	with	ADP
easat-2167	256	43	corresponding	correspond	VERB
easat-2167	256	44	loss	loss	NOUN
easat-2167	256	45	values	value	NOUN
easat-2167	256	46	of	of	ADP
easat-2167	256	47	0.7419	0.7419	NUM
easat-2167	256	48	and	and	CCONJ
easat-2167	256	49	0.8259	0.8259	NUM
easat-2167	256	50	,	,	PUNCT
easat-2167	256	51	respectively	respectively	ADV
easat-2167	256	52	.	.	PUNCT
easat-2167	257	1	2	2	X
easat-2167	257	2	.	.	X
easat-2167	257	3	overfitting	overfitte	VERB
easat-2167	257	4	concerns	concern	NOUN
easat-2167	257	5	:	:	PUNCT
easat-2167	257	6	early	early	ADV
easat-2167	257	7	in	in	ADP
easat-2167	257	8	the	the	DET
easat-2167	257	9	training	training	NOUN
easat-2167	257	10	(	(	PUNCT
easat-2167	257	11	around	around	ADP
easat-2167	257	12	epoch	epoch	NOUN
easat-2167	257	13	2	2	NUM
easat-2167	257	14	to	to	ADP
easat-2167	257	15	epoch	epoch	NOUN
easat-2167	257	16	4	4	NUM
easat-2167	257	17	)	)	PUNCT
easat-2167	257	18	,	,	PUNCT
easat-2167	257	19	the	the	DET
easat-2167	257	20	validation	validation	NOUN
easat-2167	257	21	loss	loss	NOUN
easat-2167	257	22	increases	increase	VERB
easat-2167	257	23	significantly	significantly	ADV
easat-2167	257	24	(	(	PUNCT
easat-2167	257	25	up	up	ADP
easat-2167	257	26	to	to	ADP
easat-2167	257	27	2.0053	2.0053	NUM
easat-2167	257	28	)	)	PUNCT
easat-2167	257	29	while	while	SCONJ
easat-2167	257	30	the	the	DET
easat-2167	257	31	validation	validation	NOUN
easat-2167	257	32	accuracy	accuracy	NOUN
easat-2167	257	33	drops	drop	NOUN
easat-2167	257	34	,	,	PUNCT
easat-2167	257	35	indicating	indicate	VERB
easat-2167	257	36	possible	possible	ADJ
easat-2167	257	37	overfitting	overfitting	NOUN
easat-2167	257	38	.	.	PUNCT
easat-2167	258	1	8.2.2	8.2.2	X
easat-2167	258	2	.	.	PUNCT
easat-2167	258	3	mid	mid	ADJ
easat-2167	258	4	training	training	NOUN
easat-2167	258	5	observations	observation	NOUN
easat-2167	258	6	3	3	NUM
easat-2167	258	7	.	.	PUNCT
easat-2167	258	8	improvement	improvement	NOUN
easat-2167	258	9	phase	phase	NOUN
easat-2167	258	10	:	:	PUNCT
easat-2167	258	11	by	by	ADP
easat-2167	258	12	epoch	epoch	PROPN
easat-2167	258	13	7	7	NUM
easat-2167	258	14	,	,	PUNCT
easat-2167	258	15	the	the	DET
easat-2167	258	16	model	model	NOUN
easat-2167	258	17	shows	show	VERB
easat-2167	258	18	substantial	substantial	ADJ
easat-2167	258	19	improvement	improvement	NOUN
easat-2167	258	20	,	,	PUNCT
easat-2167	258	21	with	with	ADP
easat-2167	258	22	a	a	DET
easat-2167	258	23	validation	validation	NOUN
easat-2167	258	24	accuracy	accuracy	NOUN
easat-2167	258	25	of	of	ADP
easat-2167	258	26	85.9	85.9	NUM
easat-2167	258	27	%	%	NOUN
easat-2167	258	28	and	and	CCONJ
easat-2167	258	29	a	a	DET
easat-2167	258	30	reduced	reduced	ADJ
easat-2167	258	31	loss	loss	NOUN
easat-2167	258	32	of	of	ADP
easat-2167	258	33	0.3456	0.3456	NUM
easat-2167	258	34	.	.	PUNCT
easat-2167	259	1	this	this	PRON
easat-2167	259	2	suggests	suggest	VERB
easat-2167	259	3	that	that	SCONJ
easat-2167	259	4	the	the	DET
easat-2167	259	5	model	model	NOUN
easat-2167	259	6	is	be	AUX
easat-2167	259	7	learning	learn	VERB
easat-2167	259	8	to	to	PART
easat-2167	259	9	generalize	generalize	VERB
easat-2167	259	10	better	well	ADV
easat-2167	259	11	.	.	PUNCT
easat-2167	260	1	4	4	X
easat-2167	260	2	.	.	X
easat-2167	260	3	fluctuations	fluctuation	NOUN
easat-2167	260	4	:	:	PUNCT
easat-2167	260	5	throughout	throughout	ADP
easat-2167	260	6	the	the	DET
easat-2167	260	7	training	training	NOUN
easat-2167	260	8	,	,	PUNCT
easat-2167	260	9	the	the	DET
easat-2167	260	10	validation	validation	NOUN
easat-2167	260	11	accuracy	accuracy	NOUN
easat-2167	260	12	fluctuates	fluctuate	NOUN
easat-2167	260	13	.	.	PUNCT
easat-2167	261	1	for	for	ADP
easat-2167	261	2	example	example	NOUN
easat-2167	261	3	,	,	PUNCT
easat-2167	261	4	around	around	ADP
easat-2167	261	5	epoch	epoch	NOUN
easat-2167	261	6	21	21	NUM
easat-2167	261	7	,	,	PUNCT
easat-2167	261	8	there	there	PRON
easat-2167	261	9	is	be	VERB
easat-2167	261	10	a	a	DET
easat-2167	261	11	dip	dip	NOUN
easat-2167	261	12	in	in	ADP
easat-2167	261	13	performance	performance	NOUN
easat-2167	261	14	with	with	ADP
easat-2167	261	15	a	a	DET
easat-2167	261	16	validation	validation	NOUN
easat-2167	261	17	accuracy	accuracy	NOUN
easat-2167	261	18	of	of	ADP
easat-2167	261	19	77.74	77.74	NUM
easat-2167	261	20	%	%	NOUN
easat-2167	261	21	and	and	CCONJ
easat-2167	261	22	an	an	DET
easat-2167	261	23	increase	increase	NOUN
easat-2167	261	24	in	in	ADP
easat-2167	261	25	loss	loss	NOUN
easat-2167	261	26	to	to	ADP
easat-2167	261	27	0.3079	0.3079	NUM
easat-2167	261	28	.	.	PUNCT
easat-2167	262	1	8.2.3	8.2.3	NUM
easat-2167	262	2	.	.	PUNCT
easat-2167	263	1	final	final	ADJ
easat-2167	263	2	observations	observation	NOUN
easat-2167	263	3	5	5	NUM
easat-2167	263	4	.	.	PUNCT
easat-2167	263	5	final	final	ADJ
easat-2167	263	6	epochs	epoch	NOUN
easat-2167	263	7	:	:	PUNCT
easat-2167	263	8	by	by	ADP
easat-2167	263	9	epoch	epoch	PROPN
easat-2167	263	10	47	47	NUM
easat-2167	263	11	,	,	PUNCT
easat-2167	263	12	the	the	DET
easat-2167	263	13	model	model	NOUN
easat-2167	263	14	reaches	reach	VERB
easat-2167	263	15	a	a	DET
easat-2167	263	16	validation	validation	NOUN
easat-2167	263	17	accuracy	accuracy	NOUN
easat-2167	263	18	of	of	ADP
easat-2167	263	19	84.91	84.91	NUM
easat-2167	263	20	%	%	NOUN
easat-2167	263	21	with	with	ADP
easat-2167	263	22	a	a	DET
easat-2167	263	23	validation	validation	NOUN
easat-2167	263	24	loss	loss	NOUN
easat-2167	263	25	of	of	ADP
easat-2167	263	26	0.1500	0.1500	NUM
easat-2167	263	27	,	,	PUNCT
easat-2167	263	28	indicating	indicate	VERB
easat-2167	263	29	strong	strong	ADJ
easat-2167	263	30	performance	performance	NOUN
easat-2167	263	31	.	.	PUNCT
easat-2167	264	1	figure	figure	NOUN
easat-2167	264	2	6	6	NUM
easat-2167	264	3	.	.	PUNCT
easat-2167	264	4	loss	loss	NOUN
easat-2167	264	5	and	and	CCONJ
easat-2167	264	6	accuracy	accuracy	NOUN
easat-2167	264	7	graph	graph	NOUN
easat-2167	264	8	.	.	PUNCT
easat-2167	265	1	808	808	NUM
easat-2167	265	2	edelweiss	edelweiss	PROPN
easat-2167	265	3	applied	apply	VERB
easat-2167	265	4	science	science	NOUN
easat-2167	265	5	and	and	CCONJ
easat-2167	265	6	technology	technology	NOUN
easat-2167	265	7	issn	issn	PROPN
easat-2167	265	8	:	:	PUNCT
easat-2167	265	9	2576	2576	NUM
easat-2167	265	10	-	-	SYM
easat-2167	265	11	8484	8484	NUM
easat-2167	265	12	vol	vol	NOUN
easat-2167	265	13	.	.	PROPN
easat-2167	266	1	8	8	NUM
easat-2167	266	2	,	,	PUNCT
easat-2167	266	3	no	no	INTJ
easat-2167	266	4	.	.	NOUN
easat-2167	267	1	6	6	NUM
easat-2167	267	2	:	:	PUNCT
easat-2167	267	3	793	793	NUM
easat-2167	267	4	-	-	NUM
easat-2167	267	5	813	813	NUM
easat-2167	267	6	,	,	PUNCT
easat-2167	267	7	2024	2024	NUM
easat-2167	267	8	doi	doi	NOUN
easat-2167	267	9	:	:	PUNCT
easat-2167	267	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	267	11	©	©	NOUN
easat-2167	267	12	2024	2024	NUM
easat-2167	267	13	by	by	ADP
easat-2167	267	14	the	the	DET
easat-2167	267	15	authors	author	NOUN
easat-2167	267	16	;	;	PUNCT
easat-2167	267	17	licensee	licensee	PROPN
easat-2167	267	18	learning	learn	VERB
easat-2167	267	19	gate	gate	NOUN
easat-2167	267	20	figure	figure	NOUN
easat-2167	267	21	7	7	NUM
easat-2167	267	22	.	.	PUNCT
easat-2167	267	23	blast	blast	NOUN
easat-2167	267	24	(	(	PUNCT
easat-2167	267	25	original	original	ADJ
easat-2167	267	26	image	image	NOUN
easat-2167	267	27	,	,	PUNCT
easat-2167	267	28	predicted	predict	VERB
easat-2167	267	29	mask	mask	NOUN
easat-2167	267	30	,	,	PUNCT
easat-2167	267	31	true	true	ADJ
easat-2167	267	32	mask	mask	NOUN
easat-2167	267	33	)	)	PUNCT
easat-2167	267	34	model	model	PROPN
easat-2167	267	35	-	-	PUNCT
easat-2167	267	36	ii	ii	NOUN
easat-2167	267	37	.	.	PUNCT
easat-2167	268	1	the	the	DET
easat-2167	268	2	model	model	NOUN
easat-2167	268	3	shows	show	VERB
easat-2167	268	4	a	a	DET
easat-2167	268	5	general	general	ADJ
easat-2167	268	6	trend	trend	NOUN
easat-2167	268	7	of	of	ADP
easat-2167	268	8	improving	improve	VERB
easat-2167	268	9	accuracy	accuracy	NOUN
easat-2167	268	10	and	and	CCONJ
easat-2167	268	11	reducing	reduce	VERB
easat-2167	268	12	loss	loss	NOUN
easat-2167	268	13	on	on	ADP
easat-2167	268	14	both	both	CCONJ
easat-2167	268	15	the	the	DET
easat-2167	268	16	training	training	NOUN
easat-2167	268	17	and	and	CCONJ
easat-2167	268	18	validation	validation	NOUN
easat-2167	268	19	datasets	dataset	NOUN
easat-2167	268	20	,	,	PUNCT
easat-2167	268	21	with	with	ADP
easat-2167	268	22	some	some	DET
easat-2167	268	23	fluctuations	fluctuation	NOUN
easat-2167	268	24	.	.	PUNCT
easat-2167	269	1	the	the	DET
easat-2167	269	2	final	final	ADJ
easat-2167	269	3	validation	validation	NOUN
easat-2167	269	4	accuracy	accuracy	NOUN
easat-2167	269	5	of	of	ADP
easat-2167	269	6	84.91	84.91	NUM
easat-2167	269	7	%	%	NOUN
easat-2167	269	8	suggests	suggest	VERB
easat-2167	269	9	that	that	SCONJ
easat-2167	269	10	the	the	DET
easat-2167	269	11	model	model	NOUN
easat-2167	269	12	is	be	AUX
easat-2167	269	13	performing	perform	VERB
easat-2167	269	14	quite	quite	ADV
easat-2167	269	15	well	well	ADV
easat-2167	269	16	,	,	PUNCT
easat-2167	269	17	though	though	SCONJ
easat-2167	269	18	there	there	PRON
easat-2167	269	19	may	may	AUX
easat-2167	269	20	still	still	ADV
easat-2167	269	21	be	be	AUX
easat-2167	269	22	room	room	NOUN
easat-2167	269	23	for	for	ADP
easat-2167	269	24	fine	fine	ADV
easat-2167	269	25	-	-	PUNCT
easat-2167	269	26	tuning	tuning	NOUN
easat-2167	269	27	to	to	PART
easat-2167	269	28	minimize	minimize	VERB
easat-2167	269	29	fluctuations	fluctuation	NOUN
easat-2167	269	30	and	and	CCONJ
easat-2167	269	31	further	far	ADV
easat-2167	269	32	improve	improve	VERB
easat-2167	269	33	performance	performance	NOUN
easat-2167	269	34	.	.	PUNCT
easat-2167	270	1	8.3	8.3	NUM
easat-2167	270	2	.	.	PUNCT
easat-2167	270	3	model	model	NOUN
easat-2167	270	4	iii	iii	NUM
easat-2167	270	5	-results	-result	NOUN
easat-2167	270	6	of	of	ADP
easat-2167	270	7	proposed	propose	VERB
easat-2167	270	8	unet	unet	NOUN
easat-2167	270	9	efficient	efficient	ADJ
easat-2167	270	10	netb7with	netb7with	VERB
easat-2167	270	11	attention	attention	NOUN
easat-2167	270	12	model	model	NOUN
easat-2167	270	13	the	the	DET
easat-2167	270	14	training	training	NOUN
easat-2167	270	15	log	log	NOUN
easat-2167	270	16	of	of	ADP
easat-2167	270	17	the	the	DET
easat-2167	270	18	unetefficientnetb7withattention	unetefficientnetb7withattention	NOUN
easat-2167	270	19	model	model	NOUN
easat-2167	270	20	reveals	reveal	VERB
easat-2167	270	21	several	several	ADJ
easat-2167	270	22	insights	insight	NOUN
easat-2167	270	23	into	into	ADP
easat-2167	270	24	its	its	PRON
easat-2167	270	25	performance	performance	NOUN
easat-2167	270	26	over	over	ADP
easat-2167	270	27	the	the	DET
easat-2167	270	28	course	course	NOUN
easat-2167	270	29	of	of	ADP
easat-2167	270	30	48	48	NUM
easat-2167	270	31	epochs	epoch	NOUN
easat-2167	270	32	:	:	PUNCT
easat-2167	270	33	8.3.1	8.3.1	NUM
easat-2167	270	34	.	.	PUNCT
easat-2167	270	35	accuracy	accuracy	NOUN
easat-2167	270	36	and	and	CCONJ
easat-2167	270	37	loss	loss	NOUN
easat-2167	270	38	trends	trend	NOUN
easat-2167	270	39	•	•	NOUN
easat-2167	270	40	training	training	NOUN
easat-2167	270	41	accuracy	accuracy	NOUN
easat-2167	270	42	:	:	PUNCT
easat-2167	270	43	the	the	DET
easat-2167	270	44	model	model	NOUN
easat-2167	270	45	's	's	PART
easat-2167	270	46	accuracy	accuracy	NOUN
easat-2167	270	47	improved	improve	VERB
easat-2167	270	48	significantly	significantly	ADV
easat-2167	270	49	from	from	ADP
easat-2167	270	50	0.5361	0.5361	NUM
easat-2167	270	51	in	in	ADP
easat-2167	270	52	the	the	DET
easat-2167	270	53	first	first	ADJ
easat-2167	270	54	epoch	epoch	NOUN
easat-2167	270	55	to	to	ADP
easat-2167	270	56	values	value	NOUN
easat-2167	270	57	exceeding	exceed	VERB
easat-2167	270	58	0.8	0.8	NUM
easat-2167	270	59	by	by	ADP
easat-2167	270	60	epoch	epoch	PROPN
easat-2167	270	61	43	43	NUM
easat-2167	270	62	.	.	PUNCT
easat-2167	271	1	this	this	PRON
easat-2167	271	2	indicates	indicate	VERB
easat-2167	271	3	that	that	SCONJ
easat-2167	271	4	the	the	DET
easat-2167	271	5	model	model	NOUN
easat-2167	271	6	is	be	AUX
easat-2167	271	7	learning	learn	VERB
easat-2167	271	8	effectively	effectively	ADV
