id	sid	tid	token	lemma	pos
easat-3242	1	1	edelweiss	edelweiss	PROPN
easat-3242	1	2	applied	apply	VERB
easat-3242	1	3	science	science	NOUN
easat-3242	1	4	and	and	CCONJ
easat-3242	1	5	technology	technology	NOUN
easat-3242	1	6	issn	issn	PROPN
easat-3242	1	7	:	:	PUNCT
easat-3242	1	8	2576	2576	NUM
easat-3242	1	9	-	-	SYM
easat-3242	1	10	8484	8484	NUM
easat-3242	1	11	vol	vol	NOUN
easat-3242	1	12	.	.	PROPN
easat-3242	1	13	8	8	NUM
easat-3242	1	14	,	,	PUNCT
easat-3242	1	15	no	no	INTJ
easat-3242	1	16	.	.	NOUN
easat-3242	1	17	6	6	NUM
easat-3242	1	18	,	,	PUNCT
easat-3242	1	19	5707	5707	NUM
easat-3242	1	20	-	-	SYM
easat-3242	1	21	5722	5722	NUM
easat-3242	1	22	2024	2024	NUM
easat-3242	1	23	publisher	publisher	NOUN
easat-3242	1	24	:	:	PUNCT
easat-3242	1	25	learning	learn	VERB
easat-3242	1	26	gate	gate	NOUN
easat-3242	1	27	doi	doi	PROPN
easat-3242	1	28	:	:	PUNCT
easat-3242	1	29	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	1	30	©	©	PROPN
easat-3242	1	31	2024	2024	NUM
easat-3242	1	32	by	by	ADP
easat-3242	1	33	the	the	DET
easat-3242	1	34	authors	author	NOUN
easat-3242	1	35	;	;	PUNCT
easat-3242	1	36	licensee	licensee	PROPN
easat-3242	1	37	learning	learning	NOUN
easat-3242	1	38	gate	gate	NOUN
easat-3242	1	39	©	©	PROPN
easat-3242	1	40	2024	2024	NUM
easat-3242	1	41	by	by	ADP
easat-3242	1	42	the	the	DET
easat-3242	1	43	authors	author	NOUN
easat-3242	1	44	;	;	PUNCT
easat-3242	1	45	licensee	licensee	PROPN
easat-3242	1	46	learning	learn	VERB
easat-3242	1	47	gate	gate	NOUN
easat-3242	1	48	*	*	PUNCT
easat-3242	1	49	correspondence	correspondence	NOUN
easat-3242	1	50	:	:	PUNCT
easat-3242	1	51	raju.everest8848@gmail.com	raju.everest8848@gmail.com	PROPN
easat-3242	1	52	satellite	satellite	NOUN
easat-3242	1	53	image	image	NOUN
easat-3242	1	54	segmentation	segmentation	NOUN
easat-3242	1	55	using	use	VERB
easat-3242	1	56	unet++	unet++	PROPN
easat-3242	1	57	with	with	ADP
easat-3242	1	58	vgg19	vgg19	PROPN
easat-3242	1	59	deep	deep	ADJ
easat-3242	1	60	learning	learning	NOUN
easat-3242	1	61	model	model	PROPN
easat-3242	1	62	yaragorla	yaragorla	PROPN
easat-3242	1	63	raju1	raju1	PROPN
easat-3242	1	64	*	*	NUM
easat-3242	1	65	,	,	PUNCT
easat-3242	1	66	m.	m.	NOUN
easat-3242	1	67	narayana2	narayana2	PROPN
easat-3242	2	1	1,2department	1,2department	NUM
easat-3242	2	2	of	of	ADP
easat-3242	2	3	electronics	electronic	NOUN
easat-3242	2	4	and	and	CCONJ
easat-3242	2	5	communication	communication	NOUN
easat-3242	2	6	engineering	engineering	NOUN
easat-3242	2	7	,	,	PUNCT
easat-3242	2	8	anurag	anurag	PROPN
easat-3242	2	9	university	university	PROPN
easat-3242	2	10	,	,	PUNCT
easat-3242	2	11	venkatapur	venkatapur	NOUN
easat-3242	2	12	,	,	PUNCT
easat-3242	2	13	ghatkesar	ghatkesar	PROPN
easat-3242	2	14	rd	rd	PROPN
easat-3242	2	15	,	,	PUNCT
easat-3242	2	16	hyderabad	hyderabad	PROPN
easat-3242	2	17	,	,	PUNCT
easat-3242	2	18	telangana	telangana	PROPN
easat-3242	2	19	,	,	PUNCT
easat-3242	2	20	500088	500088	NUM
easat-3242	2	21	;	;	PUNCT
easat-3242	2	22	raju.everest8848@gmail.com	raju.everest8848@gmail.com	NUM
easat-3242	2	23	(	(	PUNCT
easat-3242	2	24	y.r	y.r	PROPN
easat-3242	2	25	.	.	PROPN
easat-3242	2	26	)	)	PUNCT
easat-3242	3	1	narayanaece@anurag.edu.in	narayanaece@anurag.edu.in	NOUN
easat-3242	3	2	(	(	PUNCT
easat-3242	3	3	m.n	m.n	PROPN
easat-3242	3	4	.	.	PUNCT
easat-3242	3	5	)	)	PUNCT
easat-3242	4	1	abstract	abstract	ADJ
easat-3242	4	2	:	:	PUNCT
easat-3242	4	3	satellite	satellite	NOUN
easat-3242	4	4	image	image	NOUN
easat-3242	4	5	segmentation	segmentation	NOUN
easat-3242	4	6	is	be	AUX
easat-3242	4	7	an	an	DET
easat-3242	4	8	essential	essential	ADJ
easat-3242	4	9	step	step	NOUN
easat-3242	4	10	in	in	ADP
easat-3242	4	11	many	many	ADJ
easat-3242	4	12	applications	application	NOUN
easat-3242	4	13	,	,	PUNCT
easat-3242	4	14	including	include	VERB
easat-3242	4	15	urban	urban	ADJ
easat-3242	4	16	planning	planning	NOUN
easat-3242	4	17	,	,	PUNCT
easat-3242	4	18	disaster	disaster	NOUN
easat-3242	4	19	response	response	NOUN
easat-3242	4	20	and	and	CCONJ
easat-3242	4	21	environmental	environmental	ADJ
easat-3242	4	22	monitoring	monitoring	NOUN
easat-3242	4	23	.	.	PUNCT
easat-3242	5	1	the	the	DET
easat-3242	5	2	problem	problem	NOUN
easat-3242	5	3	though	though	ADV
easat-3242	5	4	is	be	AUX
easat-3242	5	5	that	that	SCONJ
easat-3242	5	6	existing	exist	VERB
easat-3242	5	7	methods	method	NOUN
easat-3242	5	8	suffer	suffer	VERB
easat-3242	5	9	from	from	ADP
easat-3242	5	10	high	high	ADJ
easat-3242	5	11	failure	failure	NOUN
easat-3242	5	12	rates	rate	NOUN
easat-3242	5	13	because	because	SCONJ
easat-3242	5	14	of	of	ADP
easat-3242	5	15	the	the	DET
easat-3242	5	16	intrinsic	intrinsic	ADJ
easat-3242	5	17	complexity	complexity	NOUN
easat-3242	5	18	and	and	CCONJ
easat-3242	5	19	variation	variation	NOUN
easat-3242	5	20	found	find	VERB
easat-3242	5	21	in	in	ADP
easat-3242	5	22	satellite	satellite	NOUN
easat-3242	5	23	images	image	NOUN
easat-3242	5	24	.	.	PUNCT
easat-3242	6	1	this	this	DET
easat-3242	6	2	research	research	NOUN
easat-3242	6	3	uses	use	VERB
easat-3242	6	4	deep	deep	ADJ
easat-3242	6	5	learning	learn	VERB
easat-3242	6	6	unet++	unet++	NUM
easat-3242	6	7	and	and	CCONJ
easat-3242	6	8	vgg19	vgg19	PROPN