easat-2167	271	9	over	over	ADP
easat-2167	271	10	time	time	NOUN
easat-2167	271	11	.	.	PUNCT
easat-2167	272	1	•	•	NUM
easat-2167	272	2	validation	validation	NOUN
easat-2167	272	3	accuracy	accuracy	NOUN
easat-2167	272	4	:	:	PUNCT
easat-2167	272	5	similarly	similarly	ADV
easat-2167	272	6	,	,	PUNCT
easat-2167	272	7	validation	validation	NOUN
easat-2167	272	8	accuracy	accuracy	NOUN
easat-2167	272	9	also	also	ADV
easat-2167	272	10	showed	show	VERB
easat-2167	272	11	an	an	DET
easat-2167	272	12	upward	upward	ADJ
easat-2167	272	13	trend	trend	NOUN
easat-2167	272	14	,	,	PUNCT
easat-2167	272	15	peaking	peak	VERB
easat-2167	272	16	around	around	ADP
easat-2167	272	17	0.8431	0.8431	NUM
easat-2167	272	18	in	in	ADP
easat-2167	272	19	epoch	epoch	NOUN
easat-2167	272	20	44	44	NUM
easat-2167	272	21	.	.	PUNCT
easat-2167	273	1	however	however	ADV
easat-2167	273	2	,	,	PUNCT
easat-2167	273	3	there	there	PRON
easat-2167	273	4	were	be	VERB
easat-2167	273	5	fluctuations	fluctuation	NOUN
easat-2167	273	6	,	,	PUNCT
easat-2167	273	7	such	such	ADJ
easat-2167	273	8	as	as	ADP
easat-2167	273	9	a	a	DET
easat-2167	273	10	drop	drop	NOUN
easat-2167	273	11	in	in	ADP
easat-2167	273	12	epoch	epoch	PROPN
easat-2167	273	13	21	21	NUM
easat-2167	273	14	(	(	PUNCT
easat-2167	273	15	0.2946	0.2946	NUM
easat-2167	273	16	)	)	PUNCT
easat-2167	273	17	,	,	PUNCT
easat-2167	273	18	which	which	PRON
easat-2167	273	19	might	might	AUX
easat-2167	273	20	suggest	suggest	VERB
easat-2167	273	21	overfitting	overfitte	VERB
easat-2167	273	22	or	or	CCONJ
easat-2167	273	23	noisy	noisy	ADJ
easat-2167	273	24	validation	validation	NOUN
easat-2167	273	25	data	datum	NOUN
easat-2167	273	26	.	.	PUNCT
easat-2167	274	1	•	•	NUM
easat-2167	274	2	training	training	NOUN
easat-2167	274	3	loss	loss	NOUN
easat-2167	274	4	:	:	PUNCT
easat-2167	274	5	the	the	DET
easat-2167	274	6	loss	loss	NOUN
easat-2167	274	7	consistently	consistently	ADV
easat-2167	274	8	decreased	decrease	VERB
easat-2167	274	9	,	,	PUNCT
easat-2167	274	10	signifying	signify	VERB
easat-2167	274	11	that	that	SCONJ
easat-2167	274	12	the	the	DET
easat-2167	274	13	model	model	NOUN
easat-2167	274	14	is	be	AUX
easat-2167	274	15	getting	get	VERB
easat-2167	274	16	better	well	ADJ
easat-2167	274	17	at	at	ADP
easat-2167	274	18	minimizing	minimize	VERB
easat-2167	274	19	errors	error	NOUN
easat-2167	274	20	during	during	ADP
easat-2167	274	21	training	training	NOUN
easat-2167	274	22	.	.	PUNCT
easat-2167	275	1	•	•	NUM
easat-2167	275	2	validation	validation	NOUN
easat-2167	275	3	loss	loss	NOUN
easat-2167	275	4	:	:	PUNCT
easat-2167	275	5	the	the	DET
easat-2167	275	6	validation	validation	NOUN
easat-2167	275	7	loss	loss	NOUN
easat-2167	275	8	was	be	AUX
easat-2167	275	9	more	more	ADV
easat-2167	275	10	volatile	volatile	ADJ
easat-2167	275	11	,	,	PUNCT
easat-2167	275	12	with	with	ADP
easat-2167	275	13	significant	significant	ADJ
easat-2167	275	14	spikes	spike	NOUN
easat-2167	275	15	(	(	PUNCT
easat-2167	275	16	e.g.	e.g.	ADV
easat-2167	275	17	,	,	PUNCT
easat-2167	275	18	epoch	epoch	NOUN
easat-2167	275	19	2	2	NUM
easat-2167	275	20	at	at	ADP
easat-2167	275	21	6.9256	6.9256	NUM
easat-2167	275	22	and	and	CCONJ
easat-2167	275	23	epoch	epoch	PROPN
easat-2167	275	24	28	28	NUM
easat-2167	275	25	at	at	ADP
easat-2167	275	26	5.5382	5.5382	NUM
easat-2167	275	27	)	)	PUNCT
easat-2167	275	28	.	.	PUNCT
easat-2167	276	1	this	this	PRON
easat-2167	276	2	could	could	AUX
easat-2167	276	3	indicate	indicate	VERB
easat-2167	276	4	overfitting	overfitting	NOUN
easat-2167	276	5	,	,	PUNCT
easat-2167	276	6	where	where	SCONJ
easat-2167	276	7	the	the	DET
easat-2167	276	8	model	model	NOUN
easat-2167	276	9	performs	perform	VERB
easat-2167	276	10	well	well	ADV
easat-2167	276	11	on	on	ADP
easat-2167	276	12	training	training	NOUN
easat-2167	276	13	data	datum	NOUN
easat-2167	276	14	but	but	CCONJ
easat-2167	276	15	struggles	struggle	NOUN
easat-2167	276	16	with	with	ADP
easat-2167	276	17	unseen	unseen	ADJ
easat-2167	276	18	data	datum	NOUN
easat-2167	276	19	.	.	PUNCT
easat-2167	277	1	809	809	NUM
easat-2167	277	2	edelweiss	edelweiss	PROPN
easat-2167	277	3	applied	apply	VERB
easat-2167	277	4	science	science	NOUN
easat-2167	277	5	and	and	CCONJ
easat-2167	277	6	technology	technology	NOUN
easat-2167	277	7	issn	issn	PROPN
easat-2167	277	8	:	:	PUNCT
easat-2167	277	9	2576	2576	NUM
easat-2167	277	10	-	-	SYM
easat-2167	277	11	8484	8484	NUM
easat-2167	277	12	vol	vol	NOUN
easat-2167	277	13	.	.	PROPN
easat-2167	277	14	8	8	NUM
easat-2167	277	15	,	,	PUNCT
easat-2167	277	16	no	no	INTJ
easat-2167	277	17	.	.	NOUN
easat-2167	278	1	6	6	NUM
easat-2167	278	2	:	:	PUNCT
easat-2167	278	3	793	793	NUM
easat-2167	278	4	-	-	NUM
easat-2167	278	5	813	813	NUM
easat-2167	278	6	,	,	PUNCT
easat-2167	278	7	2024	2024	NUM
easat-2167	278	8	doi	doi	NOUN
easat-2167	278	9	:	:	PUNCT
easat-2167	278	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	278	11	©	©	NOUN
easat-2167	278	12	2024	2024	NUM
easat-2167	278	13	by	by	ADP
easat-2167	278	14	the	the	DET
easat-2167	278	15	authors	author	NOUN
easat-2167	278	16	;	;	PUNCT
easat-2167	278	17	licensee	licensee	PROPN
easat-2167	278	18	learning	learn	VERB
easat-2167	278	19	gate	gate	PROPN
easat-2167	278	20	figure	figure	NOUN
easat-2167	278	21	8	8	NUM
easat-2167	278	22	.	.	PUNCT
easat-2167	279	1	blast	blast	NOUN
easat-2167	279	2	(	(	PUNCT
easat-2167	279	3	original	original	ADJ
easat-2167	279	4	image	image	NOUN
easat-2167	279	5	,	,	PUNCT
easat-2167	279	6	predicted	predict	VERB
easat-2167	279	7	mask	mask	NOUN
easat-2167	279	8	,	,	PUNCT
easat-2167	279	9	true	true	ADJ
easat-2167	279	10	mask	mask	NOUN
easat-2167	279	11	)	)	PUNCT
easat-2167	279	12	model	model	NOUN
easat-2167	279	13	-	-	PUNCT
easat-2167	279	14	iii	iii	NOUN
easat-2167	279	15	.	.	PROPN
easat-2167	279	16	8.4	8.4	NUM
easat-2167	279	17	.	.	PUNCT
easat-2167	279	18	model	model	NOUN
easat-2167	279	19	performance	performance	NOUN
easat-2167	279	20	over	over	ADP
easat-2167	279	21	time	time	NOUN
easat-2167	279	22	•	•	ADP
easat-2167	279	23	early	early	ADJ
easat-2167	279	24	improvement	improvement	NOUN
easat-2167	279	25	:	:	PUNCT
easat-2167	279	26	from	from	ADP
easat-2167	279	27	epoch	epoch	PROPN
easat-2167	279	28	1	1	NUM
easat-2167	279	29	to	to	ADP
easat-2167	279	30	epoch	epoch	NOUN
easat-2167	279	31	8	8	NUM
easat-2167	279	32	,	,	PUNCT
easat-2167	279	33	both	both	DET
easat-2167	279	34	training	training	NOUN
easat-2167	279	35	and	and	CCONJ
easat-2167	279	36	validation	validation	NOUN
easat-2167	279	37	accuracies	accuracy	NOUN
easat-2167	279	38	increased	increase	VERB
easat-2167	279	39	steadily	steadily	ADV
easat-2167	279	40	,	,	PUNCT
easat-2167	279	41	and	and	CCONJ
easat-2167	279	42	losses	loss	NOUN
easat-2167	279	43	decreased	decrease	VERB
easat-2167	279	44	,	,	PUNCT
easat-2167	279	45	indicating	indicate	VERB
easat-2167	279	46	effective	effective	ADJ
easat-2167	279	47	learning	learning	NOUN
easat-2167	279	48	.	.	PUNCT
easat-2167	280	1	•	•	NUM
easat-2167	280	2	mid	mid	ADJ
easat-2167	280	3	training	training	NOUN
easat-2167	280	4	:	:	PUNCT
easat-2167	280	5	around	around	ADP
easat-2167	280	6	epochs	epoch	NOUN
easat-2167	280	7	20	20	NUM
easat-2167	280	8	-	-	SYM
easat-2167	280	9	30	30	NUM
easat-2167	280	10	,	,	PUNCT
easat-2167	280	11	there	there	PRON
easat-2167	280	12	were	be	VERB
easat-2167	280	13	some	some	DET
easat-2167	280	14	fluctuations	fluctuation	NOUN
easat-2167	280	15	in	in	ADP
easat-2167	280	16	validation	validation	NOUN
easat-2167	280	17	metrics	metric	NOUN
easat-2167	280	18	,	,	PUNCT
easat-2167	280	19	with	with	ADP
easat-2167	280	20	notable	notable	ADJ
easat-2167	280	21	drops	drop	NOUN
easat-2167	280	22	in	in	ADP
easat-2167	280	23	accuracy	accuracy	NOUN
easat-2167	280	24	and	and	CCONJ
easat-2167	280	25	spikes	spike	NOUN
easat-2167	280	26	in	in	ADP
easat-2167	280	27	loss	loss	NOUN
easat-2167	280	28	,	,	PUNCT
easat-2167	280	29	suggesting	suggest	VERB
easat-2167	280	30	possible	possible	ADJ
easat-2167	280	31	overfitting	overfitting	NOUN
easat-2167	280	32	or	or	CCONJ
easat-2167	280	33	encountering	encounter	VERB
easat-2167	280	34	difficult	difficult	ADJ
easat-2167	280	35	validation	validation	NOUN
easat-2167	280	36	batches	batch	NOUN
easat-2167	280	37	.	.	PUNCT
easat-2167	281	1	•	•	NUM
easat-2167	281	2	later	later	ADV
easat-2167	281	3	epochs	epoch	NOUN
easat-2167	281	4	:	:	PUNCT
easat-2167	281	5	despite	despite	SCONJ
easat-2167	281	6	fluctuations	fluctuation	NOUN
easat-2167	281	7	,	,	PUNCT
easat-2167	281	8	the	the	DET
easat-2167	281	9	model	model	NOUN
easat-2167	281	10	continued	continue	VERB
easat-2167	281	11	to	to	PART
easat-2167	281	12	improve	improve	VERB
easat-2167	281	13	overall	overall	ADV
easat-2167	281	14	,	,	PUNCT
easat-2167	281	15	reaching	reach	VERB
easat-2167	281	16	a	a	DET
easat-2167	281	17	high	high	ADJ
easat-2167	281	18	validation	validation	NOUN
easat-2167	281	19	accuracy	accuracy	NOUN
easat-2167	281	20	of	of	ADP
easat-2167	281	21	0.8432	0.8432	NUM
easat-2167	281	22	in	in	ADP
easat-2167	281	23	epoch	epoch	NOUN
easat-2167	281	24	48	48	NUM
easat-2167	281	25	with	with	ADP
easat-2167	281	26	a	a	DET
easat-2167	281	27	corresponding	corresponding	ADJ
easat-2167	281	28	decrease	decrease	NOUN
easat-2167	281	29	in	in	ADP
easat-2167	281	30	validation	validation	NOUN
easat-2167	281	31	loss	loss	NOUN
easat-2167	281	32	to	to	ADP
easat-2167	281	33	0.1261	0.1261	NUM
easat-2167	281	34	,	,	PUNCT
easat-2167	281	35	indicating	indicate	VERB
easat-2167	281	36	good	good	ADJ
easat-2167	281	37	generalization	generalization	NOUN
easat-2167	281	38	by	by	ADP
easat-2167	281	39	the	the	DET
easat-2167	281	40	end	end	NOUN
easat-2167	281	41	of	of	ADP
easat-2167	281	42	the	the	DET
easat-2167	281	43	recorded	record	VERB
easat-2167	281	44	epochs	epoch	NOUN
easat-2167	281	45	.	.	PUNCT
easat-2167	282	1	810	810	NUM
easat-2167	282	2	edelweiss	edelweiss	PROPN
easat-2167	282	3	applied	apply	VERB
easat-2167	282	4	science	science	NOUN
easat-2167	282	5	and	and	CCONJ
easat-2167	282	6	technology	technology	NOUN
easat-2167	282	7	issn	issn	PROPN
easat-2167	282	8	:	:	PUNCT
easat-2167	282	9	2576	2576	NUM
easat-2167	282	10	-	-	SYM
easat-2167	282	11	8484	8484	NUM
easat-2167	282	12	vol	vol	NOUN
easat-2167	282	13	.	.	PROPN
easat-2167	282	14	8	8	NUM
easat-2167	282	15	,	,	PUNCT
easat-2167	282	16	no	no	INTJ
easat-2167	282	17	.	.	NOUN
easat-2167	283	1	6	6	NUM
easat-2167	283	2	:	:	PUNCT
easat-2167	283	3	793	793	NUM
easat-2167	283	4	-	-	NUM
easat-2167	283	5	813	813	NUM
easat-2167	283	6	,	,	PUNCT
easat-2167	283	7	2024	2024	NUM
easat-2167	283	8	doi	doi	NOUN
easat-2167	283	9	:	:	PUNCT
easat-2167	283	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	283	11	©	©	NOUN
easat-2167	283	12	2024	2024	NUM
easat-2167	283	13	by	by	ADP
easat-2167	283	14	the	the	DET
easat-2167	283	15	authors	author	NOUN
easat-2167	283	16	;	;	PUNCT
easat-2167	283	17	licensee	licensee	PROPN
easat-2167	283	18	learning	learn	VERB
easat-2167	283	19	gate	gate	NOUN
easat-2167	283	20	figure	figure	NOUN
easat-2167	283	21	9	9	NUM
easat-2167	283	22	.	.	PUNCT
easat-2167	283	23	loss	loss	NOUN
easat-2167	283	24	and	and	CCONJ
easat-2167	283	25	accuracy	accuracy	NOUN
easat-2167	283	26	graph	graph	NOUN
easat-2167	283	27	.	.	PUNCT
easat-2167	284	1	here	here	ADV
easat-2167	284	2	is	be	AUX
easat-2167	284	3	a	a	DET
easat-2167	284	4	comparative	comparative	ADJ
easat-2167	284	5	table	table	NOUN
easat-2167	284	6	summarizing	summarize	VERB
easat-2167	284	7	the	the	DET
easat-2167	284	8	results	result	NOUN
easat-2167	284	9	of	of	ADP
easat-2167	284	10	the	the	DET
easat-2167	284	11	three	three	NUM
easat-2167	284	12	image	image	NOUN
easat-2167	284	13	segmentation	segmentation	NOUN
easat-2167	284	14	models	model	NOUN
easat-2167	284	15	:	:	PUNCT
easat-2167	284	16	table	table	NOUN
easat-2167	284	17	1	1	NUM
easat-2167	284	18	.	.	PUNCT
easat-2167	284	19	comparative	comparative	ADJ
easat-2167	284	20	summary	summary	NOUN
easat-2167	284	21	of	of	ADP
easat-2167	284	22	proposed	propose	VERB
easat-2167	284	23	models	model	NOUN
easat-2167	284	24	.	.	PUNCT
easat-2167	285	1	model	model	NOUN
easat-2167	285	2	validation	validation	NOUN
easat-2167	285	3	loss	loss	NOUN