easat-3242	6	9	models	model	NOUN
easat-3242	6	10	to	to	PART
easat-3242	6	11	construct	construct	VERB
easat-3242	6	12	advanced	advanced	ADJ
easat-3242	6	13	satellite	satellite	NOUN
easat-3242	6	14	image	image	NOUN
easat-3242	6	15	segmentation	segmentation	NOUN
easat-3242	6	16	method	method	NOUN
easat-3242	6	17	as	as	SCONJ
easat-3242	6	18	we	we	PRON
easat-3242	6	19	propose	propose	VERB
easat-3242	6	20	a	a	DET
easat-3242	6	21	brand	brand	NOUN
easat-3242	6	22	-	-	PUNCT
easat-3242	6	23	new	new	ADJ
easat-3242	6	24	approach	approach	NOUN
easat-3242	6	25	of	of	ADP
easat-3242	6	26	our	our	PRON
easat-3242	6	27	own	own	ADJ
easat-3242	6	28	.	.	PUNCT
easat-3242	7	1	in	in	ADP
easat-3242	7	2	this	this	DET
easat-3242	7	3	study	study	NOUN
easat-3242	7	4	,	,	PUNCT
easat-3242	7	5	the	the	DET
easat-3242	7	6	effective	effective	ADJ
easat-3242	7	7	segmentation	segmentation	NOUN
easat-3242	7	8	method	method	NOUN
easat-3242	7	9	combines	combine	VERB
easat-3242	7	10	the	the	DET
easat-3242	7	11	powerful	powerful	ADJ
easat-3242	7	12	feature	feature	NOUN
easat-3242	7	13	extraction	extraction	NOUN
easat-3242	7	14	capability	capability	NOUN
easat-3242	7	15	of	of	ADP
easat-3242	7	16	vgg19	vgg19	PROPN
easat-3242	7	17	model	model	NOUN
easat-3242	7	18	which	which	PRON
easat-3242	7	19	is	be	AUX
easat-3242	7	20	improved	improve	VERB
easat-3242	7	21	version	version	NOUN
easat-3242	7	22	based	base	VERB
easat-3242	7	23	on	on	ADP
easat-3242	7	24	unet++	unet++	PROPN
easat-3242	7	25	approach	approach	NOUN
easat-3242	7	26	and	and	CCONJ
easat-3242	7	27	data	datum	NOUN
easat-3242	7	28	route	route	NOUN
easat-3242	7	29	aggregation	aggregation	NOUN
easat-3242	7	30	module	module	NOUN
easat-3242	7	31	are	be	AUX
easat-3242	7	32	adopted	adopt	VERB
easat-3242	7	33	to	to	PART
easat-3242	7	34	provide	provide	VERB
easat-3242	7	35	complex	complex	ADJ
easat-3242	7	36	detail	detail	NOUN
easat-3242	7	37	in	in	ADP
easat-3242	7	38	satellite	satellite	NOUN
easat-3242	7	39	images	image	NOUN
easat-3242	7	40	and	and	CCONJ
easat-3242	7	41	contextual	contextual	ADJ
easat-3242	7	42	information	information	NOUN
easat-3242	7	43	.	.	PUNCT
easat-3242	8	1	implementing	implement	VERB
easat-3242	8	2	deep	deep	ADJ
easat-3242	8	3	network	network	NOUN
easat-3242	8	4	models	model	NOUN
easat-3242	8	5	using	use	VERB
easat-3242	8	6	known	known	ADJ
easat-3242	8	7	architectures	architecture	NOUN
easat-3242	8	8	will	will	AUX
easat-3242	8	9	help	help	VERB
easat-3242	8	10	increase	increase	VERB
easat-3242	8	11	accuracy	accuracy	NOUN
easat-3242	8	12	and	and	CCONJ
easat-3242	8	13	efficiency	efficiency	NOUN
easat-3242	8	14	in	in	ADP
easat-3242	8	15	situations	situation	NOUN
easat-3242	8	16	where	where	SCONJ
easat-3242	8	17	training	training	NOUN
easat-3242	8	18	datasets	dataset	NOUN
easat-3242	8	19	can	can	AUX
easat-3242	8	20	be	be	AUX
easat-3242	8	21	limited	limit	VERB
easat-3242	8	22	.	.	PUNCT
easat-3242	9	1	apparently	apparently	ADV
easat-3242	9	2	,	,	PUNCT
easat-3242	9	3	the	the	DET
easat-3242	9	4	approach	approach	NOUN
easat-3242	9	5	was	be	AUX
easat-3242	9	6	tested	test	VERB
easat-3242	9	7	and	and	CCONJ
easat-3242	9	8	bench	bench	NOUN
easat-3242	9	9	-	-	PUNCT
easat-3242	9	10	marked	mark	VERB
easat-3242	9	11	over	over	ADP
easat-3242	9	12	a	a	DET
easat-3242	9	13	set	set	NOUN
easat-3242	9	14	of	of	ADP
easat-3242	9	15	datasets	dataset	NOUN
easat-3242	9	16	for	for	ADP
easat-3242	9	17	several	several	ADJ
easat-3242	9	18	visual	visual	ADJ
easat-3242	9	19	contexts	context	NOUN
easat-3242	9	20	confirmed	confirm	VERB
easat-3242	9	21	by	by	ADP
easat-3242	9	22	extensive	extensive	ADJ
easat-3242	9	23	testing	testing	NOUN
easat-3242	9	24	which	which	PRON
easat-3242	9	25	resulted	result	VERB
easat-3242	9	26	in	in	ADP
easat-3242	9	27	an	an	DET
easat-3242	9	28	increased	increase	VERB
easat-3242	9	29	precision	precision	NOUN
easat-3242	9	30	,	,	PUNCT
easat-3242	9	31	recall	recall	VERB
easat-3242	9	32	along	along	ADP
easat-3242	9	33	with	with	ADP
easat-3242	9	34	f1	f1	PROPN
easat-3242	9	35	scores	score	NOUN
easat-3242	9	36	.	.	PUNCT
easat-3242	10	1	keywords	keyword	NOUN
easat-3242	10	2	:	:	PUNCT
easat-3242	10	3	deep	deep	ADJ
easat-3242	10	4	learning	learning	NOUN
easat-3242	10	5	,	,	PUNCT
easat-3242	10	6	satellite	satellite	NOUN
easat-3242	10	7	image	image	NOUN
easat-3242	10	8	segmentation	segmentation	NOUN
easat-3242	10	9	,	,	PUNCT
easat-3242	10	10	semantic	semantic	ADJ
easat-3242	10	11	segmentation	segmentation	NOUN
easat-3242	10	12	,	,	PUNCT
easat-3242	10	13	unet++	unet++	PROPN
easat-3242	10	14	,	,	PUNCT
easat-3242	10	15	vgg19	vgg19	PROPN
easat-3242	10	16	.	.	PUNCT
easat-3242	11	1	1	1	X
easat-3242	11	2	.	.	X
easat-3242	11	3	introduction	introduction	NOUN
easat-3242	11	4	understanding	understanding	NOUN
easat-3242	11	5	of	of	ADP
easat-3242	11	6	satellite	satellite	NOUN
easat-3242	11	7	image	image	NOUN
easat-3242	11	8	segmentation	segmentation	NOUN
easat-3242	11	9	is	be	AUX
easat-3242	11	10	essential	essential	ADJ
easat-3242	11	11	for	for	ADP
easat-3242	11	12	geospatial	geospatial	ADJ
easat-3242	11	13	data	datum	NOUN
easat-3242	11	14	analysis	analysis	NOUN
easat-3242	11	15	and	and	CCONJ
easat-3242	11	16	remote	remote	ADJ
easat-3242	11	17	sensing	sensing	NOUN
easat-3242	11	18	.	.	PUNCT
easat-3242	12	1	this	this	PRON
easat-3242	12	2	consists	consist	VERB
easat-3242	12	3	of	of	ADP
easat-3242	12	4	segmenting	segment	VERB
easat-3242	12	5	satellite	satellite	NOUN
easat-3242	12	6	images	image	NOUN
easat-3242	12	7	to	to	PART
easat-3242	12	8	break	break	VERB
easat-3242	12	9	them	they	PRON
easat-3242	12	10	down	down	ADP
easat-3242	12	11	into	into	ADP
easat-3242	12	12	contextual	contextual	ADJ
easat-3242	12	13	patches	patch	NOUN
easat-3242	12	14	that	that	PRON
easat-3242	12	15	correspond	correspond	VERB
easat-3242	12	16	to	to	ADP
easat-3242	12	17	a	a	DET
easat-3242	12	18	particular	particular	ADJ
easat-3242	12	19	type	type	NOUN
easat-3242	12	20	of	of	ADP
easat-3242	12	21	object	object	NOUN
easat-3242	12	22	in	in	ADP
easat-3242	12	23	the	the	DET
easat-3242	12	24	scene	scene	NOUN
easat-3242	13	1	[	[	X
easat-3242	13	2	1	1	NUM
easat-3242	13	3	]	]	PUNCT
easat-3242	13	4	.	.	PUNCT
easat-3242	14	1	applications	application	NOUN
easat-3242	14	2	of	of	ADP
easat-3242	14	3	this	this	DET
easat-3242	14	4	segmentation	segmentation	NOUN
easat-3242	14	5	process	process	NOUN
easat-3242	14	6	are	be	AUX
easat-3242	14	7	wide	wide	ADV
easat-3242	14	8	-	-	PUNCT
easat-3242	14	9	ranging	range	VERB
easat-3242	14	10	and	and	CCONJ
easat-3242	14	11	include	include	VERB
easat-3242	14	12	land	land	NOUN
easat-3242	14	13	cover	cover	NOUN
easat-3242	14	14	classification	classification	NOUN
easat-3242	14	15	,	,	PUNCT
easat-3242	14	16	urban	urban	ADJ
easat-3242	14	17	planning	planning	NOUN
easat-3242	14	18	,	,	PUNCT
easat-3242	14	19	disaster	disaster	NOUN
easat-3242	14	20	management	management	NOUN
easat-3242	14	21	and	and	CCONJ
easat-3242	14	22	environmental	environmental	ADJ
easat-3242	14	23	monitoring	monitoring	NOUN
easat-3242	14	24	[	[	X
easat-3242	14	25	2	2	NUM
easat-3242	14	26	]	]	PUNCT
easat-3242	14	27	.	.	PUNCT
easat-3242	15	1	ultimately	ultimately	ADV
easat-3242	15	2	,	,	PUNCT
easat-3242	15	3	our	our	PRON
easat-3242	15	4	goal	goal	NOUN
easat-3242	15	5	was	be	AUX
easat-3242	15	6	to	to	PART
easat-3242	15	7	automatically	automatically	ADV
easat-3242	15	8	read	read	VERB
easat-3242	15	9	satellite	satellite	NOUN
easat-3242	15	10	data	datum	NOUN
easat-3242	15	11	(	(	PUNCT
easat-3242	15	12	data	datum	NOUN
easat-3242	15	13	interpretation	interpretation	NOUN
easat-3242	15	14	and	and	CCONJ
easat-3242	15	15	analysis	analysis	NOUN
easat-3242	15	16	)	)	PUNCT
easat-3242	15	17	where	where	SCONJ
easat-3242	15	18	given	give	VERB
easat-3242	15	19	a	a	DET
easat-3242	15	20	image	image	NOUN
easat-3242	15	21	all	all	DET
easat-3242	15	22	the	the	DET
easat-3242	15	23	pixels	pixel	NOUN
easat-3242	15	24	are	be	AUX
easat-3242	15	25	classified	classify	VERB
easat-3242	15	26	into	into	ADP
easat-3242	15	27	certain	certain	ADJ
easat-3242	15	28	classes	class	NOUN
easat-3242	15	29	namely	namely	ADV
easat-3242	15	30	vegetation	vegetation	NOUN
easat-3242	15	31	,	,	PUNCT
easat-3242	15	32	buildings	building	NOUN
easat-3242	15	33	,	,	PUNCT
easat-3242	15	34	water	water	NOUN
easat-3242	15	35	bodies	body	NOUN
easat-3242	15	36	,	,	PUNCT
easat-3242	15	37	roads	road	NOUN
easat-3242	15	38	&	&	CCONJ
easat-3242	15	39	barren	barren	ADJ
easat-3242	15	40	land	land	NOUN
easat-3242	15	41	[	[	X
easat-3242	15	42	3	3	NUM
easat-3242	15	43	]	]	PUNCT
easat-3242	15	44	.	.	PUNCT
easat-3242	16	1	satellite	satellite	NOUN
easat-3242	16	2	image	image	NOUN
easat-3242	16	3	segmentation	segmentation	NOUN
easat-3242	16	4	,	,	PUNCT
easat-3242	16	5	as	as	ADP
easat-3242	16	6	the	the	DET
easat-3242	16	7	basic	basic	ADJ
easat-3242	16	8	component	component	NOUN
easat-3242	16	9	of	of	ADP
easat-3242	16	10	many	many	ADJ
easat-3242	16	11	fields	field	NOUN
easat-3242	16	12	relying	rely	VERB
easat-3242	16	13	on	on	ADP
easat-3242	16	14	geospatial	geospatial	ADJ
easat-3242	16	15	data	datum	NOUN
easat-3242	16	16	for	for	ADP
easat-3242	16	17	decision	decision	NOUN
easat-3242	16	18	-	-	PUNCT
easat-3242	16	19	making	make	VERB
easat-3242	16	20	purposes	purpose	NOUN
easat-3242	16	21	,	,	PUNCT
easat-3242	16	22	is	be	AUX
easat-3242	16	23	highly	highly	ADV
easat-3242	16	24	effective	effective	ADJ
easat-3242	16	25	in	in	ADP
easat-3242	16	26	various	various	ADJ
easat-3242	16	27	applications	application	NOUN
easat-3242	16	28	whose	whose	DET
easat-3242	16	29	precision	precision	NOUN
easat-3242	16	30	directly	directly	ADV
easat-3242	16	31	influences	influence	VERB
easat-3242	16	32	the	the	DET
easat-3242	16	33	overall	overall	ADJ
easat-3242	16	34	result	result	NOUN
easat-3242	16	35	.	.	PUNCT
easat-3242	17	1	ultimately	ultimately	ADV
easat-3242	17	2	,	,	PUNCT
easat-3242	17	3	our	our	PRON
easat-3242	17	4	goal	goal	NOUN
easat-3242	17	5	was	be	AUX
easat-3242	17	6	to	to	PART
easat-3242	17	7	automatically	automatically	ADV
easat-3242	17	8	read	read	VERB
easat-3242	17	9	satellite	satellite	NOUN
easat-3242	17	10	data	datum	NOUN
easat-3242	17	11	where	where	SCONJ
easat-3242	17	12	given	give	VERB
easat-3242	17	13	a	a	DET
easat-3242	17	14	image	image	NOUN
easat-3242	17	15	all	all	DET
easat-3242	17	16	the	the	DET
easat-3242	17	17	pixels	pixel	NOUN
easat-3242	17	18	are	be	AUX
easat-3242	17	19	classified	classify	VERB