easat-2167	285	4	validation	validation	NOUN
easat-2167	285	5	accuracy	accuracy	NOUN
easat-2167	285	6	efficiency	efficiency	NOUN
easat-2167	285	7	complexity	complexity	NOUN
easat-2167	285	8	segmentation	segmentation	NOUN
easat-2167	285	9	quality	quality	NOUN
easat-2167	285	10	u	u	NOUN
easat-2167	286	1	-	-	NOUN
easat-2167	287	1	net	net	ADJ
easat-2167	287	2	with	with	ADP
easat-2167	287	3	mobilenetv2	mobilenetv2	PROPN
easat-2167	287	4	0.2052	0.2052	NUM
easat-2167	287	5	0.7595	0.7595	NUM
easat-2167	287	6	highest	high	ADJ
easat-2167	287	7	efficiency	efficiency	NOUN
easat-2167	287	8	;	;	PUNCT
easat-2167	287	9	ideal	ideal	ADJ
easat-2167	287	10	for	for	ADP
easat-2167	287	11	real	real	ADJ
easat-2167	287	12	-	-	PUNCT
easat-2167	287	13	time	time	NOUN
easat-2167	287	14	use	use	NOUN
easat-2167	287	15	moderate	moderate	ADJ
easat-2167	287	16	;	;	PUNCT
easat-2167	287	17	lightweight	lightweight	ADJ
easat-2167	287	18	and	and	CCONJ
easat-2167	287	19	fast	fast	ADJ
easat-2167	287	20	good	good	ADJ
easat-2167	287	21	for	for	ADP
easat-2167	287	22	simpler	simple	ADJ
easat-2167	287	23	tasks	task	NOUN
easat-2167	287	24	;	;	PUNCT
easat-2167	287	25	may	may	AUX
easat-2167	287	26	struggle	struggle	VERB
easat-2167	287	27	with	with	ADP
easat-2167	287	28	complex	complex	ADJ
easat-2167	287	29	details	detail	NOUN
easat-2167	287	30	deeplabv3	deeplabv3	ADV
easat-2167	287	31	+	+	X
easat-2167	287	32	with	with	ADP
easat-2167	287	33	efficientnetb4	efficientnetb4	ADJ
easat-2167	287	34	0.2342	0.2342	NUM
easat-2167	287	35	0.7879	0.7879	NUM
easat-2167	287	36	moderate	moderate	ADJ
easat-2167	287	37	efficiency	efficiency	NOUN
easat-2167	287	38	;	;	PUNCT
easat-2167	287	39	balanced	balanced	ADJ
easat-2167	287	40	performance	performance	NOUN
easat-2167	287	41	higher	high	ADJ
easat-2167	287	42	than	than	ADP
easat-2167	287	43	mobilenetv2	mobilenetv2	NOUN
easat-2167	287	44	;	;	PUNCT
easat-2167	287	45	efficient	efficient	ADJ
easat-2167	287	46	and	and	CCONJ
easat-2167	287	47	powerful	powerful	ADJ
easat-2167	287	48	superior	superior	ADJ
easat-2167	287	49	to	to	ADP
easat-2167	287	50	mobilenetv2	mobilenetv2	PROPN
easat-2167	287	51	;	;	PUNCT
easat-2167	287	52	handles	handle	VERB
easat-2167	287	53	complex	complex	ADJ
easat-2167	287	54	boundaries	boundary	NOUN
easat-2167	287	55	well	well	ADV
easat-2167	287	56	u	u	NOUN
easat-2167	287	57	-	-	NOUN
easat-2167	287	58	net	net	ADJ
easat-2167	287	59	with	with	ADP
easat-2167	287	60	efficientnetb7	efficientnetb7	NOUN
easat-2167	287	61	and	and	CCONJ
easat-2167	287	62	attention	attention	NOUN
easat-2167	287	63	gates	gate	VERB
easat-2167	287	64	0.172	0.172	NUM
easat-2167	287	65	0.8028	0.8028	NUM
easat-2167	287	66	lowest	low	ADJ
easat-2167	287	67	efficiency	efficiency	NOUN
easat-2167	287	68	;	;	PUNCT
easat-2167	287	69	high	high	ADJ
easat-2167	287	70	computational	computational	ADJ
easat-2167	287	71	cost	cost	NOUN
easat-2167	287	72	highest	high	ADJ
easat-2167	287	73	;	;	PUNCT
easat-2167	287	74	most	most	ADV
easat-2167	287	75	powerful	powerful	ADJ
easat-2167	287	76	and	and	CCONJ
easat-2167	287	77	complex	complex	ADJ
easat-2167	287	78	best	good	ADJ
easat-2167	287	79	-	-	PUNCT
easat-2167	287	80	in	in	ADP
easat-2167	287	81	-	-	PUNCT
easat-2167	287	82	class	class	NOUN
easat-2167	287	83	;	;	PUNCT
easat-2167	287	84	excels	excel	NOUN
easat-2167	287	85	at	at	ADP
easat-2167	287	86	fine	fine	ADJ
easat-2167	287	87	details	detail	NOUN
easat-2167	287	88	and	and	CCONJ
easat-2167	287	89	complex	complex	ADJ
easat-2167	287	90	segmentation	segmentation	NOUN
easat-2167	287	91	811	811	NUM
easat-2167	287	92	edelweiss	edelweiss	PROPN
easat-2167	287	93	applied	apply	VERB
easat-2167	287	94	science	science	NOUN
easat-2167	287	95	and	and	CCONJ
easat-2167	287	96	technology	technology	NOUN
easat-2167	287	97	issn	issn	PROPN
easat-2167	287	98	:	:	PUNCT
easat-2167	287	99	2576	2576	NUM
easat-2167	287	100	-	-	SYM
easat-2167	287	101	8484	8484	NUM
easat-2167	287	102	vol	vol	NOUN
easat-2167	287	103	.	.	PROPN
easat-2167	288	1	8	8	NUM
easat-2167	288	2	,	,	PUNCT
easat-2167	288	3	no	no	INTJ
easat-2167	288	4	.	.	NOUN
easat-2167	289	1	6	6	NUM
easat-2167	289	2	:	:	PUNCT
easat-2167	289	3	793	793	NUM
easat-2167	289	4	-	-	NUM
easat-2167	289	5	813	813	NUM
easat-2167	289	6	,	,	PUNCT
easat-2167	289	7	2024	2024	NUM
easat-2167	289	8	doi	doi	NOUN
easat-2167	289	9	:	:	PUNCT
easat-2167	289	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	289	11	©	©	NOUN
easat-2167	289	12	2024	2024	NUM
easat-2167	289	13	by	by	ADP
easat-2167	289	14	the	the	DET
easat-2167	289	15	authors	author	NOUN
easat-2167	289	16	;	;	PUNCT
easat-2167	289	17	licensee	licensee	PROPN
easat-2167	289	18	learning	learning	NOUN
easat-2167	289	19	gate	gate	NOUN
easat-2167	289	20	this	this	DET
easat-2167	289	21	table	table	NOUN
easat-2167	289	22	provides	provide	VERB
easat-2167	289	23	a	a	DET
easat-2167	289	24	clear	clear	ADJ
easat-2167	289	25	comparison	comparison	NOUN
easat-2167	289	26	of	of	ADP
easat-2167	289	27	the	the	DET
easat-2167	289	28	three	three	NUM
easat-2167	289	29	models	model	NOUN
easat-2167	289	30	based	base	VERB
easat-2167	289	31	on	on	ADP
easat-2167	289	32	their	their	PRON
easat-2167	289	33	validation	validation	NOUN
easat-2167	289	34	results	result	NOUN
easat-2167	289	35	,	,	PUNCT
easat-2167	289	36	efficiency	efficiency	NOUN
easat-2167	289	37	,	,	PUNCT
easat-2167	289	38	complexity	complexity	NOUN
easat-2167	289	39	,	,	PUNCT
easat-2167	289	40	and	and	CCONJ
easat-2167	289	41	segmentation	segmentation	NOUN
easat-2167	289	42	quality	quality	NOUN
easat-2167	289	43	9	9	NUM
easat-2167	289	44	.	.	PUNCT
easat-2167	290	1	conclusions	conclusion	NOUN
easat-2167	290	2	and	and	CCONJ
easat-2167	290	3	future	future	ADJ
easat-2167	290	4	plans	plan	NOUN
easat-2167	290	5	this	this	DET
easat-2167	290	6	analysis	analysis	NOUN
easat-2167	290	7	focuses	focus	VERB
easat-2167	290	8	on	on	ADP
easat-2167	290	9	three	three	NUM
easat-2167	290	10	advanced	advanced	ADJ
easat-2167	290	11	models	model	NOUN
easat-2167	290	12	for	for	ADP
easat-2167	290	13	image	image	NOUN
easat-2167	290	14	segmentation	segmentation	NOUN
easat-2167	290	15	,	,	PUNCT
easat-2167	290	16	evaluating	evaluate	VERB
easat-2167	290	17	their	their	PRON
easat-2167	290	18	architecture	architecture	NOUN
easat-2167	290	19	,	,	PUNCT
easat-2167	290	20	efficiency	efficiency	NOUN
easat-2167	290	21	,	,	PUNCT
easat-2167	290	22	prediction	prediction	NOUN
easat-2167	290	23	accuracy	accuracy	NOUN
easat-2167	290	24	,	,	PUNCT
easat-2167	290	25	segmentation	segmentation	NOUN
easat-2167	290	26	performance	performance	NOUN
easat-2167	290	27	,	,	PUNCT
easat-2167	290	28	and	and	CCONJ
easat-2167	290	29	comparative	comparative	ADJ
easat-2167	290	30	results	result	NOUN
easat-2167	290	31	after	after	ADP
easat-2167	290	32	100	100	NUM
easat-2167	290	33	epochs	epoch	NOUN
easat-2167	290	34	of	of	ADP
easat-2167	290	35	training	training	NOUN
easat-2167	290	36	.	.	PUNCT
easat-2167	291	1	the	the	DET
easat-2167	291	2	models	model	NOUN
easat-2167	291	3	are	be	AUX
easat-2167	291	4	:	:	PUNCT
easat-2167	291	5	1	1	X
easat-2167	291	6	.	.	X
easat-2167	291	7	u	u	NOUN
easat-2167	291	8	-	-	NOUN
easat-2167	291	9	net	net	ADJ
easat-2167	291	10	with	with	ADP
easat-2167	291	11	mobilenetv2	mobilenetv2	PROPN
easat-2167	291	12	2	2	NUM
easat-2167	291	13	.	.	PUNCT
easat-2167	292	1	deeplabv3	deeplabv3	NOUN
easat-2167	292	2	+	+	X
easat-2167	292	3	with	with	ADP
easat-2167	292	4	efficientnetb4	efficientnetb4	PROPN
easat-2167	292	5	3	3	NUM
easat-2167	292	6	.	.	X
easat-2167	292	7	u	u	NOUN
easat-2167	292	8	-	-	NOUN
easat-2167	292	9	net	net	ADJ
easat-2167	292	10	with	with	ADP
easat-2167	292	11	efficientnetb7	efficientnetb7	NOUN
easat-2167	292	12	and	and	CCONJ
easat-2167	292	13	attention	attention	NOUN
easat-2167	292	14	gates	gate	NOUN
easat-2167	292	15	9.1	9.1	NUM
easat-2167	292	16	.	.	PUNCT
easat-2167	293	1	model	model	NOUN
easat-2167	293	2	architecture	architecture	NOUN
easat-2167	293	3	and	and	CCONJ
easat-2167	293	4	components	component	NOUN
easat-2167	293	5	•	•	PRON
easat-2167	293	6	u	u	NOUN
easat-2167	293	7	-	-	NOUN
easat-2167	293	8	net	net	ADJ
easat-2167	293	9	with	with	ADP
easat-2167	293	10	mobilenetv2	mobilenetv2	PROPN
easat-2167	293	11	:	:	PUNCT
easat-2167	293	12	•	•	NUM
easat-2167	293	13	encoder	encoder	NOUN
easat-2167	293	14	:	:	PUNCT
easat-2167	293	15	mobilenetv2	mobilenetv2	PROPN
easat-2167	293	16	,	,	PUNCT
easat-2167	293	17	optimized	optimize	VERB
easat-2167	293	18	for	for	ADP
easat-2167	293	19	efficiency	efficiency	NOUN
easat-2167	293	20	.	.	PUNCT
easat-2167	294	1	•	•	NUM
easat-2167	294	2	decoder	decoder	NOUN
easat-2167	294	3	:	:	PUNCT
easat-2167	294	4	standard	standard	ADJ
easat-2167	294	5	u	u	ADJ
easat-2167	294	6	-	-	ADJ
easat-2167	294	7	net	net	ADJ
easat-2167	294	8	decoder	decoder	NOUN
easat-2167	294	9	with	with	ADP
easat-2167	294	10	skip	skip	ADJ
easat-2167	294	11	connections	connection	NOUN
easat-2167	294	12	.	.	PUNCT
easat-2167	295	1	•	•	NUM
easat-2167	295	2	complexity	complexity	NOUN
easat-2167	295	3	:	:	PUNCT
easat-2167	295	4	moderate	moderate	ADJ
easat-2167	295	5	,	,	PUNCT
easat-2167	295	6	with	with	ADP
easat-2167	295	7	a	a	DET
easat-2167	295	8	focus	focus	NOUN
easat-2167	295	9	on	on	ADP
easat-2167	295	10	lightweight	lightweight	ADJ
easat-2167	295	11	and	and	CCONJ
easat-2167	295	12	fast	fast	ADJ
easat-2167	295	13	performance	performance	NOUN
easat-2167	295	14	.	.	PUNCT
easat-2167	296	1	•	•	NUM
easat-2167	296	2	key	key	ADJ
easat-2167	296	3	features	feature	NOUN
easat-2167	296	4	:	:	PUNCT
easat-2167	296	5	efficient	efficient	ADJ
easat-2167	296	6	for	for	ADP
easat-2167	296	7	mobile	mobile	ADJ
easat-2167	296	8	and	and	CCONJ
easat-2167	296	9	resource	resource	NOUN
easat-2167	296	10	-	-	PUNCT
easat-2167	296	11	constrained	constrain	VERB
easat-2167	296	12	environments	environment	NOUN
easat-2167	296	13	,	,	PUNCT
easat-2167	296	14	but	but	CCONJ
easat-2167	296	15	may	may	AUX
easat-2167	296	16	sacrifice	sacrifice	VERB
easat-2167	296	17	some	some	DET
easat-2167	296	18	accuracy	accuracy	NOUN
easat-2167	296	19	.	.	PUNCT
easat-2167	297	1	•	•	NUM
easat-2167	298	1	deeplabv3	deeplabv3	ADV
easat-2167	298	2	+	+	X
easat-2167	298	3	with	with	ADP
easat-2167	298	4	efficientnetb4	efficientnetb4	PROPN
easat-2167	298	5	:	:	PUNCT
easat-2167	298	6	•	•	NUM
easat-2167	298	7	encoder	encoder	NOUN
easat-2167	298	8	:	:	PUNCT
easat-2167	298	9	efficientnetb4	efficientnetb4	PROPN
easat-2167	298	10	,	,	PUNCT
easat-2167	298	11	designed	design	VERB
easat-2167	298	12	for	for	ADP
easat-2167	298	13	a	a	DET
easat-2167	298	14	balance	balance	NOUN
easat-2167	298	15	between	between	ADP
easat-2167	298	16	performance	performance	NOUN
easat-2167	298	17	and	and	CCONJ
easat-2167	298	18	efficiency	efficiency	NOUN
easat-2167	298	19	.	.	PUNCT
easat-2167	299	1	•	•	NUM
easat-2167	299	2	decoder	decoder	NOUN
easat-2167	299	3	:	:	PUNCT
easat-2167	299	4	convolutional	convolutional	ADJ
easat-2167	299	5	upsampling	upsample	VERB
easat-2167	299	6	layers	layer	NOUN
easat-2167	299	7	.	.	PUNCT
easat-2167	300	1	•	•	NUM
easat-2167	300	2	complexity	complexity	NOUN
easat-2167	300	3	:	:	PUNCT
easat-2167	300	4	higher	high	ADJ
easat-2167	300	5	than	than	ADP
easat-2167	300	6	u	u	NOUN
easat-2167	300	7	-	-	NOUN
easat-2167	300	8	net	net	ADJ
easat-2167	300	9	with	with	ADP
easat-2167	300	10	mobilenetv2	mobilenetv2	PROPN
easat-2167	300	11	,	,	PUNCT
easat-2167	300	12	balancing	balance	VERB
easat-2167	300	13	feature	feature	NOUN
easat-2167	300	14	extraction	extraction	NOUN
easat-2167	300	15	and	and	CCONJ
easat-2167	300	16	efficiency	efficiency	NOUN
easat-2167	300	17	.	.	PUNCT
easat-2167	301	1	•	•	NUM
easat-2167	301	2	key	key	ADJ
easat-2167	301	3	features	feature	NOUN
easat-2167	301	4	:	:	PUNCT
easat-2167	301	5	strong	strong	ADJ
easat-2167	301	6	backbone	backbone	NOUN
easat-2167	301	7	and	and	CCONJ
easat-2167	301	8	optimized	optimize	VERB
easat-2167	301	9	for	for	ADP
easat-2167	301	10	accurate	accurate	ADJ
easat-2167	301	11	segmentation	segmentation	NOUN
easat-2167	301	12	.	.	PUNCT
easat-2167	302	1	•	•	NUM
easat-2167	302	2	u	u	NOUN
easat-2167	302	3	-	-	NOUN
easat-2167	302	4	net	net	ADJ
easat-2167	302	5	with	with	ADP