easat-3242	17	20	into	into	ADP
easat-3242	17	21	certain	certain	ADJ
easat-3242	17	22	classes	class	NOUN
easat-3242	17	23	namely	namely	ADV
easat-3242	17	24	vegetation	vegetation	NOUN
easat-3242	17	25	,	,	PUNCT
easat-3242	17	26	buildings	building	NOUN
easat-3242	17	27	,	,	PUNCT
easat-3242	17	28	water	water	NOUN
easat-3242	17	29	bodies	body	NOUN
easat-3242	17	30	,	,	PUNCT
easat-3242	17	31	roads	road	NOUN
easat-3242	17	32	&	&	CCONJ
easat-3242	17	33	barren	barren	ADJ
easat-3242	17	34	land	land	NOUN
easat-3242	17	35	[	[	X
easat-3242	17	36	3	3	NUM
easat-3242	17	37	]	]	PUNCT
easat-3242	17	38	.	.	PUNCT
easat-3242	18	1	satellite	satellite	NOUN
easat-3242	18	2	image	image	NOUN
easat-3242	18	3	segmentation	segmentation	NOUN
easat-3242	18	4	,	,	PUNCT
easat-3242	18	5	as	as	ADP
easat-3242	18	6	the	the	DET
easat-3242	18	7	basic	basic	ADJ
easat-3242	18	8	component	component	NOUN
easat-3242	18	9	of	of	ADP
easat-3242	18	10	many	many	ADJ
easat-3242	18	11	fields	field	NOUN
easat-3242	18	12	relying	rely	VERB
easat-3242	18	13	on	on	ADP
easat-3242	18	14	geospatial	geospatial	ADJ
easat-3242	18	15	data	datum	NOUN
easat-3242	18	16	for	for	ADP
easat-3242	18	17	decision	decision	NOUN
easat-3242	18	18	-	-	PUNCT
easat-3242	18	19	making	make	VERB
easat-3242	18	20	purposes	purpose	NOUN
easat-3242	18	21	,	,	PUNCT
easat-3242	18	22	is	be	AUX
easat-3242	18	23	highly	highly	ADV
easat-3242	18	24	effective	effective	ADJ
easat-3242	18	25	in	in	ADP
easat-3242	18	26	various	various	ADJ
easat-3242	18	27	applications	application	NOUN
easat-3242	18	28	whose	whose	DET
easat-3242	18	29	precision	precision	NOUN
easat-3242	18	30	directly	directly	ADV
easat-3242	18	31	influences	influence	VERB
easat-3242	18	32	the	the	DET
easat-3242	18	33	overall	overall	ADJ
easat-3242	18	34	result	result	NOUN
easat-3242	18	35	.	.	PUNCT
easat-3242	19	1	this	this	PRON
easat-3242	19	2	is	be	AUX
easat-3242	19	3	a	a	DET
easat-3242	19	4	result	result	NOUN
easat-3242	19	5	of	of	ADP
easat-3242	19	6	differences	difference	NOUN
easat-3242	19	7	in	in	ADP
easat-3242	19	8	captured	capture	VERB
easat-3242	19	9	satellite	satellite	NOUN
easat-3242	19	10	images	image	NOUN
easat-3242	19	11	,	,	PUNCT
easat-3242	19	12	as	as	SCONJ
easat-3242	19	13	they	they	PRON
easat-3242	19	14	are	be	AUX
easat-3242	19	15	taken	take	VERB
easat-3242	19	16	at	at	ADP
easat-3242	19	17	different	different	ADJ
easat-3242	19	18	altitudes	altitude	NOUN
easat-3242	19	19	and	and	CCONJ
easat-3242	19	20	under	under	ADP
easat-3242	19	21	various	various	ADJ
easat-3242	19	22	environmental	environmental	ADJ
easat-3242	19	23	conditions	condition	NOUN
easat-3242	19	24	;	;	PUNCT
easat-3242	19	25	hence	hence	ADV
easat-3242	19	26	one	one	NUM
easat-3242	19	27	algorithmic	algorithmic	ADJ
easat-3242	19	28	approach	approach	NOUN
easat-3242	19	29	may	may	AUX
easat-3242	19	30	not	not	PART
easat-3242	19	31	work	work	VERB
easat-3242	19	32	well	well	ADV
easat-3242	19	33	across	across	ADP
easat-3242	19	34	all	all	DET
easat-3242	19	35	datasets	dataset	NOUN
easat-3242	19	36	.	.	PUNCT
easat-3242	20	1	this	this	PRON
easat-3242	20	2	also	also	ADV
easat-3242	20	3	leads	lead	VERB
easat-3242	20	4	to	to	ADP
easat-3242	20	5	the	the	DET
easat-3242	20	6	need	need	NOUN
easat-3242	20	7	for	for	ADP
easat-3242	20	8	powerful	powerful	ADJ
easat-3242	20	9	algorithms	algorithm	NOUN
easat-3242	20	10	that	that	PRON
easat-3242	20	11	work	work	VERB
easat-3242	20	12	with	with	ADP
easat-3242	20	13	a	a	DET
easat-3242	20	14	large	large	ADJ
easat-3242	20	15	input	input	NOUN
easat-3242	20	16	dataset	dataset	NOUN
easat-3242	20	17	,	,	PUNCT
easat-3242	20	18	which	which	PRON
easat-3242	20	19	can	can	AUX
easat-3242	20	20	truly	truly	ADV
easat-3242	20	21	represent	represent	VERB
easat-3242	20	22	slight	slight	ADJ
easat-3242	20	23	variations	variation	NOUN
easat-3242	20	24	in	in	ADP
easat-3242	20	25	pixel	pixel	PROPN
easat-3242	20	26	intensities	intensity	NOUN
easat-3242	20	27	.	.	PUNCT
easat-3242	21	1	one	one	NUM
easat-3242	21	2	of	of	ADP
easat-3242	21	3	the	the	DET
easat-3242	21	4	major	major	ADJ
easat-3242	21	5	limitations	limitation	NOUN
easat-3242	21	6	in	in	ADP
easat-3242	21	7	segmentation	segmentation	NOUN
easat-3242	21	8	of	of	ADP
easat-3242	21	9	satellite	satellite	NOUN
easat-3242	21	10	images	image	NOUN
easat-3242	21	11	is	be	AUX
easat-3242	21	12	their	their	PRON
easat-3242	21	13	inconsistent	inconsistent	ADJ
easat-3242	21	14	occurrence	occurrence	NOUN
easat-3242	21	15	.	.	PUNCT
easat-3242	22	1	there	there	PRON
easat-3242	22	2	is	be	VERB
easat-3242	22	3	high	high	ADJ
easat-3242	22	4	variation	variation	NOUN
easat-3242	22	5	of	of	ADP
easat-3242	22	6	images	image	NOUN
easat-3242	22	7	even	even	ADV
easat-3242	22	8	within	within	ADP
easat-3242	22	9	the	the	DET
easat-3242	22	10	same	same	ADJ
easat-3242	22	11	class	class	NOUN
easat-3242	22	12	due	due	ADP
easat-3242	22	13	to	to	ADP
easat-3242	22	14	unique	unique	ADJ
easat-3242	22	15	geographical	geographical	ADJ
easat-3242	22	16	features	feature	NOUN
easat-3242	22	17	,	,	PUNCT
easat-3242	22	18	weather	weather	NOUN
easat-3242	22	19	patterns	pattern	NOUN
easat-3242	22	20	or	or	CCONJ
easat-3242	22	21	just	just	ADV
easat-3242	22	22	different	different	ADJ