easat-2167	302	6	efficientnetb7	efficientnetb7	NOUN
easat-2167	302	7	and	and	CCONJ
easat-2167	302	8	attention	attention	NOUN
easat-2167	302	9	gates	gate	NOUN
easat-2167	302	10	:	:	PUNCT
easat-2167	302	11	•	•	NUM
easat-2167	302	12	encoder	encoder	NOUN
easat-2167	302	13	:	:	PUNCT
easat-2167	302	14	efficientnetb7	efficientnetb7	ADV
easat-2167	302	15	,	,	PUNCT
easat-2167	302	16	the	the	DET
easat-2167	302	17	most	most	ADV
easat-2167	302	18	powerful	powerful	ADJ
easat-2167	302	19	of	of	ADP
easat-2167	302	20	the	the	DET
easat-2167	302	21	three	three	NUM
easat-2167	302	22	.	.	PUNCT
easat-2167	302	23	•	•	NOUN
easat-2167	302	24	decoder	decoder	NOUN
easat-2167	302	25	:	:	PUNCT
easat-2167	302	26	u	u	ADJ
easat-2167	302	27	-	-	ADJ
easat-2167	302	28	net	net	ADJ
easat-2167	302	29	-	-	PUNCT
easat-2167	302	30	style	style	NOUN
easat-2167	302	31	upsampling	upsampling	NOUN
easat-2167	302	32	with	with	ADP
easat-2167	302	33	attention	attention	NOUN
easat-2167	302	34	gates	gate	NOUN
easat-2167	302	35	.	.	PUNCT
easat-2167	303	1	•	•	NUM
easat-2167	303	2	complexity	complexity	NOUN
easat-2167	303	3	:	:	PUNCT
easat-2167	303	4	highest	high	ADJ
easat-2167	303	5	,	,	PUNCT
easat-2167	303	6	with	with	ADP
easat-2167	303	7	attention	attention	NOUN
easat-2167	303	8	mechanisms	mechanism	NOUN
easat-2167	303	9	to	to	PART
easat-2167	303	10	enhance	enhance	VERB
easat-2167	303	11	feature	feature	NOUN
easat-2167	303	12	focus	focus	NOUN
easat-2167	303	13	.	.	PUNCT
easat-2167	304	1	•	•	NUM
easat-2167	304	2	key	key	ADJ
easat-2167	304	3	features	feature	NOUN
easat-2167	304	4	:	:	PUNCT
easat-2167	304	5	high	high	ADJ
easat-2167	304	6	accuracy	accuracy	NOUN
easat-2167	304	7	due	due	ADP
easat-2167	304	8	to	to	ADP
easat-2167	304	9	attention	attention	NOUN
easat-2167	304	10	gates	gate	NOUN
easat-2167	304	11	,	,	PUNCT
easat-2167	304	12	but	but	CCONJ
easat-2167	304	13	at	at	ADP
easat-2167	304	14	the	the	DET
easat-2167	304	15	cost	cost	NOUN
easat-2167	304	16	of	of	ADP
easat-2167	304	17	increased	increase	VERB
easat-2167	304	18	computational	computational	ADJ
easat-2167	304	19	requirements	requirement	NOUN
easat-2167	304	20	.	.	PUNCT
easat-2167	305	1	9.2	9.2	NUM
easat-2167	305	2	.	.	PUNCT
easat-2167	305	3	efficiency	efficiency	NOUN
easat-2167	305	4	(	(	PUNCT
easat-2167	305	5	computational	computational	ADJ
easat-2167	305	6	cost	cost	NOUN
easat-2167	305	7	and	and	CCONJ
easat-2167	305	8	speed	speed	NOUN
easat-2167	305	9	)	)	PUNCT
easat-2167	305	10	•	•	PRON
easat-2167	305	11	u	u	NOUN
easat-2167	305	12	-	-	NOUN
easat-2167	305	13	net	net	ADJ
easat-2167	305	14	with	with	ADP
easat-2167	305	15	mobilenetv2	mobilenetv2	PROPN
easat-2167	305	16	:	:	PUNCT
easat-2167	305	17	•	•	NUM
easat-2167	305	18	efficiency	efficiency	NOUN
easat-2167	305	19	:	:	PUNCT
easat-2167	305	20	highest	high	ADJ
easat-2167	305	21	due	due	ADP
easat-2167	305	22	to	to	ADP
easat-2167	305	23	the	the	DET
easat-2167	305	24	lightweight	lightweight	ADJ
easat-2167	305	25	backbone	backbone	NOUN
easat-2167	305	26	,	,	PUNCT
easat-2167	305	27	ideal	ideal	NOUN
easat-2167	305	28	for	for	ADP
easat-2167	305	29	real	real	ADJ
easat-2167	305	30	-	-	PUNCT
easat-2167	305	31	time	time	NOUN
easat-2167	305	32	and	and	CCONJ
easat-2167	305	33	resourceconstrained	resourceconstraine	VERB
easat-2167	305	34	applications	application	NOUN
easat-2167	305	35	.	.	PUNCT
easat-2167	306	1	•	•	NUM
easat-2167	306	2	computational	computational	ADJ
easat-2167	306	3	cost	cost	NOUN
easat-2167	306	4	:	:	PUNCT
easat-2167	306	5	low	low	ADJ
easat-2167	306	6	,	,	PUNCT
easat-2167	306	7	making	make	VERB
easat-2167	306	8	it	it	PRON
easat-2167	306	9	suitable	suitable	ADJ
easat-2167	306	10	for	for	ADP
easat-2167	306	11	scenarios	scenario	NOUN
easat-2167	306	12	where	where	SCONJ
easat-2167	306	13	computational	computational	ADJ
easat-2167	306	14	resources	resource	NOUN
easat-2167	306	15	are	be	AUX
easat-2167	306	16	limited	limited	ADJ
easat-2167	306	17	.	.	PUNCT
easat-2167	307	1	•	•	NUM
easat-2167	308	1	deeplabv3	deeplabv3	ADV
easat-2167	308	2	+	+	X
easat-2167	308	3	with	with	ADP
easat-2167	308	4	efficientnetb4	efficientnetb4	PROPN
easat-2167	308	5	:	:	PUNCT
easat-2167	308	6	•	•	NUM
easat-2167	309	1	efficiency	efficiency	NOUN
easat-2167	309	2	:	:	PUNCT
easat-2167	309	3	moderate	moderate	ADJ
easat-2167	309	4	,	,	PUNCT
easat-2167	309	5	balancing	balance	VERB
easat-2167	309	6	performance	performance	NOUN
easat-2167	309	7	with	with	ADP
easat-2167	309	8	computational	computational	ADJ
easat-2167	309	9	cost	cost	NOUN
easat-2167	309	10	.	.	PUNCT
easat-2167	310	1	•	•	NUM
easat-2167	310	2	computational	computational	ADJ
easat-2167	310	3	cost	cost	NOUN
easat-2167	310	4	:	:	PUNCT
easat-2167	310	5	moderate	moderate	ADJ
easat-2167	310	6	,	,	PUNCT
easat-2167	310	7	making	make	VERB
easat-2167	310	8	it	it	PRON
easat-2167	310	9	feasible	feasible	ADJ
easat-2167	310	10	for	for	ADP
easat-2167	310	11	more	more	ADV
easat-2167	310	12	demanding	demanding	ADJ
easat-2167	310	13	applications	application	NOUN
easat-2167	310	14	.	.	PUNCT
easat-2167	311	1	•	•	NUM
easat-2167	311	2	u	u	NOUN
easat-2167	311	3	-	-	NOUN
easat-2167	311	4	net	net	ADJ
easat-2167	311	5	with	with	ADP
easat-2167	311	6	efficientnetb7	efficientnetb7	NOUN
easat-2167	311	7	and	and	CCONJ
easat-2167	311	8	attention	attention	NOUN
easat-2167	311	9	gates	gate	NOUN
easat-2167	311	10	:	:	PUNCT
easat-2167	311	11	•	•	NUM
easat-2167	311	12	efficiency	efficiency	NOUN
easat-2167	311	13	:	:	PUNCT
easat-2167	311	14	lowest	low	ADJ
easat-2167	311	15	due	due	ADP
easat-2167	311	16	to	to	ADP
easat-2167	311	17	the	the	DET
easat-2167	311	18	complexity	complexity	NOUN
easat-2167	311	19	of	of	ADP
easat-2167	311	20	the	the	DET
easat-2167	311	21	model	model	NOUN
easat-2167	311	22	.	.	PUNCT
easat-2167	312	1	•	•	ADJ
easat-2167	312	2	computational	computational	ADJ
easat-2167	312	3	cost	cost	NOUN
easat-2167	312	4	:	:	PUNCT
easat-2167	312	5	high	high	ADJ
easat-2167	312	6	,	,	PUNCT
easat-2167	312	7	making	make	VERB
easat-2167	312	8	it	it	PRON
easat-2167	312	9	less	less	ADV
easat-2167	312	10	ideal	ideal	ADJ
easat-2167	312	11	for	for	ADP
easat-2167	312	12	real	real	ADJ
easat-2167	312	13	-	-	PUNCT
easat-2167	312	14	time	time	NOUN
easat-2167	312	15	applications	application	NOUN
easat-2167	312	16	unless	unless	SCONJ
easat-2167	312	17	on	on	ADP
easat-2167	312	18	highperformance	highperformance	NOUN
easat-2167	312	19	hardware	hardware	NOUN
easat-2167	312	20	.	.	PUNCT
easat-2167	313	1	9.3	9.3	NUM
easat-2167	313	2	.	.	PUNCT
easat-2167	313	3	prediction	prediction	NOUN
easat-2167	313	4	accuracy	accuracy	NOUN
easat-2167	313	5	•	•	PRON
easat-2167	313	6	u	u	NOUN
easat-2167	313	7	-	-	NOUN
easat-2167	313	8	net	net	ADJ
easat-2167	313	9	with	with	ADP
easat-2167	313	10	mobilenetv2	mobilenetv2	PROPN
easat-2167	313	11	:	:	PUNCT
easat-2167	313	12	•	•	NUM
easat-2167	313	13	accuracy	accuracy	NOUN
easat-2167	313	14	:	:	PUNCT
easat-2167	313	15	good	good	ADJ
easat-2167	313	16	,	,	PUNCT
easat-2167	313	17	particularly	particularly	ADV
easat-2167	313	18	in	in	ADP
easat-2167	313	19	simpler	simple	ADJ
easat-2167	313	20	tasks	task	NOUN
easat-2167	313	21	or	or	CCONJ
easat-2167	313	22	where	where	SCONJ
easat-2167	313	23	efficiency	efficiency	NOUN
easat-2167	313	24	is	be	AUX
easat-2167	313	25	prioritized	prioritize	VERB
easat-2167	313	26	.	.	PUNCT
easat-2167	314	1	812	812	NUM
easat-2167	314	2	edelweiss	edelweiss	PROPN
easat-2167	314	3	applied	apply	VERB
easat-2167	314	4	science	science	NOUN
easat-2167	314	5	and	and	CCONJ
easat-2167	314	6	technology	technology	NOUN
easat-2167	314	7	issn	issn	PROPN
easat-2167	314	8	:	:	PUNCT
easat-2167	314	9	2576	2576	NUM
easat-2167	314	10	-	-	SYM
easat-2167	314	11	8484	8484	NUM
easat-2167	314	12	vol	vol	NOUN
easat-2167	314	13	.	.	PROPN
easat-2167	314	14	8	8	NUM
easat-2167	314	15	,	,	PUNCT
easat-2167	314	16	no	no	INTJ
easat-2167	314	17	.	.	NOUN
easat-2167	315	1	6	6	NUM
easat-2167	315	2	:	:	PUNCT
easat-2167	315	3	793	793	NUM
easat-2167	315	4	-	-	NUM
easat-2167	315	5	813	813	NUM
easat-2167	315	6	,	,	PUNCT
easat-2167	315	7	2024	2024	NUM
easat-2167	315	8	doi	doi	NOUN
easat-2167	315	9	:	:	PUNCT
easat-2167	315	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	315	11	©	©	NOUN
easat-2167	315	12	2024	2024	NUM
easat-2167	315	13	by	by	ADP
easat-2167	315	14	the	the	DET
easat-2167	315	15	authors	author	NOUN
easat-2167	315	16	;	;	PUNCT
easat-2167	315	17	licensee	licensee	PROPN
easat-2167	315	18	learning	learning	NOUN
easat-2167	315	19	gate	gate	VERB
easat-2167	315	20	•	•	NOUN
easat-2167	315	21	validation	validation	NOUN
easat-2167	315	22	accuracy	accuracy	NOUN
easat-2167	315	23	:	:	PUNCT
easat-2167	315	24	0.7595	0.7595	NUM
easat-2167	315	25	•	•	NOUN
easat-2167	315	26	deeplabv3	deeplabv3	ADV
easat-2167	315	27	+	+	X
easat-2167	315	28	with	with	ADP
easat-2167	315	29	efficientnetb4	efficientnetb4	PROPN
easat-2167	315	30	:	:	PUNCT
easat-2167	315	31	•	•	NUM
easat-2167	315	32	accuracy	accuracy	NOUN
easat-2167	315	33	:	:	PUNCT
easat-2167	315	34	higher	high	ADJ
easat-2167	315	35	accuracy	accuracy	NOUN
easat-2167	315	36	than	than	ADP
easat-2167	315	37	mobilenetv2	mobilenetv2	PROPN
easat-2167	315	38	,	,	PUNCT
easat-2167	315	39	suitable	suitable	ADJ
easat-2167	315	40	for	for	ADP
easat-2167	315	41	more	more	ADJ
easat-2167	315	42	complex	complex	ADJ
easat-2167	315	43	tasks	task	NOUN
easat-2167	315	44	.	.	PUNCT
easat-2167	316	1	•	•	NUM
easat-2167	316	2	validation	validation	NOUN
easat-2167	316	3	accuracy	accuracy	NOUN
easat-2167	316	4	:	:	PUNCT
easat-2167	316	5	0.7879	0.7879	NUM
easat-2167	316	6	•	•	NUM
easat-2167	316	7	u	u	NOUN
easat-2167	316	8	-	-	NOUN
easat-2167	316	9	net	net	ADJ
easat-2167	316	10	with	with	ADP
easat-2167	316	11	efficientnetb7	efficientnetb7	NOUN
easat-2167	316	12	and	and	CCONJ
easat-2167	316	13	attention	attention	NOUN
easat-2167	316	14	gates	gate	NOUN
easat-2167	316	15	:	:	PUNCT
easat-2167	316	16	•	•	NUM
easat-2167	316	17	accuracy	accuracy	NOUN
easat-2167	316	18	:	:	PUNCT
easat-2167	316	19	highest	high	ADJ
easat-2167	316	20	accuracy	accuracy	NOUN
easat-2167	316	21	due	due	ADP
easat-2167	316	22	to	to	ADP
easat-2167	316	23	attention	attention	NOUN
easat-2167	316	24	gates	gate	NOUN
easat-2167	316	25	and	and	CCONJ
easat-2167	316	26	advanced	advanced	ADJ
easat-2167	316	27	feature	feature	NOUN
easat-2167	316	28	extraction	extraction	NOUN
easat-2167	316	29	.	.	PUNCT
easat-2167	317	1	•	•	NUM
easat-2167	317	2	validation	validation	NOUN
easat-2167	317	3	accuracy	accuracy	NOUN
easat-2167	317	4	:	:	PUNCT
easat-2167	317	5	0.8028	0.8028	NUM
easat-2167	317	6	4	4	NUM
easat-2167	317	7	.	.	PUNCT
easat-2167	317	8	segmentation	segmentation	NOUN
easat-2167	317	9	performance	performance	NOUN
easat-2167	317	10	•	•	NOUN
easat-2167	317	11	u	u	NOUN
easat-2167	317	12	-	-	NOUN
easat-2167	317	13	net	net	ADJ
easat-2167	317	14	with	with	ADP
easat-2167	317	15	mobilenetv2	mobilenetv2	PROPN
easat-2167	317	16	:	:	PUNCT
easat-2167	317	17	•	•	NUM
easat-2167	317	18	segmentation	segmentation	NOUN
easat-2167	317	19	quality	quality	NOUN
easat-2167	317	20	:	:	PUNCT
easat-2167	317	21	good	good	ADJ
easat-2167	317	22	,	,	PUNCT
easat-2167	317	23	but	but	CCONJ
easat-2167	317	24	may	may	AUX
easat-2167	317	25	struggle	struggle	VERB
easat-2167	317	26	with	with	ADP
easat-2167	317	27	fine	fine	ADJ
easat-2167	317	28	details	detail	NOUN
easat-2167	317	29	in	in	ADP
easat-2167	317	30	complex	complex	ADJ
easat-2167	317	31	images	image	NOUN
easat-2167	317	32	.	.	PUNCT
easat-2167	318	1	•	•	NUM
easat-2167	318	2	validation	validation	NOUN
easat-2167	318	3	loss	loss	NOUN
easat-2167	318	4	:	:	PUNCT
easat-2167	318	5	0.2052	0.2052	NUM
easat-2167	318	6	•	•	NOUN
easat-2167	318	7	deeplabv3	deeplabv3	ADJ
easat-2167	318	8	+	+	X
easat-2167	318	9	with	with	ADP
easat-2167	318	10	efficientnetb4	efficientnetb4	PROPN
easat-2167	318	11	:	:	PUNCT
easat-2167	318	12	•	•	NUM
easat-2167	318	13	segmentation	segmentation	NOUN
easat-2167	318	14	quality	quality	NOUN
easat-2167	318	15	:	:	PUNCT
easat-2167	318	16	superior	superior	ADJ
easat-2167	318	17	to	to	ADP
easat-2167	318	18	mobilenetv2	mobilenetv2	PROPN
easat-2167	318	19	,	,	PUNCT
easat-2167	318	20	with	with	ADP
easat-2167	318	21	better	well	ADJ
easat-2167	318	22	handling	handling	NOUN
easat-2167	318	23	of	of	ADP
easat-2167	318	24	complex	complex	ADJ
easat-2167	318	25	boundaries	boundary	NOUN
easat-2167	318	26	.	.	PUNCT
easat-2167	319	1	•	•	NUM
easat-2167	319	2	validation	validation	NOUN
easat-2167	319	3	loss	loss	NOUN
easat-2167	319	4	:	:	PUNCT
easat-2167	319	5	0.2342	0.2342	NUM
easat-2167	319	6	•	•	NUM
easat-2167	319	7	u	u	NOUN
easat-2167	319	8	-	-	NOUN
easat-2167	319	9	net	net	ADJ
easat-2167	319	10	with	with	ADP
easat-2167	319	11	efficientnetb7	efficientnetb7	NOUN
easat-2167	319	12	and	and	CCONJ
easat-2167	319	13	attention	attention	NOUN
easat-2167	319	14	gates	gate	NOUN
easat-2167	319	15	:	:	PUNCT
easat-2167	319	16	•	•	NUM
easat-2167	319	17	segmentation	segmentation	NOUN
easat-2167	319	18	quality	quality	NOUN
easat-2167	319	19	:	:	PUNCT
easat-2167	319	20	best	good	ADJ
easat-2167	319	21	-	-	PUNCT
easat-2167	319	22	in	in	ADP
easat-2167	319	23	-	-	PUNCT
easat-2167	319	24	class	class	NOUN
easat-2167	319	25	,	,	PUNCT
easat-2167	319	26	excelling	excel	VERB
easat-2167	319	27	in	in	ADP
easat-2167	319	28	distinguishing	distinguish	VERB
easat-2167	319	29	subtle	subtle	ADJ
easat-2167	319	30	details	detail	NOUN
easat-2167	319	31	.	.	PUNCT
easat-2167	320	1	•	•	NUM
easat-2167	320	2	validation	validation	NOUN
easat-2167	320	3	loss	loss	NOUN
easat-2167	320	4	:	:	PUNCT
easat-2167	320	5	0.1720	0.1720	NUM
easat-2167	320	6	9.4	9.4	NUM
easat-2167	320	7	.	.	PUNCT
easat-2167	321	1	comparison	comparison	NOUN
easat-2167	321	2	with	with	ADP
easat-2167	321	3	earlier	early	ADJ
easat-2167	321	4	work	work	NOUN
easat-2167	321	5	the	the	DET
easat-2167	321	6	three	three	NUM
easat-2167	321	7	models	model	NOUN
easat-2167	321	8	discussed	discuss	VERB
easat-2167	321	9	represent	represent	VERB
easat-2167	321	10	significant	significant	ADJ
easat-2167	321	11	advancements	advancement	NOUN
easat-2167	321	12	over	over	ADP
easat-2167	321	13	earlier	early	ADJ
easat-2167	321	14	image	image	NOUN
easat-2167	321	15	segmentation	segmentation	NOUN
easat-2167	321	16	models	model	NOUN
easat-2167	321	17	that	that	PRON
easat-2167	321	18	typically	typically	ADV
easat-2167	321	19	relied	rely	VERB
easat-2167	321	20	on	on	ADP
easat-2167	321	21	simpler	simple	ADJ
easat-2167	321	22	backbones	backbone	NOUN
easat-2167	321	23	like	like	ADP
easat-2167	321	24	vgg	vgg	NOUN
easat-2167	321	25	or	or	CCONJ
easat-2167	321	26	resnet	resnet	NOUN
easat-2167	321	27	.	.	PUNCT
easat-2167	322	1	these	these	DET
easat-2167	322	2	earlier	early	ADJ
easat-2167	322	3	models	model	NOUN
easat-2167	322	4	often	often	ADV
easat-2167	322	5	required	require	VERB
easat-2167	322	6	a	a	DET
easat-2167	322	7	trade	trade	NOUN
easat-2167	322	8	-	-	PUNCT
easat-2167	322	9	off	off	NOUN
easat-2167	322	10	between	between	ADP
easat-2167	322	11	accuracy	accuracy	NOUN
easat-2167	322	12	and	and	CCONJ
easat-2167	322	13	computational	computational	ADJ
easat-2167	322	14	efficiency	efficiency	NOUN
easat-2167	322	15	.	.	PUNCT
easat-2167	323	1	•	•	NUM
easat-2167	323	2	advancements	advancement	NOUN
easat-2167	323	3	:	:	PUNCT
easat-2167	323	4	o	o	NOUN
easat-2167	323	5	efficientnet	efficientnet	NOUN
easat-2167	323	6	integration	integration	NOUN
easat-2167	323	7	:	:	PUNCT
easat-2167	323	8	offers	offer	VERB
easat-2167	323	9	a	a	DET
easat-2167	323	10	strong	strong	ADJ
easat-2167	323	11	balance	balance	NOUN
easat-2167	323	12	between	between	ADP
easat-2167	323	13	accuracy	accuracy	NOUN
easat-2167	323	14	and	and	CCONJ
easat-2167	323	15	resource	resource	NOUN
easat-2167	323	16	use	use	NOUN
easat-2167	323	17	.	.	PUNCT
easat-2167	324	1	o	o	NOUN
easat-2167	324	2	attention	attention	NOUN
easat-2167	324	3	mechanisms	mechanism	NOUN
easat-2167	324	4	:	:	PUNCT
easat-2167	324	5	in	in	ADP
easat-2167	324	6	u	u	NOUN
easat-2167	324	7	-	-	NOUN
easat-2167	324	8	net	net	ADJ
easat-2167	324	9	with	with	ADP
easat-2167	324	10	efficientnetb7	efficientnetb7	NOUN
easat-2167	324	11	,	,	PUNCT
easat-2167	324	12	attention	attention	NOUN
easat-2167	324	13	gates	gate	NOUN
easat-2167	324	14	enhance	enhance	VERB
easat-2167	324	15	segmentation	segmentation	NOUN
easat-2167	324	16	quality	quality	NOUN
easat-2167	324	17	by	by	ADP
easat-2167	324	18	focusing	focus	VERB
easat-2167	324	19	on	on	ADP
easat-2167	324	20	relevant	relevant	ADJ
easat-2167	324	21	features	feature	NOUN
easat-2167	324	22	.	.	PUNCT
easat-2167	325	1	10	10	NUM
easat-2167	325	2	.	.	X
easat-2167	325	3	conclusion	conclusion	NOUN
easat-2167	325	4	and	and	CCONJ
easat-2167	325	5	recommendations	recommendation	NOUN
easat-2167	325	6	•	•	ADV
easat-2167	325	7	best	good	ADJ
easat-2167	325	8	for	for	ADP
easat-2167	325	9	efficiency	efficiency	NOUN
easat-2167	325	10	:	:	PUNCT
easat-2167	325	11	u	u	NOUN
easat-2167	325	12	-	-	NOUN
easat-2167	325	13	net	net	ADJ
easat-2167	325	14	with	with	ADP
easat-2167	325	15	mobilenetv2	mobilenetv2	PROPN
easat-2167	325	16	is	be	AUX
easat-2167	325	17	recommended	recommend	VERB
easat-2167	325	18	for	for	ADP
easat-2167	325	19	scenarios	scenario	NOUN
easat-2167	325	20	where	where	SCONJ
easat-2167	325	21	efficiency	efficiency	NOUN
easat-2167	325	22	and	and	CCONJ
easat-2167	325	23	computational	computational	ADJ
easat-2167	325	24	cost	cost	NOUN
easat-2167	325	25	are	be	AUX
easat-2167	325	26	paramount	paramount	ADJ
easat-2167	325	27	,	,	PUNCT
easat-2167	325	28	such	such	ADJ
easat-2167	325	29	as	as	ADP
easat-2167	325	30	mobile	mobile	ADJ
easat-2167	325	31	applications	application	NOUN
easat-2167	325	32	.	.	PUNCT
easat-2167	326	1	•	•	NUM
easat-2167	326	2	best	good	ADJ
easat-2167	326	3	for	for	ADP
easat-2167	326	4	balance	balance	NOUN
easat-2167	326	5	:	:	PUNCT
easat-2167	327	1	deeplabv3	deeplabv3	ADJ
easat-2167	327	2	+	+	X
easat-2167	327	3	with	with	ADP
easat-2167	327	4	efficientnetb4	efficientnetb4	PROPN
easat-2167	327	5	provides	provide	VERB
easat-2167	327	6	a	a	DET
easat-2167	327	7	good	good	ADJ
easat-2167	327	8	compromise	compromise	NOUN
easat-2167	327	9	between	between	ADP
easat-2167	327	10	accuracy	accuracy	NOUN
easat-2167	327	11	and	and	CCONJ
easat-2167	327	12	efficiency	efficiency	NOUN
easat-2167	327	13	,	,	PUNCT
easat-2167	327	14	making	make	VERB
easat-2167	327	15	it	it	PRON
easat-2167	327	16	suitable	suitable	ADJ
easat-2167	327	17	for	for	ADP
easat-2167	327	18	general	general	ADJ
easat-2167	327	19	-	-	PUNCT
easat-2167	327	20	purpose	purpose	NOUN
easat-2167	327	21	applications	application	NOUN
easat-2167	327	22	.	.	PUNCT
easat-2167	328	1	•	•	NUM
easat-2167	328	2	best	good	ADJ
easat-2167	328	3	for	for	ADP
easat-2167	328	4	accuracy	accuracy	NOUN
easat-2167	328	5	:	:	PUNCT
easat-2167	328	6	u	u	NOUN
easat-2167	328	7	-	-	NOUN
easat-2167	328	8	net	net	ADJ
easat-2167	328	9	with	with	ADP
easat-2167	328	10	efficientnetb7	efficientnetb7	NOUN
easat-2167	328	11	and	and	CCONJ
easat-2167	328	12	attention	attention	NOUN
easat-2167	328	13	gates	gate	NOUN
easat-2167	328	14	is	be	AUX
easat-2167	328	15	ideal	ideal	ADJ
easat-2167	328	16	for	for	ADP
easat-2167	328	17	high	high	ADJ
easat-2167	328	18	-	-	PUNCT
easat-2167	328	19	stakes	stake	NOUN
easat-2167	328	20	,	,	PUNCT
easat-2167	328	21	precision	precision	NOUN
easat-2167	328	22	-	-	PUNCT
easat-2167	328	23	critical	critical	ADJ
easat-2167	328	24	tasks	task	NOUN
easat-2167	328	25	,	,	PUNCT
easat-2167	328	26	although	although	SCONJ
easat-2167	328	27	it	it	PRON
easat-2167	328	28	demands	demand	VERB
easat-2167	328	29	more	more	ADJ
easat-2167	328	30	computational	computational	ADJ
easat-2167	328	31	resources	resource	NOUN
easat-2167	328	32	.	.	PUNCT
easat-2167	329	1	overall	overall	ADV
easat-2167	329	2	,	,	PUNCT
easat-2167	329	3	these	these	DET
easat-2167	329	4	models	model	NOUN
easat-2167	329	5	demonstrate	demonstrate	VERB
easat-2167	329	6	impressive	impressive	ADJ
easat-2167	329	7	capabilities	capability	NOUN
easat-2167	329	8	in	in	ADP
easat-2167	329	9	modern	modern	ADJ
easat-2167	329	10	image	image	NOUN
easat-2167	329	11	segmentation	segmentation	NOUN
easat-2167	329	12	tasks	task	NOUN
easat-2167	329	13	,	,	PUNCT
easat-2167	329	14	each	each	DET
easat-2167	329	15	excelling	excel	VERB
easat-2167	329	16	in	in	ADP
easat-2167	329	17	different	different	ADJ
easat-2167	329	18	aspects	aspect	NOUN
easat-2167	329	19	depending	depend	VERB
easat-2167	329	20	on	on	ADP
easat-2167	329	21	the	the	DET
easat-2167	329	22	specific	specific	ADJ
easat-2167	329	23	requirements	requirement	NOUN
easat-2167	329	24	of	of	ADP
easat-2167	329	25	the	the	DET
easat-2167	329	26	application	application	NOUN
easat-2167	329	27	.	.	PUNCT
easat-2167	330	1	the	the	DET
easat-2167	330	2	choice	choice	NOUN
easat-2167	330	3	of	of	ADP
easat-2167	330	4	model	model	NOUN
easat-2167	330	5	should	should	AUX
easat-2167	330	6	align	align	VERB
easat-2167	330	7	with	with	ADP
easat-2167	330	8	the	the	DET
easat-2167	330	9	balance	balance	NOUN
easat-2167	330	10	of	of	ADP
easat-2167	330	11	accuracy	accuracy	NOUN
easat-2167	330	12	,	,	PUNCT
easat-2167	330	13	efficiency	efficiency	NOUN
easat-2167	330	14	,	,	PUNCT
easat-2167	330	15	and	and	CCONJ
easat-2167	330	16	computational	computational	ADJ
easat-2167	330	17	resources	resource	NOUN
easat-2167	330	18	available	available	ADJ
easat-2167	330	19	.	.	PUNCT
easat-2167	331	1	challenges	challenge	NOUN
easat-2167	331	2	and	and	CCONJ
easat-2167	331	3	future	future	ADJ
easat-2167	331	4	directions	direction	NOUN
easat-2167	331	5	:	:	PUNCT
easat-2167	331	6	while	while	SCONJ
easat-2167	331	7	significant	significant	ADJ
easat-2167	331	8	progress	progress	NOUN
easat-2167	331	9	has	have	AUX
easat-2167	331	10	been	be	AUX
easat-2167	331	11	made	make	VERB
easat-2167	331	12	in	in	ADP
easat-2167	331	13	the	the	DET
easat-2167	331	14	field	field	NOUN
easat-2167	331	15	of	of	ADP
easat-2167	331	16	image	image	NOUN
easat-2167	331	17	segmentation	segmentation	NOUN
easat-2167	331	18	for	for	ADP
easat-2167	331	19	leaf	leaf	NOUN
easat-2167	331	20	diseases	disease	NOUN
easat-2167	331	21	,	,	PUNCT
easat-2167	331	22	challenges	challenge	NOUN
easat-2167	331	23	remain	remain	VERB
easat-2167	331	24	.	.	PUNCT
easat-2167	332	1	the	the	DET
easat-2167	332	2	variability	variability	NOUN
easat-2167	332	3	in	in	ADP
easat-2167	332	4	environmental	environmental	ADJ
easat-2167	332	5	conditions	condition	NOUN
easat-2167	332	6	,	,	PUNCT
easat-2167	332	7	the	the	DET
easat-2167	332	8	presence	presence	NOUN
easat-2167	332	9	of	of	ADP
easat-2167	332	10	noise	noise	NOUN
easat-2167	332	11	in	in	ADP
easat-2167	332	12	images	image	NOUN
easat-2167	332	13	,	,	PUNCT
easat-2167	332	14	and	and	CCONJ
easat-2167	332	15	the	the	DET
easat-2167	332	16	need	need	NOUN
easat-2167	332	17	for	for	ADP
easat-2167	332	18	large	large	ADJ
easat-2167	332	19	,	,	PUNCT
easat-2167	332	20	annotated	annotate	VERB
easat-2167	332	21	datasets	dataset	NOUN
easat-2167	332	22	for	for	ADP
easat-2167	332	23	training	train	VERB
easat-2167	332	24	deep	deep	ADJ
easat-2167	332	25	learning	learning	NOUN
easat-2167	332	26	models	model	NOUN
easat-2167	332	27	are	be	AUX
easat-2167	332	28	ongoing	ongoing	ADJ
easat-2167	332	29	concerns	concern	NOUN
easat-2167	332	30	.	.	PUNCT
easat-2167	333	1	future	future	ADJ
easat-2167	333	2	research	research	NOUN
easat-2167	333	3	is	be	AUX
easat-2167	333	4	likely	likely	ADJ
easat-2167	333	5	to	to	PART
easat-2167	333	6	focus	focus	VERB
easat-2167	333	7	on	on	ADP
easat-2167	333	8	developing	develop	VERB
easat-2167	333	9	more	more	ADJ
easat-2167	333	10	generalized	generalized	ADJ
easat-2167	333	11	models	model	NOUN
easat-2167	333	12	that	that	PRON
easat-2167	333	13	can	can	AUX
easat-2167	333	14	perform	perform	VERB
easat-2167	333	15	well	well	ADV
easat-2167	333	16	across	across	ADP
easat-2167	333	17	different	different	ADJ
easat-2167	333	18	conditions	condition	NOUN
easat-2167	333	19	,	,	PUNCT
easat-2167	333	20	improving	improve	VERB
easat-2167	333	21	the	the	DET
easat-2167	333	22	computational	computational	ADJ
easat-2167	333	23	efficiency	efficiency	NOUN
easat-2167	333	24	of	of	ADP
easat-2167	333	25	these	these	DET
easat-2167	333	26	models	model	NOUN
easat-2167	333	27	,	,	PUNCT
easat-2167	333	28	and	and	CCONJ