easat-3242	22	23	sensors	sensor	NOUN
easat-3242	22	24	used	use	VERB
easat-3242	22	25	[	[	X
easat-3242	22	26	5	5	NUM
easat-3242	22	27	]	]	PUNCT
easat-3242	22	28	.	.	PUNCT
easat-3242	23	1	this	this	DET
easat-3242	23	2	unpredictability	unpredictability	NOUN
easat-3242	23	3	has	have	VERB
easat-3242	23	4	implications	implication	NOUN
easat-3242	23	5	for	for	ADP
easat-3242	23	6	traditional	traditional	ADJ
easat-3242	23	7	image	image	NOUN
easat-3242	23	8	processing	processing	NOUN
easat-3242	23	9	techniques	technique	NOUN
easat-3242	23	10	,	,	PUNCT
easat-3242	23	11	which	which	PRON
easat-3242	23	12	require	require	VERB
easat-3242	23	13	hand	hand	NOUN
easat-3242	23	14	-	-	PUNCT
easat-3242	23	15	crafted	craft	VERB
easat-3242	23	16	rules	rule	NOUN
easat-3242	23	17	and	and	CCONJ
easat-3242	23	18	functions	function	NOUN
easat-3242	23	19	to	to	PART
easat-3242	23	20	specify	specify	VERB
easat-3242	23	21	the	the	DET
easat-3242	23	22	boundaries	boundary	NOUN
easat-3242	23	23	of	of	ADP
easat-3242	23	24	objects	object	NOUN
easat-3242	23	25	[	[	X
easat-3242	23	26	6	6	NUM
easat-3242	23	27	]	]	PUNCT
easat-3242	23	28	.	.	PUNCT
easat-3242	24	1	https://orcid.org/0009-0005-2443-9135	https://orcid.org/0009-0005-2443-9135	NOUN
easat-3242	24	2	5708	5708	NUM
easat-3242	24	3	edelweiss	edelweiss	PROPN
easat-3242	24	4	applied	apply	VERB
easat-3242	24	5	science	science	NOUN
easat-3242	24	6	and	and	CCONJ
easat-3242	24	7	technology	technology	NOUN
easat-3242	24	8	issn	issn	PROPN
easat-3242	24	9	:	:	PUNCT
easat-3242	24	10	2576	2576	NUM
easat-3242	24	11	-	-	SYM
easat-3242	24	12	8484	8484	NUM
easat-3242	24	13	vol	vol	NOUN
easat-3242	24	14	.	.	PROPN
easat-3242	25	1	8	8	NUM
easat-3242	25	2	,	,	PUNCT
easat-3242	25	3	no	no	INTJ
easat-3242	25	4	.	.	NOUN
easat-3242	26	1	6	6	NUM
easat-3242	26	2	:	:	SYM
easat-3242	26	3	5707	5707	NUM
easat-3242	26	4	-	-	SYM
easat-3242	26	5	5722	5722	NUM
easat-3242	26	6	,	,	PUNCT
easat-3242	26	7	2024	2024	NUM
easat-3242	26	8	doi	doi	NOUN
easat-3242	26	9	:	:	PUNCT
easat-3242	26	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	26	11	©	©	PROPN
easat-3242	26	12	2024	2024	NUM
easat-3242	26	13	by	by	ADP
easat-3242	26	14	the	the	DET
easat-3242	26	15	authors	author	NOUN
easat-3242	26	16	;	;	PUNCT
easat-3242	26	17	licensee	licensee	PROPN
easat-3242	26	18	learning	learning	NOUN
easat-3242	26	19	gate	gate	VERB
easat-3242	26	20	the	the	DET
easat-3242	26	21	other	other	ADJ
easat-3242	26	22	issue	issue	NOUN
easat-3242	26	23	that	that	PRON
easat-3242	26	24	is	be	AUX
easat-3242	26	25	seen	see	VERB
easat-3242	26	26	with	with	ADP
easat-3242	26	27	satellite	satellite	NOUN
easat-3242	26	28	image	image	NOUN
easat-3242	26	29	segmentation	segmentation	NOUN
easat-3242	26	30	is	be	AUX
easat-3242	26	31	the	the	DET
easat-3242	26	32	problem	problem	NOUN
easat-3242	26	33	of	of	ADP
easat-3242	26	34	handling	handle	VERB
easat-3242	26	35	so	so	ADV
easat-3242	26	36	much	much	ADJ
easat-3242	26	37	data	datum	NOUN
easat-3242	26	38	.	.	PUNCT
easat-3242	27	1	thousands	thousand	NOUN
easat-3242	27	2	or	or	CCONJ
easat-3242	27	3	more	more	ADJ
easat-3242	27	4	of	of	ADP
easat-3242	27	5	satellite	satellite	NOUN
easat-3242	27	6	images	image	NOUN
easat-3242	27	7	are	be	AUX
easat-3242	27	8	produced	produce	VERB
easat-3242	27	9	,	,	PUNCT
easat-3242	27	10	and	and	CCONJ
easat-3242	27	11	as	as	ADP
easat-3242	27	12	a	a	DET
easat-3242	27	13	time	time	NOUN
easat-3242	27	14	-	-	PUNCT
easat-3242	27	15	consuming	consume	VERB
easat-3242	27	16	task	task	NOUN
easat-3242	27	17	has	have	VERB
easat-3242	27	18	to	to	PART
easat-3242	27	19	be	be	AUX
easat-3242	27	20	manually	manually	ADV
easat-3242	27	21	labelled	label	VERB
easat-3242	27	22	for	for	ADP
easat-3242	27	23	training	training	NOUN
easat-3242	27	24	models	model	NOUN
easat-3242	27	25	in	in	ADP
easat-3242	27	26	machine	machine	NOUN
easat-3242	27	27	learning	learn	VERB
easat-3242	28	1	[	[	X
easat-3242	28	2	7	7	NUM
easat-3242	28	3	]	]	PUNCT
easat-3242	28	4	.	.	PUNCT
easat-3242	29	1	this	this	PRON
easat-3242	29	2	not	not	PART
easat-3242	29	3	only	only	ADV
easat-3242	29	4	makes	make	VERB
easat-3242	29	5	labelled	label	VERB
easat-3242	29	6	data	datum	NOUN
easat-3242	29	7	scarce	scarce	ADJ
easat-3242	29	8	,	,	PUNCT
easat-3242	29	9	an	an	DET
easat-3242	29	10	essential	essential	ADJ
easat-3242	29	11	ingredient	ingredient	NOUN
easat-3242	29	12	to	to	PART
easat-3242	29	13	train	train	VERB
easat-3242	29	14	regular	regular	ADJ
easat-3242	29	15	supervised	supervised	ADJ
easat-3242	29	16	machine	machine	NOUN
easat-3242	29	17	learning	learning	NOUN
easat-3242	29	18	models	model	NOUN
easat-3242	29	19	.	.	PUNCT
easat-3242	30	1	to	to	PART
easat-3242	30	2	address	address	VERB
easat-3242	30	3	this	this	DET
easat-3242	30	4	issue	issue	NOUN
easat-3242	30	5	,	,	PUNCT
easat-3242	30	6	automated	automated	ADJ
easat-3242	30	7	and	and	CCONJ
easat-3242	30	8	scalable	scalable	ADJ
easat-3242	30	9	solutions	solution	NOUN
easat-3242	30	10	should	should	AUX
easat-3242	30	11	be	be	AUX
easat-3242	30	12	used	use	VERB
easat-3242	30	13	to	to	PART
easat-3242	30	14	make	make	VERB
easat-3242	30	15	use	use	NOUN
easat-3242	30	16	of	of	ADP
easat-3242	30	17	the	the	DET