easat-2167	333	29	integrating	integrate	VERB
easat-2167	333	30	segmentation	segmentation	NOUN
easat-2167	333	31	techniques	technique	NOUN
easat-2167	333	32	with	with	ADP
easat-2167	333	33	other	other	ADJ
easat-2167	333	34	data	datum	NOUN
easat-2167	333	35	sources	source	NOUN
easat-2167	333	36	,	,	PUNCT
easat-2167	333	37	such	such	ADJ
easat-2167	333	38	as	as	ADP
easat-2167	333	39	meteorological	meteorological	ADJ
easat-2167	333	40	data	datum	NOUN
easat-2167	333	41	,	,	PUNCT
easat-2167	333	42	to	to	PART
easat-2167	333	43	enhance	enhance	VERB
easat-2167	333	44	disease	disease	NOUN
easat-2167	333	45	prediction	prediction	NOUN
easat-2167	333	46	and	and	CCONJ
easat-2167	333	47	management	management	NOUN
easat-2167	333	48	.	.	PUNCT
easat-2167	334	1	813	813	NUM
easat-2167	334	2	edelweiss	edelweiss	PROPN
easat-2167	334	3	applied	apply	VERB
easat-2167	334	4	science	science	NOUN
easat-2167	334	5	and	and	CCONJ
easat-2167	334	6	technology	technology	NOUN
easat-2167	334	7	issn	issn	PROPN
easat-2167	334	8	:	:	PUNCT
easat-2167	334	9	2576	2576	NUM
easat-2167	334	10	-	-	SYM
easat-2167	334	11	8484	8484	NUM
easat-2167	334	12	vol	vol	NOUN
easat-2167	334	13	.	.	PROPN
easat-2167	334	14	8	8	NUM
easat-2167	334	15	,	,	PUNCT
easat-2167	334	16	no	no	INTJ
easat-2167	334	17	.	.	NOUN
easat-2167	335	1	6	6	NUM
easat-2167	335	2	:	:	PUNCT
easat-2167	335	3	793	793	NUM
easat-2167	335	4	-	-	NUM
easat-2167	335	5	813	813	NUM
easat-2167	335	6	,	,	PUNCT
easat-2167	335	7	2024	2024	NUM
easat-2167	335	8	doi	doi	NOUN
easat-2167	335	9	:	:	PUNCT
easat-2167	335	10	10.55214/25768484.v8i6.2167	10.55214/25768484.v8i6.2167	ADJ
easat-2167	335	11	©	©	NOUN
easat-2167	335	12	2024	2024	NUM
easat-2167	335	13	by	by	ADP
easat-2167	335	14	the	the	DET
easat-2167	335	15	authors	author	NOUN
easat-2167	335	16	;	;	PUNCT
easat-2167	336	1	licensee	licensee	PROPN
easat-2167	336	2	learning	learn	VERB
easat-2167	336	3	gate	gate	NOUN
easat-2167	336	4	copyright	copyright	NOUN
easat-2167	336	5	:	:	PUNCT
easat-2167	336	6	©	©	PROPN
easat-2167	336	7	2024	2024	NUM
easat-2167	336	8	by	by	ADP
easat-2167	336	9	the	the	DET
easat-2167	336	10	authors	author	NOUN
easat-2167	336	11	.	.	PUNCT
easat-2167	337	1	this	this	DET
easat-2167	337	2	article	article	NOUN
easat-2167	337	3	is	be	AUX
easat-2167	337	4	an	an	DET
easat-2167	337	5	open	open	ADJ
easat-2167	337	6	access	access	NOUN
easat-2167	337	7	article	article	NOUN
easat-2167	337	8	distributed	distribute	VERB
easat-2167	337	9	under	under	ADP
easat-2167	337	10	the	the	DET
easat-2167	337	11	terms	term	NOUN
easat-2167	337	12	and	and	CCONJ
easat-2167	337	13	conditions	condition	NOUN
easat-2167	337	14	of	of	ADP
easat-2167	337	15	the	the	DET
easat-2167	337	16	creative	creative	ADJ
easat-2167	337	17	commons	common	NOUN
easat-2167	337	18	attribution	attribution	NOUN
easat-2167	337	19	(	(	PUNCT
easat-2167	337	20	cc	cc	NOUN
easat-2167	337	21	by	by	ADP
easat-2167	337	22	)	)	PUNCT
easat-2167	337	23	license	license	NOUN
easat-2167	337	24	(	(	PUNCT
easat-2167	337	25	https://creativecommons.org/licenses/by/4.0/	https://creativecommons.org/licenses/by/4.0/	PROPN
easat-2167	337	26	)	)	PUNCT
easat-2167	337	27	.	.	PUNCT
easat-2167	338	1	references	reference	NOUN
easat-2167	338	2	[	[	X
easat-2167	338	3	1	1	X
easat-2167	338	4	]	]	PUNCT
easat-2167	338	5	t.	t.	PROPN
easat-2167	338	6	archana	archana	PROPN
easat-2167	338	7	and	and	CCONJ
easat-2167	338	8	a.	a.	PROPN
easat-2167	338	9	sahayadhas	sahayadhas	PROPN
easat-2167	338	10	,	,	PUNCT
easat-2167	338	11	"	"	PUNCT
easat-2167	338	12	automatic	automatic	ADJ
easat-2167	338	13	rice	rice	NOUN
easat-2167	338	14	leaf	leaf	NOUN
easat-2167	338	15	disease	disease	NOUN
easat-2167	338	16	segmentation	segmentation	NOUN
easat-2167	338	17	:	:	PUNCT
easat-2167	338	18	detection	detection	NOUN
easat-2167	338	19	and	and	CCONJ
easat-2167	338	20	segmentation	segmentation	NOUN
easat-2167	338	21	of	of	ADP
easat-2167	338	22	bacterial	bacterial	ADJ
easat-2167	338	23	leaf	leaf	NOUN
easat-2167	338	24	blight	blight	NOUN
easat-2167	338	25	and	and	CCONJ
easat-2167	338	26	brown	brown	ADJ
easat-2167	338	27	spot	spot	NOUN
easat-2167	338	28	in	in	ADP
easat-2167	338	29	rice	rice	NOUN
easat-2167	338	30	leaves	leave	VERB
easat-2167	338	31	under	under	ADP
easat-2167	338	32	varying	vary	VERB
easat-2167	338	33	environmental	environmental	ADJ
easat-2167	338	34	conditions	condition	NOUN
easat-2167	338	35	,	,	PUNCT
easat-2167	338	36	"	"	PUNCT
easat-2167	338	37	international	international	ADJ
easat-2167	338	38	journal	journal	NOUN
easat-2167	338	39	of	of	ADP
easat-2167	338	40	image	image	NOUN
easat-2167	338	41	processing	processing	NOUN
easat-2167	338	42	and	and	CCONJ
easat-2167	338	43	vision	vision	NOUN
easat-2167	338	44	sciences	science	NOUN
easat-2167	338	45	,	,	PUNCT
easat-2167	338	46	2018	2018	NUM
easat-2167	338	47	.	.	PUNCT
easat-2167	339	1	[	[	X
easat-2167	339	2	2	2	NUM
easat-2167	339	3	]	]	PUNCT
easat-2167	339	4	p.	p.	NOUN
easat-2167	339	5	k.	k.	PROPN
easat-2167	340	1	sethy	sethy	ADJ
easat-2167	340	2	,	,	PUNCT
easat-2167	340	3	a.	a.	PROPN
easat-2167	340	4	k.	k.	PROPN
easat-2167	340	5	rath	rath	PROPN
easat-2167	340	6	,	,	PUNCT
easat-2167	340	7	and	and	CCONJ
easat-2167	340	8	s.	s.	PROPN
easat-2167	340	9	k.	k.	PROPN
easat-2167	340	10	behera	behera	PROPN
easat-2167	340	11	,	,	PUNCT
easat-2167	340	12	"	"	PUNCT
easat-2167	340	13	a	a	DET
easat-2167	340	14	survey	survey	NOUN
easat-2167	340	15	on	on	ADP
easat-2167	340	16	image	image	NOUN
easat-2167	340	17	processing	processing	NOUN
easat-2167	340	18	techniques	technique	NOUN
easat-2167	340	19	for	for	ADP
easat-2167	340	20	rice	rice	NOUN
easat-2167	340	21	plant	plant	NOUN
easat-2167	340	22	disease	disease	NOUN
easat-2167	340	23	detection	detection	NOUN
easat-2167	340	24	and	and	CCONJ
easat-2167	340	25	classification	classification	NOUN
easat-2167	340	26	,	,	PUNCT
easat-2167	340	27	"	"	PUNCT
easat-2167	340	28	journal	journal	NOUN
easat-2167	340	29	of	of	ADP
easat-2167	340	30	agricultural	agricultural	ADJ
easat-2167	340	31	engineering	engineering	NOUN
easat-2167	340	32	,	,	PUNCT
easat-2167	340	33	2020	2020	NUM
easat-2167	340	34	.	.	PUNCT
easat-2167	341	1	[	[	X
easat-2167	341	2	3	3	X
easat-2167	341	3	]	]	X
easat-2167	341	4	c.	c.	PROPN
easat-2167	341	5	thanawiparat	thanawiparat	PROPN
easat-2167	341	6	,	,	PUNCT
easat-2167	341	7	n.	n.	PROPN
easat-2167	341	8	iam	iam	PROPN
easat-2167	341	9	-	-	PUNCT
easat-2167	341	10	on	on	ADP
easat-2167	341	11	,	,	PUNCT
easat-2167	341	12	and	and	CCONJ
easat-2167	341	13	v.	v.	ADP
easat-2167	341	14	boonjing	boonje	VERB
easat-2167	341	15	,	,	PUNCT
easat-2167	341	16	"	"	PUNCT
easat-2167	341	17	improving	improve	VERB
easat-2167	341	18	rice	rice	NOUN
easat-2167	341	19	disease	disease	NOUN
easat-2167	341	20	detection	detection	NOUN
easat-2167	341	21	using	use	VERB
easat-2167	341	22	deep	deep	ADJ
easat-2167	341	23	learning	learning	NOUN
easat-2167	341	24	and	and	CCONJ
easat-2167	341	25	level	level	VERB
easat-2167	341	26	set	set	VERB
easat-2167	341	27	methods	method	NOUN
easat-2167	341	28	,	,	PUNCT
easat-2167	341	29	"	"	PUNCT
easat-2167	341	30	expert	expert	NOUN
easat-2167	341	31	systems	system	NOUN
easat-2167	341	32	with	with	ADP
easat-2167	341	33	applications	application	NOUN
easat-2167	341	34	,	,	PUNCT
easat-2167	341	35	2023	2023	NUM
easat-2167	341	36	.	.	PUNCT
easat-2167	342	1	[	[	X
easat-2167	342	2	4	4	X
easat-2167	342	3	]	]	PUNCT
easat-2167	342	4	s.	s.	PROPN
easat-2167	342	5	m.	m.	PROPN
easat-2167	342	6	abdullah	abdullah	PROPN
easat-2167	342	7	,	,	PUNCT
easat-2167	342	8	et	et	PROPN
easat-2167	342	9	al	al	PROPN
easat-2167	342	10	.	.	PROPN
easat-2167	342	11	,	,	PUNCT
easat-2167	342	12	"	"	PUNCT
easat-2167	342	13	bi	bi	ADJ
easat-2167	342	14	-	-	ADJ
easat-2167	342	15	level	level	ADJ
easat-2167	342	16	thresholding	thresholding	NOUN
easat-2167	342	17	for	for	ADP
easat-2167	342	18	disease	disease	NOUN
easat-2167	342	19	diagnosis	diagnosis	NOUN
easat-2167	342	20	:	:	PUNCT
easat-2167	342	21	a	a	DET
easat-2167	342	22	portable	portable	ADJ
easat-2167	342	23	tool	tool	NOUN
easat-2167	342	24	for	for	ADP
easat-2167	342	25	predicting	predict	VERB
easat-2167	342	26	rice	rice	NOUN
easat-2167	342	27	diseases	disease	NOUN
easat-2167	342	28	,	,	PUNCT
easat-2167	342	29	"	"	PUNCT
easat-2167	342	30	proceedings	proceeding	NOUN
easat-2167	342	31	of	of	ADP
easat-2167	342	32	the	the	DET
easat-2167	342	33	international	international	ADJ
easat-2167	342	34	conference	conference	NOUN
easat-2167	342	35	on	on	ADP
easat-2167	342	36	computer	computer	NOUN
easat-2167	342	37	vision	vision	NOUN
easat-2167	342	38	and	and	CCONJ
easat-2167	342	39	graphics	graphic	NOUN
easat-2167	342	40	,	,	PUNCT
easat-2167	342	41	2016	2016	NUM
easat-2167	342	42	.	.	PUNCT
easat-2167	343	1	[	[	X
easat-2167	343	2	5	5	X
easat-2167	343	3	]	]	PUNCT
easat-2167	343	4	d.	d.	PROPN
easat-2167	343	5	m.	m.	PROPN
easat-2167	343	6	mai	mai	PROPN
easat-2167	343	7	and	and	CCONJ
easat-2167	343	8	q.	q.	PROPN
easat-2167	343	9	meng	meng	PROPN
easat-2167	343	10	,	,	PUNCT
easat-2167	343	11	"	"	PUNCT
easat-2167	343	12	lesion	lesion	NOUN
easat-2167	343	13	segmentation	segmentation	NOUN
easat-2167	343	14	in	in	ADP
easat-2167	343	15	rice	rice	NOUN
easat-2167	343	16	leaf	leaf	NOUN
easat-2167	343	17	blast	blast	NOUN
easat-2167	343	18	images	image	NOUN
easat-2167	343	19	using	use	VERB
easat-2167	343	20	superpixel	superpixel	ADJ
easat-2167	343	21	segmentation	segmentation	NOUN
easat-2167	343	22	and	and	CCONJ
easat-2167	343	23	random	random	ADJ
easat-2167	343	24	forest	forest	NOUN
easat-2167	343	25	classifiers	classifier	NOUN
easat-2167	343	26	,	,	PUNCT
easat-2167	343	27	"	"	PUNCT
easat-2167	343	28	biosystems	biosystems	PROPN
easat-2167	343	29	engineering	engineering	NOUN
easat-2167	343	30	,	,	PUNCT
easat-2167	343	31	2016	2016	NUM
easat-2167	343	32	.	.	PUNCT
easat-2167	344	1	[	[	X
easat-2167	344	2	6	6	NUM
easat-2167	344	3	]	]	PUNCT
easat-2167	344	4	p.	p.	PROPN
easat-2167	344	5	devi	devi	PROPN
easat-2167	344	6	and	and	CCONJ
easat-2167	344	7	m.	m.	PROPN
easat-2167	344	8	muthukannan	muthukannan	PROPN
easat-2167	344	9	,	,	PUNCT
easat-2167	344	10	"	"	PUNCT
easat-2167	344	11	comparison	comparison	NOUN
easat-2167	344	12	of	of	ADP
easat-2167	344	13	image	image	NOUN
easat-2167	344	14	segmentation	segmentation	NOUN
easat-2167	344	15	methods	method	NOUN
easat-2167	344	16	for	for	ADP
easat-2167	344	17	identifying	identify	VERB
easat-2167	344	18	diseased	diseased	ADJ
easat-2167	344	19	rice	rice	NOUN
easat-2167	344	20	leaves	leave	NOUN
easat-2167	344	21	,	,	PUNCT
easat-2167	344	22	"	"	PUNCT
easat-2167	344	23	international	international	ADJ
easat-2167	344	24	journal	journal	NOUN
easat-2167	344	25	of	of	ADP
easat-2167	344	26	computer	computer	NOUN
easat-2167	344	27	applications	application	NOUN
easat-2167	344	28	,	,	PUNCT
easat-2167	344	29	2014	2014	NUM
easat-2167	344	30	.	.	PUNCT
easat-2167	345	1	[	[	X
easat-2167	345	2	7	7	X
easat-2167	345	3	]	]	X
easat-2167	345	4	y.	y.	PROPN
easat-2167	345	5	jiang	jiang	PROPN
easat-2167	345	6	,	,	PUNCT
easat-2167	345	7	et	et	PROPN
easat-2167	345	8	al	al	PROPN
easat-2167	345	9	.	.	PROPN
easat-2167	345	10	,	,	PUNCT
easat-2167	345	11	"	"	PUNCT
easat-2167	345	12	machine	machine	NOUN
easat-2167	345	13	vision	vision	NOUN
easat-2167	345	14	for	for	ADP
easat-2167	345	15	disease	disease	NOUN
easat-2167	345	16	spot	spot	NOUN
easat-2167	345	17	extraction	extraction	NOUN
easat-2167	345	18	in	in	ADP
easat-2167	345	19	rice	rice	NOUN
easat-2167	345	20	leaves	leave	NOUN
easat-2167	345	21	:	:	PUNCT
easat-2167	345	22	a	a	DET
easat-2167	345	23	gray	gray	ADJ
easat-2167	345	24	-	-	PUNCT
easat-2167	345	25	level	level	NOUN
easat-2167	345	26	transformation	transformation	NOUN
easat-2167	345	27	approach	approach	NOUN
easat-2167	345	28	,	,	PUNCT
easat-2167	345	29	"	"	PUNCT
easat-2167	345	30	computers	computer	NOUN
easat-2167	345	31	and	and	CCONJ
easat-2167	345	32	electronics	electronic	NOUN
easat-2167	345	33	in	in	ADP
easat-2167	345	34	agriculture	agriculture	NOUN
easat-2167	345	35	,	,	PUNCT
easat-2167	345	36	2015	2015	NUM
easat-2167	345	37	.	.	PUNCT
easat-2167	346	1	[	[	X
easat-2167	346	2	8	8	NUM
easat-2167	346	3	]	]	X
easat-2167	346	4	f.	f.	PROPN
easat-2167	346	5	g.	g.	PROPN
easat-2167	346	6	pascual	pascual	PROPN
easat-2167	346	7	,	,	PUNCT
easat-2167	346	8	et	et	PROPN
easat-2167	346	9	al	al	PROPN
easat-2167	346	10	.	.	PROPN
easat-2167	346	11	,	,	PUNCT
easat-2167	346	12	"	"	PUNCT
easat-2167	346	13	automated	automate	VERB
easat-2167	346	14	detection	detection	NOUN
easat-2167	346	15	of	of	ADP
easat-2167	346	16	rice	rice	NOUN
easat-2167	346	17	diseases	disease	NOUN
easat-2167	346	18	using	use	VERB
easat-2167	346	19	svm	svm	PROPN
easat-2167	346	20	and	and	CCONJ
easat-2167	346	21	random	random	ADJ
easat-2167	346	22	forest	forest	NOUN
easat-2167	346	23	classifiers	classifier	NOUN
easat-2167	346	24	,	,	PUNCT
easat-2167	346	25	"	"	PUNCT
easat-2167	346	26	international	international	ADJ
easat-2167	346	27	conference	conference	NOUN
easat-2167	346	28	on	on	ADP
easat-2167	346	29	advances	advance	NOUN
easat-2167	346	30	in	in	ADP
easat-2167	346	31	computing	computing	NOUN
easat-2167	346	32	,	,	PUNCT
easat-2167	346	33	communications	communication	NOUN
easat-2167	346	34	and	and	CCONJ
easat-2167	346	35	informatics	informatic	NOUN
easat-2167	346	36	(	(	PUNCT
easat-2167	346	37	icacci	icacci	PROPN
easat-2167	346	38	)	)	PUNCT
easat-2167	346	39	,	,	PUNCT
easat-2167	346	40	2019	2019	NUM
easat-2167	346	41	.	.	PUNCT
easat-2167	347	1	[	[	X
easat-2167	347	2	9	9	NUM
easat-2167	347	3	]	]	PUNCT
easat-2167	347	4	x.	x.	NOUN
easat-2167	347	5	li	li	PROPN
easat-2167	347	6	,	,	PUNCT
easat-2167	347	7	et	et	PROPN
easat-2167	347	8	al	al	PROPN
easat-2167	347	9	.	.	PROPN
easat-2167	347	10	,	,	PUNCT
easat-2167	347	11	"	"	PUNCT
easat-2167	347	12	copy	copy	VERB
easat-2167	347	13	paste	paste	NOUN
easat-2167	347	14	and	and	CCONJ
easat-2167	347	15	semantic	semantic	ADJ
easat-2167	347	16	segmentation	segmentation	NOUN
easat-2167	347	17	for	for	ADP
easat-2167	347	18	accurate	accurate	ADJ
easat-2167	347	19	disease	disease	NOUN
easat-2167	347	20	region	region	NOUN
easat-2167	347	21	segmentation	segmentation	NOUN
easat-2167	347	22	and	and	CCONJ
easat-2167	347	23	severity	severity	NOUN
easat-2167	347	24	assessment	assessment	NOUN
easat-2167	347	25	in	in	ADP
easat-2167	347	26	rice	rice	NOUN
easat-2167	347	27	leaves	leave	NOUN
easat-2167	347	28	,	,	PUNCT
easat-2167	347	29	"	"	PUNCT
easat-2167	347	30	international	international	ADJ
easat-2167	347	31	conference	conference	NOUN
easat-2167	347	32	on	on	ADP
easat-2167	347	33	image	image	NOUN
easat-2167	347	34	processing	processing	NOUN
easat-2167	347	35	(	(	PUNCT
easat-2167	347	36	icip	icip	PROPN
easat-2167	347	37	)	)	PUNCT
easat-2167	347	38	,	,	PUNCT
easat-2167	347	39	2022	2022	NUM
easat-2167	347	40	.	.	PUNCT
easat-2167	348	1	[	[	X
easat-2167	348	2	10	10	NUM
easat-2167	348	3	]	]	PUNCT
easat-2167	348	4	t.	t.	PROPN
easat-2167	348	5	daniya	daniya	PROPN
easat-2167	348	6	and	and	CCONJ
easat-2167	348	7	s.	s.	PROPN
easat-2167	348	8	vigneshwari	vigneshwari	PROPN
easat-2167	348	9	,	,	PUNCT
easat-2167	348	10	"	"	PUNCT
easat-2167	348	11	texture	texture	ADJ
easat-2167	348	12	and	and	CCONJ
easat-2167	348	13	deep	deep	ADJ
easat-2167	348	14	features	feature	NOUN
easat-2167	348	15	for	for	ADP
easat-2167	348	16	disease	disease	NOUN
easat-2167	348	17	detection	detection	NOUN
easat-2167	348	18	in	in	ADP
easat-2167	348	19	rice	rice	NOUN
easat-2167	348	20	plants	plant	NOUN
easat-2167	348	21	:	:	PUNCT
easat-2167	348	22	a	a	DET
easat-2167	348	23	deep	deep	ADJ
easat-2167	348	24	learning	learning	NOUN
easat-2167	348	25	approach	approach	NOUN
easat-2167	348	26	using	use	VERB
easat-2167	348	27	segnet	segnet	NOUN
easat-2167	348	28	and	and	CCONJ
easat-2167	348	29	deep	deep	ADJ
easat-2167	348	30	rnn	rnn	NOUN
easat-2167	348	31	,	,	PUNCT
easat-2167	348	32	"	"	PUNCT
easat-2167	348	33	ieee	ieee	NOUN
easat-2167	348	34	access	access	NOUN
easat-2167	348	35	,	,	PUNCT
easat-2167	348	36	2021	2021	NUM
easat-2167	348	37	.	.	PUNCT
easat-2167	349	1	[	[	X
easat-2167	349	2	11	11	NUM
easat-2167	349	3	]	]	X
easat-2167	349	4	r.	r.	PROPN
easat-2167	349	5	krishnamoorthy	krishnamoorthy	PROPN
easat-2167	349	6	,	,	PUNCT
easat-2167	349	7	et	et	PROPN
easat-2167	349	8	al	al	PROPN
easat-2167	349	9	.	.	PROPN
easat-2167	349	10	,	,	PUNCT
easat-2167	349	11	"	"	PUNCT
easat-2167	349	12	transfer	transfer	NOUN
easat-2167	349	13	learning	learning	NOUN
easat-2167	349	14	with	with	ADP
easat-2167	349	15	cnn	cnn	PROPN
easat-2167	349	16	for	for	ADP
easat-2167	349	17	predicting	predict	VERB
easat-2167	349	18	rice	rice	NOUN
easat-2167	349	19	leaf	leaf	NOUN
easat-2167	349	20	diseases	disease	NOUN
easat-2167	349	21	,	,	PUNCT
easat-2167	349	22	"	"	PUNCT
easat-2167	349	23	international	international	ADJ
easat-2167	349	24	journal	journal	NOUN
easat-2167	349	25	of	of	ADP
easat-2167	349	26	computational	computational	ADJ
easat-2167	349	27	intelligence	intelligence	NOUN
easat-2167	349	28	and	and	CCONJ
easat-2167	349	29	applications	application	NOUN
easat-2167	349	30	,	,	PUNCT
easat-2167	349	31	2021	2021	NUM
easat-2167	349	32	.	.	PUNCT
easat-2167	350	1	[	[	X
easat-2167	350	2	12	12	NUM
easat-2167	350	3	]	]	PUNCT
easat-2167	350	4	l.	l.	PROPN
easat-2167	350	5	pothen	pothen	PROPN
easat-2167	350	6	and	and	CCONJ
easat-2167	350	7	m.	m.	PROPN
easat-2167	350	8	m.	m.	PROPN
easat-2167	350	9	pai	pai	PROPN
easat-2167	350	10	,	,	PUNCT
easat-2167	350	11	"	"	PUNCT
easat-2167	350	12	otsu	otsu	PROPN
easat-2167	350	13	’s	’s	PART
easat-2167	350	14	method	method	NOUN
easat-2167	350	15	combined	combine	VERB
easat-2167	350	16	with	with	ADP
easat-2167	350	17	local	local	ADJ
easat-2167	350	18	binary	binary	ADJ
easat-2167	350	19	patterns	pattern	NOUN
easat-2167	350	20	and	and	CCONJ
easat-2167	350	21	histogram	histogram	NOUN
easat-2167	350	22	of	of	ADP
easat-2167	350	23	oriented	orient	VERB
easat-2167	350	24	gradients	gradient	NOUN
easat-2167	350	25	for	for	ADP
easat-2167	350	26	rice	rice	NOUN
easat-2167	350	27	disease	disease	NOUN
easat-2167	350	28	classification	classification	NOUN
easat-2167	350	29	,	,	PUNCT
easat-2167	350	30	"	"	PUNCT
easat-2167	350	31	pattern	pattern	NOUN
easat-2167	350	32	recognition	recognition	NOUN
easat-2167	350	33	letters	letter	NOUN
easat-2167	350	34	,	,	PUNCT
easat-2167	350	35	2020	2020	NUM
easat-2167	350	36	.	.	PUNCT
easat-2167	351	1	[	[	X
easat-2167	351	2	13	13	NUM
easat-2167	351	3	]	]	PUNCT
easat-2167	351	4	s.	s.	PROPN
easat-2167	351	5	zhou	zhou	PROPN
easat-2167	351	6	,	,	PUNCT
easat-2167	351	7	et	et	PROPN
easat-2167	351	8	al	al	PROPN
easat-2167	351	9	.	.	PROPN
easat-2167	351	10	,	,	PUNCT
easat-2167	351	11	"	"	PUNCT
easat-2167	351	12	fusion	fusion	NOUN
easat-2167	351	13	of	of	ADP
easat-2167	351	14	fcm	fcm	ADJ
easat-2167	351	15	-	-	ADJ
easat-2167	351	16	km	km	NOUN
easat-2167	351	17	clustering	clustering	NOUN
easat-2167	351	18	and	and	CCONJ
easat-2167	351	19	faster	fast	ADJ
easat-2167	351	20	r	r	NOUN
easat-2167	351	21	-	-	PUNCT
easat-2167	351	22	cnn	cnn	PROPN
easat-2167	351	23	for	for	ADP
easat-2167	351	24	rapid	rapid	ADJ
easat-2167	351	25	detection	detection	NOUN
easat-2167	351	26	of	of	ADP
easat-2167	351	27	rice	rice	NOUN
easat-2167	351	28	diseases	disease	NOUN
easat-2167	351	29	,	,	PUNCT
easat-2167	351	30	"	"	PUNCT
easat-2167	351	31	ieee	ieee	NOUN
easat-2167	351	32	transactions	transaction	NOUN
easat-2167	351	33	on	on	ADP
easat-2167	351	34	geoscience	geoscience	NOUN
easat-2167	351	35	and	and	CCONJ
easat-2167	351	36	remote	remote	ADJ
easat-2167	351	37	sensing	sensing	NOUN
easat-2167	351	38	,	,	PUNCT
easat-2167	351	39	2019	2019	NUM
easat-2167	351	40	.	.	PUNCT
easat-2167	352	1	[	[	X
easat-2167	352	2	14	14	NUM
easat-2167	352	3	]	]	X
easat-2167	352	4	y.	y.	PROPN
easat-2167	352	5	lu	lu	PROPN
easat-2167	352	6	,	,	PUNCT
easat-2167	352	7	et	et	PROPN
easat-2167	352	8	al	al	PROPN
easat-2167	352	9	.	.	PROPN
easat-2167	352	10	,	,	PUNCT
easat-2167	352	11	"	"	PUNCT
easat-2167	352	12	deep	deep	ADJ
easat-2167	352	13	cnn	cnn	NOUN
easat-2167	352	14	for	for	ADP
easat-2167	352	15	identifying	identify	VERB
easat-2167	352	16	10	10	NUM
easat-2167	352	17	common	common	ADJ
easat-2167	352	18	rice	rice	NOUN
easat-2167	352	19	diseases	disease	NOUN
easat-2167	352	20	:	:	PUNCT
easat-2167	352	21	an	an	DET
easat-2167	352	22	evaluation	evaluation	NOUN
easat-2167	352	23	of	of	ADP
easat-2167	352	24	deep	deep	ADJ
easat-2167	352	25	learning	learning	NOUN
easat-2167	352	26	approaches	approach	NOUN
easat-2167	352	27	,	,	PUNCT
easat-2167	352	28	"	"	PUNCT
easat-2167	352	29	computers	computer	NOUN
easat-2167	352	30	and	and	CCONJ
easat-2167	352	31	electronics	electronic	NOUN
easat-2167	352	32	in	in	ADP
easat-2167	352	33	agriculture	agriculture	NOUN
easat-2167	352	34	,	,	PUNCT
easat-2167	352	35	2017	2017	NUM
easat-2167	352	36	.	.	PUNCT
easat-2167	353	1	[	[	X
easat-2167	353	2	15	15	NUM
easat-2167	353	3	]	]	X
easat-2167	353	4	j.	j.	PROPN
easat-2167	353	5	chen	chen	PROPN
easat-2167	353	6	,	,	PUNCT
easat-2167	353	7	et	et	PROPN
easat-2167	353	8	al	al	PROPN
easat-2167	353	9	.	.	PROPN
easat-2167	353	10	,	,	PUNCT
easat-2167	353	11	"	"	PUNCT
easat-2167	353	12	attention	attention	NOUN
easat-2167	353	13	networks	network	NOUN
easat-2167	353	14	for	for	ADP
easat-2167	353	15	rice	rice	NOUN
easat-2167	353	16	disease	disease	NOUN
easat-2167	353	17	detection	detection	NOUN
easat-2167	353	18	:	:	PUNCT
easat-2167	353	19	using	use	VERB
easat-2167	353	20	lightweight	lightweight	ADJ
easat-2167	353	21	models	model	NOUN
easat-2167	353	22	based	base	VERB
easat-2167	353	23	on	on	ADP
easat-2167	353	24	mobilenet	mobilenet	NOUN
easat-2167	353	25	-	-	PUNCT
easat-2167	353	26	v2	v2	NOUN
easat-2167	353	27	,	,	PUNCT
easat-2167	353	28	"	"	PUNCT
easat-2167	353	29	computers	computer	NOUN
easat-2167	353	30	and	and	CCONJ
easat-2167	353	31	electronics	electronic	NOUN
easat-2167	353	32	in	in	ADP
easat-2167	353	33	agriculture	agriculture	NOUN
easat-2167	353	34	,	,	PUNCT
easat-2167	353	35	2021	2021	NUM
easat-2167	353	36	.	.	PUNCT
easat-2167	354	1	[	[	X
easat-2167	354	2	16	16	NUM
easat-2167	354	3	]	]	X
easat-2167	354	4	c.	c.	PROPN
easat-2167	354	5	orillo	orillo	PROPN
easat-2167	354	6	,	,	PUNCT
easat-2167	354	7	et	et	PROPN
easat-2167	354	8	al	al	PROPN
easat-2167	354	9	.	.	PROPN
easat-2167	354	10	,	,	PUNCT
easat-2167	354	11	"	"	PUNCT
easat-2167	354	12	backpropagation	backpropagation	NOUN
easat-2167	354	13	neural	neural	ADJ
easat-2167	354	14	network	network	NOUN
easat-2167	354	15	for	for	ADP
easat-2167	354	16	detecting	detect	VERB
easat-2167	354	17	three	three	NUM
easat-2167	354	18	common	common	ADJ
easat-2167	354	19	rice	rice	NOUN
easat-2167	354	20	diseases	disease	NOUN
easat-2167	354	21	,	,	PUNCT
easat-2167	354	22	"	"	PUNCT
easat-2167	354	23	international	international	ADJ
easat-2167	354	24	journal	journal	NOUN
easat-2167	354	25	of	of	ADP
easat-2167	354	26	computer	computer	NOUN
easat-2167	354	27	applications	application	NOUN
easat-2167	354	28	,	,	PUNCT
easat-2167	354	29	2014	2014	NUM
easat-2167	354	30	.	.	PUNCT
easat-2167	355	1	[	[	X
easat-2167	355	2	17	17	NUM
easat-2167	355	3	]	]	X
easat-2167	355	4	v.	v.	CCONJ
easat-2167	355	5	s.	s.	PROPN
easat-2167	355	6	shah	shah	PROPN
easat-2167	355	7	,	,	PUNCT
easat-2167	355	8	et	et	PROPN
easat-2167	355	9	al	al	PROPN
easat-2167	355	10	.	.	PROPN
easat-2167	355	11	,	,	PUNCT
easat-2167	355	12	"	"	PUNCT
easat-2167	355	13	a	a	DET
easat-2167	355	14	survey	survey	NOUN
easat-2167	355	15	on	on	ADP
easat-2167	355	16	rice	rice	NOUN
easat-2167	355	17	disease	disease	NOUN
easat-2167	355	18	detection	detection	NOUN
easat-2167	355	19	using	use	VERB
easat-2167	355	20	image	image	NOUN
easat-2167	355	21	processing	processing	NOUN
easat-2167	355	22	and	and	CCONJ
easat-2167	355	23	machine	machine	NOUN
easat-2167	355	24	learning	learn	VERB
easat-2167	355	25	techniques	technique	NOUN
easat-2167	355	26	,	,	PUNCT
easat-2167	355	27	"	"	PUNCT
easat-2167	355	28	journal	journal	NOUN
easat-2167	355	29	of	of	ADP
easat-2167	355	30	computational	computational	ADJ
easat-2167	355	31	and	and	CCONJ
easat-2167	355	32	theoretical	theoretical	ADJ
easat-2167	355	33	nanoscience	nanoscience	NOUN
easat-2167	355	34	,	,	PUNCT
easat-2167	355	35	2016	2016	NUM
easat-2167	355	36	.	.	PUNCT
easat-2167	356	1	[	[	X