easat-3242	30	18	large	large	ADJ
easat-3242	30	19	amount	amount	NOUN
easat-3242	30	20	of	of	ADP
easat-3242	30	21	unlabeled	unlabeled	ADJ
easat-3242	30	22	experience	experience	NOUN
easat-3242	30	23	data	datum	NOUN
easat-3242	30	24	[	[	X
easat-3242	30	25	8	8	NUM
easat-3242	30	26	]	]	PUNCT
easat-3242	30	27	.	.	PUNCT
easat-3242	31	1	these	these	DET
easat-3242	31	2	works	work	NOUN
easat-3242	31	3	can	can	AUX
easat-3242	31	4	remedy	remedy	VERB
easat-3242	31	5	the	the	DET
easat-3242	31	6	production	production	NOUN
easat-3242	31	7	of	of	ADP
easat-3242	31	8	incomplete	incomplete	ADJ
easat-3242	31	9	pictures	picture	NOUN
easat-3242	31	10	and	and	CCONJ
easat-3242	31	11	improve	improve	VERB
easat-3242	31	12	segmentation	segmentation	NOUN
easat-3242	31	13	efficiency	efficiency	NOUN
easat-3242	31	14	in	in	ADP
easat-3242	31	15	novel	novel	NOUN
easat-3242	31	16	or	or	CCONJ
easat-3242	31	17	difficult	difficult	ADJ
easat-3242	31	18	situations	situation	NOUN
easat-3242	31	19	.	.	PUNCT
easat-3242	32	1	this	this	PRON
easat-3242	32	2	is	be	AUX
easat-3242	32	3	where	where	SCONJ
easat-3242	32	4	deep	deep	ADJ
easat-3242	32	5	learning	learning	NOUN
easat-3242	32	6	excels	excel	NOUN
easat-3242	32	7	in	in	ADP
easat-3242	32	8	solving	solve	VERB
easat-3242	32	9	the	the	DET
easat-3242	32	10	problem	problem	NOUN
easat-3242	32	11	of	of	ADP
easat-3242	32	12	satellite	satellite	NOUN
easat-3242	32	13	image	image	NOUN
easat-3242	32	14	segmentation	segmentation	NOUN
easat-3242	32	15	.	.	PUNCT
easat-3242	33	1	in	in	ADP
easat-3242	33	2	deep	deep	ADJ
easat-3242	33	3	learning	learning	NOUN
easat-3242	33	4	,	,	PUNCT
easat-3242	33	5	convolutional	convolutional	ADJ
easat-3242	33	6	neural	neural	ADJ
easat-3242	33	7	networks	network	NOUN
easat-3242	33	8	(	(	PUNCT
easat-3242	33	9	cnns	cnns	PROPN
easat-3242	33	10	)	)	PUNCT
easat-3242	33	11	have	have	AUX
easat-3242	33	12	demonstrated	demonstrate	VERB
easat-3242	33	13	impressive	impressive	ADJ
easat-3242	33	14	results	result	NOUN
easat-3242	33	15	by	by	ADP
easat-3242	33	16	being	be	AUX
easat-3242	33	17	able	able	ADJ
easat-3242	33	18	to	to	PART
easat-3242	33	19	learn	learn	VERB
easat-3242	33	20	hierarchical	hierarchical	ADJ
easat-3242	33	21	representations	representation	NOUN
easat-3242	33	22	from	from	ADP
easat-3242	33	23	visual	visual	ADJ
easat-3242	33	24	data	datum	NOUN
easat-3242	33	25	.	.	PUNCT
easat-3242	34	1	deep	deep	ADJ
easat-3242	34	2	learning	learning	NOUN
easat-3242	34	3	is	be	AUX
easat-3242	34	4	unique	unique	ADJ
easat-3242	34	5	among	among	ADP
easat-3242	34	6	traditional	traditional	ADJ
easat-3242	34	7	methods	method	NOUN
easat-3242	34	8	as	as	SCONJ
easat-3242	34	9	it	it	PRON
easat-3242	34	10	does	do	AUX
easat-3242	34	11	not	not	PART
easat-3242	34	12	require	require	VERB
easat-3242	34	13	complicated	complicated	ADJ
easat-3242	34	14	handmade	handmade	ADJ
easat-3242	34	15	feature	feature	NOUN
easat-3242	34	16	extraction	extraction	NOUN
easat-3242	34	17	process	process	NOUN
easat-3242	34	18	and	and	CCONJ
easat-3242	34	19	allows	allow	VERB
easat-3242	34	20	the	the	DET
easat-3242	34	21	network	network	NOUN
easat-3242	34	22	to	to	PART
easat-3242	34	23	learn	learn	VERB
easat-3242	34	24	discriminative	discriminative	NOUN
easat-3242	34	25	features	feature	NOUN
easat-3242	34	26	directly	directly	ADV
easat-3242	34	27	from	from	ADP
easat-3242	34	28	pixel	pixel	PROPN
easat-3242	34	29	intensities	intensity	NOUN
easat-3242	34	30	.	.	PUNCT
easat-3242	35	1	this	this	DET
easat-3242	35	2	article	article	NOUN
easat-3242	35	3	presents	present	VERB
easat-3242	35	4	a	a	DET
easat-3242	35	5	new	new	ADJ
easat-3242	35	6	fusion	fusion	NOUN
easat-3242	35	7	of	of	ADP
easat-3242	35	8	modern	modern	ADJ
easat-3242	35	9	deep	deep	ADJ
easat-3242	35	10	learning	learning	NOUN
easat-3242	35	11	frameworks	framework	NOUN
easat-3242	35	12	like	like	ADP
easat-3242	35	13	unet++	unet++	PRON
easat-3242	35	14	and	and	CCONJ
easat-3242	35	15	vgg19	vgg19	VERB
easat-3242	35	16	for	for	ADP
easat-3242	35	17	improving	improve	VERB
easat-3242	35	18	satellite	satellite	NOUN
easat-3242	35	19	image	image	NOUN
easat-3242	35	20	segmentation	segmentation	NOUN
easat-3242	35	21	in	in	ADP
easat-3242	35	22	this	this	DET
easat-3242	35	23	perspective	perspective	NOUN
easat-3242	35	24	.	.	PUNCT
easat-3242	36	1	the	the	DET
easat-3242	36	2	vgg19	vgg19	PROPN
easat-3242	36	3	model	model	NOUN
easat-3242	36	4	helps	help	VERB
easat-3242	36	5	boost	boost	VERB
easat-3242	36	6	accuracy	accuracy	NOUN
easat-3242	36	7	of	of	ADP
easat-3242	36	8	the	the	DET
easat-3242	36	9	system	system	NOUN
easat-3242	36	10	to	to	ADP
easat-3242	36	11	a	a	DET
easat-3242	36	12	good	good	ADJ
easat-3242	36	13	extent	extent	NOUN
easat-3242	36	14	in	in	ADP
easat-3242	36	15	settings	setting	NOUN
easat-3242	36	16	where	where	SCONJ
easat-3242	36	17	our	our	PRON
easat-3242	36	18	data	datum	NOUN
easat-3242	36	19	happens	happen	VERB
easat-3242	36	20	to	to	PART
easat-3242	36	21	be	be	AUX
easat-3242	36	22	less	less	ADJ
easat-3242	36	23	,	,	PUNCT
easat-3242	36	24	by	by	ADP
easat-3242	36	25	using	use	VERB
easat-3242	36	26	pre	pre	ADJ
easat-3242	36	27	-	-	ADJ
easat-3242	36	28	trained	train	VERB
easat-3242	36	29	weights	weight	NOUN
easat-3242	36	30	with	with	ADP
easat-3242	36	31	small	small	ADJ
easat-3242	36	32	input	input	NOUN
easat-3242	36	33	perturbations	perturbation	NOUN
easat-3242	36	34	for	for	ADP
easat-3242	36	35	noise	noise	NOUN
easat-3242	36	36	finding	find	VERB
easat-3242	36	37	neural	neural	ADJ
easat-3242	36	38	networks	network	NOUN
easat-3242	36	39	up	up	ADP
easat-3242	36	40	to	to	ADP
easat-3242	36	41	this	this	DET
easat-3242	36	42	point	point	NOUN
easat-3242	36	43	.	.	PUNCT
easat-3242	37	1	utilizing	utilize	VERB
easat-3242	37	2	this	this	DET
easat-3242	37	3	blend	blend	NOUN
easat-3242	37	4	of	of	ADP
easat-3242	37	5	unet++	unet++	PROPN
easat-3242	37	6	and	and	CCONJ
easat-3242	37	7	vgg19	vgg19	PROPN
easat-3242	37	8	is	be	AUX
easat-3242	37	9	a	a	DET
easat-3242	37	10	very	very	ADV
easat-3242	37	11	productive	productive	ADJ
easat-3242	37	12	way	way	NOUN
easat-3242	37	13	for	for	ADP
easat-3242	37	14	aerial	aerial	ADJ
easat-3242	37	15	image	image	NOUN
easat-3242	37	16	segmentation	segmentation	NOUN
easat-3242	37	17	accurately	accurately	ADV
easat-3242	37	18	.	.	PUNCT
easat-3242	38	1	2	2	X
easat-3242	38	2	.	.	X
easat-3242	38	3	literature	literature	NOUN
easat-3242	38	4	kunhao	kunhao	PROPN
easat-3242	38	5	yuan	yuan	PROPN
easat-3242	38	6	et	et	PROPN
easat-3242	38	7	al	al	PROPN
easat-3242	39	1	[	[	X
easat-3242	39	2	9	9	NUM
easat-3242	39	3	]	]	PUNCT
easat-3242	39	4	proposed	propose	VERB
easat-3242	39	5	an	an	DET
easat-3242	39	6	original	original	ADJ
easat-3242	39	7	dcnn	dcnn	ADJ
easat-3242	39	8	model	model	PROPN
easat-3242	39	9	mc	mc	PROPN
easat-3242	39	10	-	-	PUNCT
easat-3242	39	11	wbdn	wbdn	PROPN
easat-3242	39	12	(	(	PUNCT
easat-3242	39	13	multichannel	multichannel	NOUN
easat-3242	39	14	water	water	NOUN
easat-3242	39	15	body	body	NOUN
easat-3242	39	16	detection	detection	NOUN
easat-3242	39	17	network	network	NOUN
easat-3242	39	18	)	)	PUNCT
easat-3242	39	19	which	which	PRON
easat-3242	39	20	outperforms	outperform	VERB
easat-3242	39	21	the	the	DET
easat-3242	39	22	state	state	NOUN
easat-3242	39	23	-	-	PUNCT
easat-3242	39	24	of	of	ADP
easat-3242	39	25	-	-	PUNCT
easat-3242	39	26	the	the	DET
easat-3242	39	27	-	-	PUNCT
easat-3242	39	28	art	art	NOUN
easat-3242	39	29	dcnn	dcnn	NOUN
easat-3242	39	30	-	-	PUNCT
easat-3242	39	31	based	base	VERB
easat-3242	39	32	wbd	wbd	PROPN
easat-3242	39	33	techniques	technique	NOUN
easat-3242	39	34	by	by	ADP
easat-3242	39	35	utilizing	utilize	VERB
easat-3242	39	36	these	these	DET
easat-3242	39	37	three	three	NUM
easat-3242	39	38	advanced	advanced	ADJ
easat-3242	39	39	features	feature	NOUN
easat-3242	39	40	:	:	PUNCT
easat-3242	39	41	a	a	DET
easat-3242	39	42	multichannel	multichannel	ADJ
easat-3242	39	43	fusion	fusion	NOUN
easat-3242	39	44	module	module	NOUN
easat-3242	39	45	,	,	PUNCT
easat-3242	39	46	enhanced	enhance	VERB
easat-3242	39	47	atrous	atrous	ADJ
easat-3242	39	48	spatial	spatial	ADJ
easat-3242	39	49	pyramid	pyramid	NOUN
easat-3242	39	50	pooling	pooling	NOUN
easat-3242	39	51	and	and	CCONJ
easat-3242	39	52	space	space	NOUN
easat-3242	39	53	-	-	PUNCT
easat-3242	39	54	to	to	ADP
easat-3242	39	55	-	-	PUNCT
easat-3242	39	56	depth	depth	NOUN
easat-3242	39	57	/	/	SYM
easat-3242	39	58	depth	depth	NOUN
easat-3242	39	59	-	-	PUNCT
easat-3242	39	60	to	to	ADP
easat-3242	39	61	-	-	PUNCT
easat-3242	39	62	space	space	NOUN
easat-3242	39	63	operations	operation	NOUN
easat-3242	39	64	.	.	PUNCT
easat-3242	40	1	zhuokun	zhuokun	PROPN
easat-3242	40	2	pan	pan	PROPN
easat-3242	40	3	et	et	PROPN
easat-3242	40	4	al	al	PROPN
easat-3242	41	1	[	[	X
easat-3242	41	2	10	10	NUM
easat-3242	41	3	]	]	PUNCT
easat-3242	41	4	presented	present	VERB
easat-3242	41	5	the	the	DET
easat-3242	41	6	urban	urban	ADJ
easat-3242	41	7	village	village	NOUN
easat-3242	41	8	mapping	mapping	NOUN
easat-3242	41	9	paradigm	paradigm	NOUN
easat-3242	41	10	which	which	PRON
easat-3242	41	11	is	be	AUX
easat-3242	41	12	build	build	VERB
easat-3242	41	13	by	by	ADP
easat-3242	41	14	u	u	NOUN
easat-3242	41	15	-	-	ADJ
easat-3242	41	16	net	net	ADJ
easat-3242	41	17	deep	deep	ADJ
easat-3242	41	18	learning	learning	NOUN
easat-3242	41	19	architecture	architecture	NOUN
easat-3242	41	20	.	.	PUNCT
easat-3242	42	1	the	the	DET
easat-3242	42	2	study	study	NOUN
easat-3242	42	3	was	be	AUX
easat-3242	42	4	carried	carry	VERB
easat-3242	42	5	out	out	ADP
easat-3242	42	6	in	in	ADP
easat-3242	42	7	guangzhou	guangzhou	PROPN
easat-3242	42	8	city	city	PROPN
easat-3242	42	9	,	,	PUNCT
easat-3242	42	10	china	china	PROPN
easat-3242	42	11	.	.	PUNCT
easat-3242	43	1	for	for	ADP
easat-3242	43	2	this	this	DET
easat-3242	43	3	study	study	NOUN
easat-3242	43	4	,	,	PUNCT
easat-3242	43	5	worldview	worldview	PROPN
easat-3242	43	6	satellite	satellite	NOUN
easat-3242	43	7	imagery	imagery	NOUN
easat-3242	43	8	with	with	ADP
easat-3242	43	9	8	8	NUM
easat-3242	43	10	pan	pan	ADJ
easat-3242	43	11	-	-	ADJ
easat-3242	43	12	sharpened	sharpen	VERB
easat-3242	43	13	band	band	NOUN
easat-3242	43	14	is	be	AUX
easat-3242	43	15	employed	employ	VERB
easat-3242	43	16	at	at	ADP
easat-3242	43	17	spatial	spatial	ADJ
easat-3242	43	18	resolution	resolution	NOUN
easat-3242	43	19	of	of	ADP
easat-3242	43	20	0.5	0.5	NUM
easat-3242	43	21	m	m	NOUN
easat-3242	43	22	and	and	CCONJ