easat-2167	356	2	18	18	NUM
easat-2167	356	3	]	]	X
easat-2167	356	4	n.	n.	PROPN
easat-2167	356	5	s.	s.	PROPN
easat-2167	356	6	patil	patil	PROPN
easat-2167	356	7	and	and	CCONJ
easat-2167	356	8	a.	a.	PROPN
easat-2167	356	9	kumar	kumar	PROPN
easat-2167	356	10	,	,	PUNCT
easat-2167	356	11	"	"	PUNCT
easat-2167	356	12	rice	rice	NOUN
easat-2167	356	13	-	-	PUNCT
easat-2167	356	14	fusion	fusion	NOUN
easat-2167	356	15	:	:	PUNCT
easat-2167	356	16	a	a	DET
easat-2167	356	17	multimodality	multimodality	NOUN
easat-2167	356	18	data	datum	NOUN
easat-2167	356	19	fusion	fusion	NOUN
easat-2167	356	20	framework	framework	NOUN
easat-2167	356	21	for	for	ADP
easat-2167	356	22	rice	rice	NOUN
easat-2167	356	23	disease	disease	NOUN
easat-2167	356	24	diagnosis	diagnosis	NOUN
easat-2167	356	25	,	,	PUNCT
easat-2167	356	26	"	"	PUNCT
easat-2167	356	27	journal	journal	NOUN
easat-2167	356	28	of	of	ADP
easat-2167	356	29	agricultural	agricultural	ADJ
easat-2167	356	30	and	and	CCONJ
easat-2167	356	31	food	food	NOUN
easat-2167	356	32	information	information	NOUN
easat-2167	356	33	,	,	PUNCT
easat-2167	356	34	2022	2022	NUM
easat-2167	356	35	.	.	PUNCT
easat-2167	357	1	[	[	X
easat-2167	357	2	19	19	NUM
easat-2167	357	3	]	]	PUNCT
easat-2167	357	4	s.	s.	PROPN
easat-2167	357	5	s.	s.	PROPN
easat-2167	357	6	chawathe	chawathe	PROPN
easat-2167	357	7	,	,	PUNCT
easat-2167	357	8	"	"	PUNCT
easat-2167	357	9	image	image	NOUN
easat-2167	357	10	analysis	analysis	NOUN
easat-2167	357	11	for	for	ADP
easat-2167	357	12	rice	rice	NOUN
easat-2167	357	13	disease	disease	NOUN
easat-2167	357	14	detection	detection	NOUN
easat-2167	357	15	using	use	VERB
easat-2167	357	16	leaf	leaf	NOUN
easat-2167	357	17	and	and	CCONJ
easat-2167	357	18	lesion	lesion	NOUN
easat-2167	357	19	attributes	attribute	VERB
easat-2167	357	20	,	,	PUNCT
easat-2167	357	21	"	"	PUNCT
easat-2167	357	22	proceedings	proceeding	NOUN
easat-2167	357	23	of	of	ADP
easat-2167	357	24	the	the	DET
easat-2167	357	25	international	international	ADJ
easat-2167	357	26	conference	conference	NOUN
easat-2167	357	27	on	on	ADP
easat-2167	357	28	image	image	NOUN
easat-2167	357	29	processing	processing	NOUN
easat-2167	357	30	and	and	CCONJ
easat-2167	357	31	vision	vision	NOUN
easat-2167	357	32	science	science	NOUN
easat-2167	357	33	,	,	PUNCT
easat-2167	357	34	2020	2020	NUM
easat-2167	357	35	.	.	PUNCT
easat-2167	358	1	[	[	X
easat-2167	358	2	20	20	NUM
easat-2167	358	3	]	]	PUNCT
easat-2167	358	4	p.	p.	PROPN
easat-2167	358	5	joshi	joshi	PROPN
easat-2167	358	6	and	and	CCONJ
easat-2167	358	7	r.	r.	PROPN
easat-2167	358	8	jadhav	jadhav	PROPN
easat-2167	358	9	,	,	PUNCT
easat-2167	358	10	"	"	PUNCT
easat-2167	358	11	an	an	DET
easat-2167	358	12	automated	automate	VERB
easat-2167	358	13	system	system	NOUN
easat-2167	358	14	for	for	ADP
easat-2167	358	15	monitoring	monitor	VERB
easat-2167	358	16	and	and	CCONJ
easat-2167	358	17	controlling	control	VERB
easat-2167	358	18	rice	rice	NOUN
easat-2167	358	19	diseases	disease	NOUN
easat-2167	358	20	using	use	VERB
easat-2167	358	21	image	image	NOUN
easat-2167	358	22	processing	processing	NOUN
easat-2167	358	23	techniques	technique	NOUN
easat-2167	358	24	,	,	PUNCT
easat-2167	358	25	"	"	PUNCT
easat-2167	358	26	international	international	ADJ
easat-2167	358	27	journal	journal	NOUN
easat-2167	358	28	of	of	ADP
easat-2167	358	29	computer	computer	NOUN
easat-2167	358	30	applications	application	NOUN
easat-2167	358	31	,	,	PUNCT
easat-2167	358	32	2016	2016	NUM
easat-2167	358	33	.	.	PUNCT
easat-2167	359	1	[	[	X
easat-2167	359	2	online	online	X
easat-2167	359	3	]	]	X
easat-2167	359	4	.	.	PUNCT
easat-2167	360	1	available	available	ADJ
easat-2167	360	2	:	:	PUNCT
easat-2167	360	3	https://www.kaggle.com/datasets/slygirl/rice-leaf-disease-with-segmentation-labels	https://www.kaggle.com/datasets/slygirl/rice-leaf-disease-with-segmentation-label	VERB
easat-2167	360	4	.	.	PUNCT
easat-2167	361	1	[	[	X
easat-2167	361	2	21	21	NUM
easat-2167	361	3	]	]	X
easat-2167	361	4	m.	m.	NOUN
easat-2167	361	5	sandler	sandler	PROPN
easat-2167	361	6	,	,	PUNCT
easat-2167	361	7	a.	a.	PROPN
easat-2167	361	8	howard	howard	PROPN
easat-2167	361	9	,	,	PUNCT
easat-2167	361	10	m.	m.	PROPN
easat-2167	361	11	zhu	zhu	PROPN
easat-2167	361	12	,	,	PUNCT
easat-2167	361	13	a.	a.	NOUN
easat-2167	361	14	zhmoginov	zhmoginov	PROPN
easat-2167	361	15	,	,	PUNCT
easat-2167	361	16	and	and	CCONJ
easat-2167	361	17	l.	l.	PROPN
easat-2167	361	18	c.	c.	PROPN
easat-2167	361	19	chen	chen	PROPN
easat-2167	361	20	,	,	PUNCT
easat-2167	361	21	"	"	PUNCT
easat-2167	361	22	mobilenetv2	mobilenetv2	NOUN
easat-2167	361	23	:	:	PUNCT
easat-2167	361	24	inverted	inverted	ADJ
easat-2167	361	25	residuals	residual	NOUN
easat-2167	361	26	and	and	CCONJ
easat-2167	361	27	linear	linear	ADJ
easat-2167	361	28	bottlenecks	bottleneck	NOUN
easat-2167	361	29	,	,	PUNCT
easat-2167	361	30	"	"	PUNCT
easat-2167	361	31	proceedings	proceeding	NOUN
easat-2167	361	32	of	of	ADP
easat-2167	361	33	the	the	DET
easat-2167	361	34	ieee	ieee	NOUN
easat-2167	361	35	/	/	SYM
easat-2167	361	36	cvf	cvf	NOUN
easat-2167	361	37	conference	conference	NOUN
easat-2167	361	38	on	on	ADP
easat-2167	361	39	computer	computer	NOUN
easat-2167	361	40	vision	vision	NOUN
easat-2167	361	41	and	and	CCONJ
easat-2167	361	42	pattern	pattern	NOUN
easat-2167	361	43	recognition	recognition	NOUN
easat-2167	361	44	(	(	PUNCT
easat-2167	361	45	cvpr	cvpr	NOUN
easat-2167	361	46	)	)	PUNCT
easat-2167	361	47	,	,	PUNCT
easat-2167	361	48	2018	2018	NUM
easat-2167	361	49	.	.	PUNCT
easat-2167	362	1	[	[	X
easat-2167	362	2	22	22	NUM
easat-2167	362	3	]	]	X
easat-2167	362	4	o.	o.	NOUN
easat-2167	362	5	ronneberger	ronneberger	NOUN
easat-2167	362	6	,	,	PUNCT
easat-2167	362	7	p.	p.	NOUN
easat-2167	362	8	fischer	fischer	PROPN
easat-2167	362	9	,	,	PUNCT
easat-2167	362	10	and	and	CCONJ
easat-2167	362	11	t.	t.	PROPN
easat-2167	362	12	brox	brox	PROPN
easat-2167	362	13	,	,	PUNCT
easat-2167	362	14	"	"	PUNCT
easat-2167	362	15	u	u	NOUN
easat-2167	362	16	-	-	NOUN
easat-2167	362	17	net	net	ADJ
easat-2167	362	18	:	:	PUNCT
easat-2167	362	19	convolutional	convolutional	ADJ
easat-2167	362	20	networks	network	NOUN
easat-2167	362	21	for	for	ADP
easat-2167	362	22	biomedical	biomedical	ADJ
easat-2167	362	23	image	image	NOUN
easat-2167	362	24	segmentation	segmentation	NOUN
easat-2167	362	25	,	,	PUNCT
easat-2167	362	26	"	"	PUNCT
easat-2167	362	27	international	international	ADJ
easat-2167	362	28	conference	conference	NOUN
easat-2167	362	29	on	on	ADP
easat-2167	362	30	medical	medical	ADJ
easat-2167	362	31	image	image	NOUN
easat-2167	362	32	computing	computing	NOUN
easat-2167	362	33	and	and	CCONJ
easat-2167	362	34	computer	computer	NOUN
easat-2167	362	35	-	-	PUNCT
easat-2167	362	36	assisted	assist	VERB
easat-2167	362	37	intervention	intervention	NOUN
easat-2167	362	38	(	(	PUNCT
easat-2167	362	39	miccai	miccai	PROPN
easat-2167	362	40	)	)	PUNCT
easat-2167	362	41	,	,	PUNCT
easat-2167	362	42	2015	2015	NUM
easat-2167	362	43	.	.	PUNCT
easat-2167	363	1	[	[	X
easat-2167	363	2	23	23	NUM
easat-2167	363	3	]	]	PUNCT
easat-2167	363	4	l.	l.	PROPN
easat-2167	363	5	c.	c.	PROPN
easat-2167	363	6	chen	chen	PROPN
easat-2167	363	7	,	,	PUNCT
easat-2167	363	8	y.	y.	PROPN
easat-2167	363	9	zhu	zhu	PROPN
easat-2167	363	10	,	,	PUNCT
easat-2167	363	11	g.	g.	PROPN
easat-2167	363	12	papandreou	papandreou	PROPN
easat-2167	363	13	,	,	PUNCT
easat-2167	363	14	f.	f.	PROPN
easat-2167	363	15	schroff	schroff	PROPN
easat-2167	363	16	,	,	PUNCT
easat-2167	363	17	and	and	CCONJ
easat-2167	363	18	h.	h.	PROPN
easat-2167	363	19	adam	adam	PROPN
easat-2167	363	20	,	,	PUNCT
easat-2167	363	21	"	"	PUNCT
easat-2167	363	22	encoder	encoder	NOUN
easat-2167	363	23	-	-	PUNCT
easat-2167	363	24	decoder	decoder	NOUN
easat-2167	363	25	with	with	ADP
easat-2167	363	26	atrousseparable	atrousseparable	ADJ
easat-2167	363	27	convolution	convolution	NOUN
easat-2167	363	28	for	for	ADP
easat-2167	363	29	semantic	semantic	ADJ
easat-2167	363	30	image	image	NOUN
easat-2167	363	31	segmentation	segmentation	NOUN
easat-2167	363	32	,	,	PUNCT
easat-2167	363	33	"	"	PUNCT
easat-2167	363	34	proceedings	proceeding	NOUN
easat-2167	363	35	of	of	ADP
easat-2167	363	36	the	the	DET
easat-2167	363	37	european	european	PROPN
easat-2167	363	38	conference	conference	PROPN
easat-2167	363	39	on	on	ADP
easat-2167	363	40	computer	computer	NOUN
easat-2167	363	41	vision	vision	NOUN
easat-2167	363	42	(	(	PUNCT
easat-2167	363	43	eccv	eccv	ADV
easat-2167	363	44	)	)	PUNCT
easat-2167	363	45	,	,	PUNCT
easat-2167	363	46	2018	2018	NUM
easat-2167	363	47	.	.	PUNCT
easat-2167	364	1	[	[	X
easat-2167	364	2	24	24	NUM
easat-2167	364	3	]	]	PUNCT
easat-2167	364	4	m.	m.	NOUN
easat-2167	364	5	tan	tan	PROPN
easat-2167	364	6	and	and	CCONJ
easat-2167	364	7	q.	q.	PROPN
easat-2167	364	8	le	le	PROPN
easat-2167	364	9	,	,	PUNCT
easat-2167	364	10	"	"	PUNCT
easat-2167	364	11	efficientnet	efficientnet	NOUN
easat-2167	364	12	:	:	PUNCT
easat-2167	364	13	rethinking	rethink	VERB
easat-2167	364	14	model	model	NOUN
easat-2167	364	15	scaling	scale	VERB
easat-2167	364	16	for	for	ADP
easat-2167	364	17	convolutional	convolutional	ADJ
easat-2167	364	18	neural	neural	ADJ
easat-2167	364	19	networks	network	NOUN
easat-2167	364	20	,	,	PUNCT
easat-2167	364	21	"	"	PUNCT
easat-2167	364	22	proceedings	proceeding	NOUN
easat-2167	364	23	of	of	ADP
easat-2167	364	24	the	the	DET
easat-2167	364	25	36th	36th	ADJ
easat-2167	364	26	international	international	ADJ
easat-2167	364	27	conference	conference	NOUN
easat-2167	364	28	on	on	ADP
easat-2167	364	29	machine	machine	NOUN
easat-2167	364	30	learning	learning	NOUN
easat-2167	364	31	(	(	PUNCT
easat-2167	364	32	icml	icml	PROPN
easat-2167	364	33	)	)	PUNCT
easat-2167	364	34	,	,	PUNCT
easat-2167	364	35	2019	2019	NUM
easat-2167	364	36	.	.	PUNCT
easat-2167	365	1	[	[	X
easat-2167	365	2	25	25	NUM
easat-2167	365	3	]	]	X
easat-2167	365	4	o.	o.	PROPN
easat-2167	365	5	oktay	oktay	PROPN
easat-2167	365	6	,	,	PUNCT
easat-2167	365	7	j.	j.	PROPN
easat-2167	365	8	schlemper	schlemper	PROPN
easat-2167	365	9	,	,	PUNCT
easat-2167	365	10	l.	l.	PROPN
easat-2167	365	11	le	le	PROPN
easat-2167	365	12	folgoc	folgoc	PROPN
easat-2167	365	13	,	,	PUNCT
easat-2167	365	14	m.	m.	PROPN
easat-2167	365	15	lee	lee	PROPN
easat-2167	365	16	,	,	PUNCT
easat-2167	365	17	m.	m.	NOUN
easat-2167	365	18	heinrich	heinrich	PROPN
easat-2167	365	19	,	,	PUNCT
easat-2167	365	20	k.	k.	PROPN
easat-2167	365	21	misawa	misawa	PROPN
easat-2167	365	22	,	,	PUNCT
easat-2167	365	23	k.	k.	PROPN
easat-2167	365	24	mori	mori	PROPN
easat-2167	365	25	,	,	PUNCT
easat-2167	365	26	s.	s.	PROPN
easat-2167	365	27	mcdonagh	mcdonagh	PROPN
easat-2167	365	28	,	,	PUNCT
easat-2167	365	29	n.	n.	PROPN
easat-2167	365	30	y.	y.	PROPN
easat-2167	365	31	hammerla	hammerla	PROPN
easat-2167	365	32	,	,	PUNCT
easat-2167	365	33	b.	b.	PROPN
easat-2167	365	34	kainz	kainz	PROPN
easat-2167	365	35	,	,	PUNCT
easat-2167	365	36	b.	b.	PROPN
easat-2167	365	37	glocker	glocker	PROPN
easat-2167	365	38	,	,	PUNCT
easat-2167	365	39	and	and	CCONJ
easat-2167	365	40	d.	d.	PROPN
easat-2167	365	41	rueckert	rueckert	PROPN
easat-2167	365	42	,	,	PUNCT
easat-2167	365	43	"	"	PUNCT
easat-2167	365	44	attention	attention	NOUN
easat-2167	365	45	u	u	NOUN
easat-2167	365	46	-	-	NOUN
easat-2167	365	47	net	net	ADJ
easat-2167	365	48	:	:	PUNCT
easat-2167	365	49	learning	learn	VERB
easat-2167	365	50	where	where	SCONJ
easat-2167	365	51	to	to	PART
easat-2167	365	52	look	look	VERB
easat-2167	365	53	for	for	ADP
easat-2167	365	54	the	the	DET
easat-2167	365	55	pancreas	pancrea	NOUN
easat-2167	365	56	,	,	PUNCT
easat-2167	365	57	"	"	PUNCT
easat-2167	365	58	arxiv	arxiv	PROPN
easat-2167	365	59	preprint	preprint	NOUN
easat-2167	365	60	arxiv:1804.03999	arxiv:1804.03999	NOUN
easat-2167	365	61	,	,	PUNCT
easat-2167	365	62	2018	2018	NUM
easat-2167	365	63	.	.	PUNCT
easat-2167	366	1	https://creativecommons.org/licenses/by/4.0/	https://creativecommons.org/licenses/by/4.0/	PROPN
easat-2167	366	2	https://www.kaggle.com/datasets/slygirl/rice-leaf-disease-with-segmentation-labels	https://www.kaggle.com/datasets/slygirl/rice-leaf-disease-with-segmentation-label	VERB