easat-3242	43	23	building	build	VERB
easat-3242	43	24	boundary	boundary	ADJ
easat-3242	43	25	vector	vector	NOUN
easat-3242	43	26	layer	layer	NOUN
easat-3242	43	27	used	use	VERB
easat-3242	43	28	in	in	ADP
easat-3242	43	29	the	the	DET
easat-3242	43	30	research	research	NOUN
easat-3242	43	31	process	process	NOUN
easat-3242	43	32	.	.	PUNCT
easat-3242	44	1	this	this	DET
easat-3242	44	2	world	world	NOUN
easat-3242	44	3	image	image	NOUN
easat-3242	44	4	scene	scene	NOUN
easat-3242	44	5	encompasses	encompass	VERB
easat-3242	44	6	ten	ten	NUM
easat-3242	44	7	sites	site	NOUN
easat-3242	44	8	of	of	ADP
easat-3242	44	9	urban	urban	ADJ
easat-3242	44	10	villages	village	NOUN
easat-3242	44	11	.	.	PUNCT
easat-3242	45	1	urban	urban	ADJ
easat-3242	45	2	village	village	NOUN
easat-3242	45	3	six	six	NUM
easat-3242	45	4	and	and	CCONJ
easat-3242	45	5	four	four	NUM
easat-3242	45	6	selected	select	VERB
easat-3242	45	7	sites	site	NOUN
easat-3242	45	8	were	be	AUX
easat-3242	45	9	subjected	subject	VERB
easat-3242	45	10	to	to	ADP
easat-3242	45	11	training	training	NOUN
easat-3242	45	12	and	and	CCONJ
easat-3242	45	13	testing	testing	NOUN
easat-3242	45	14	of	of	ADP
easat-3242	45	15	the	the	DET
easat-3242	45	16	deep	deep	ADJ
easat-3242	45	17	neural	neural	ADJ
easat-3242	45	18	network	network	NOUN
easat-3242	45	19	model	model	NOUN
easat-3242	45	20	.	.	PUNCT
easat-3242	46	1	both	both	PRON
easat-3242	46	2	building	build	VERB
easat-3242	46	3	segmentation	segmentation	NOUN
easat-3242	46	4	and	and	CCONJ
easat-3242	46	5	classification	classification	NOUN
easat-3242	46	6	models	model	NOUN
easat-3242	46	7	are	be	AUX
easat-3242	46	8	trained	train	VERB
easat-3242	46	9	and	and	CCONJ
easat-3242	46	10	tested	test	VERB
easat-3242	46	11	.	.	PUNCT
easat-3242	47	1	ramesh	ramesh	PROPN
easat-3242	47	2	raghavan	raghavan	PROPN
easat-3242	47	3	et	et	PROPN
easat-3242	47	4	al	al	PROPN
easat-3242	48	1	[	[	X
easat-3242	48	2	11	11	NUM
easat-3242	48	3	]	]	PUNCT
easat-3242	48	4	demonstrated	demonstrate	VERB
easat-3242	48	5	approach	approach	NOUN
easat-3242	48	6	enhances	enhance	VERB
easat-3242	48	7	the	the	DET
easat-3242	48	8	convolutional	convolutional	ADJ
easat-3242	48	9	neural	neural	ADJ
easat-3242	48	10	network	network	NOUN
easat-3242	48	11	for	for	ADP
easat-3242	48	12	pixel	pixel	ADJ
easat-3242	48	13	-	-	PUNCT
easat-3242	48	14	level	level	NOUN
easat-3242	48	15	image	image	NOUN
easat-3242	48	16	classification	classification	NOUN
easat-3242	48	17	.	.	PUNCT
easat-3242	49	1	meanwhile	meanwhile	ADV
easat-3242	49	2	,	,	PUNCT
easat-3242	49	3	the	the	PRON
easat-3242	49	4	mask	mask	NOUN
easat-3242	49	5	-	-	PUNCT
easat-3242	49	6	rcnn	rcnn	PROPN
easat-3242	49	7	(	(	PUNCT
easat-3242	49	8	mask	mask	NOUN
easat-3242	49	9	region	region	NOUN
easat-3242	49	10	-	-	PUNCT
easat-3242	49	11	based	base	VERB
easat-3242	49	12	convolutional	convolutional	ADJ
easat-3242	49	13	neural	neural	ADJ
easat-3242	49	14	networks	network	NOUN
easat-3242	49	15	)	)	PUNCT
easat-3242	49	16	performs	perform	VERB
easat-3242	49	17	image	image	NOUN
easat-3242	49	18	extraction	extraction	NOUN
easat-3242	49	19	and	and	CCONJ
easat-3242	49	20	the	the	DET
easat-3242	49	21	masking	masking	NOUN
easat-3242	49	22	problem	problem	NOUN
easat-3242	49	23	which	which	PRON
easat-3242	49	24	can	can	AUX
easat-3242	49	25	be	be	AUX
easat-3242	49	26	resolved	resolve	VERB
easat-3242	49	27	so	so	SCONJ
easat-3242	49	28	that	that	SCONJ
easat-3242	49	29	it	it	PRON
easat-3242	49	30	is	be	AUX
easat-3242	49	31	easier	easy	ADJ
easat-3242	49	32	and	and	CCONJ
easat-3242	49	33	faster	fast	ADV
easat-3242	49	34	in	in	ADP
easat-3242	49	35	preventing	prevent	VERB
easat-3242	49	36	this	this	DET
easat-3242	49	37	matter	matter	NOUN
easat-3242	49	38	.	.	PUNCT
easat-3242	50	1	it	it	PRON
easat-3242	50	2	involves	involve	VERB
easat-3242	50	3	training	train	VERB
easat-3242	50	4	the	the	DET
easat-3242	50	5	model	model	NOUN
easat-3242	50	6	with	with	ADP
easat-3242	50	7	a	a	DET
easat-3242	50	8	much	much	ADV
easat-3242	50	9	larger	large	ADJ
easat-3242	50	10	set	set	NOUN
easat-3242	50	11	of	of	ADP
easat-3242	50	12	data	datum	NOUN
easat-3242	50	13	points	point	NOUN
easat-3242	50	14	,	,	PUNCT
easat-3242	50	15	which	which	PRON
easat-3242	50	16	makes	make	VERB
easat-3242	50	17	it	it	PRON
easat-3242	50	18	more	more	ADV
easat-3242	50	19	complicated	complicated	ADJ
easat-3242	50	20	.	.	PUNCT
easat-3242	51	1	finally	finally	ADV
easat-3242	51	2	,	,	PUNCT
easat-3242	51	3	a	a	DET
easat-3242	51	4	sophisticated	sophisticated	ADJ
easat-3242	51	5	image	image	NOUN
easat-3242	51	6	augmentation	augmentation	NOUN
easat-3242	51	7	technique	technique	NOUN
easat-3242	51	8	has	have	AUX
easat-3242	51	9	been	be	AUX
easat-3242	51	10	added	add	VERB
easat-3242	51	11	in	in	ADP
easat-3242	51	12	the	the	DET
easat-3242	51	13	preprocessing	preprocessing	NOUN
easat-3242	51	14	to	to	PART
easat-3242	51	15	enhance	enhance	VERB
easat-3242	51	16	this	this	DET
easat-3242	51	17	algorithm	algorithm	NOUN
easat-3242	51	18	towards	towards	ADP
easat-3242	51	19	scalability	scalability	NOUN
easat-3242	51	20	.	.	PUNCT
easat-3242	52	1	jakub	jakub	PROPN
easat-3242	52	2	nalepa	nalepa	PROPN
easat-3242	52	3	et	et	PROPN
easat-3242	52	4	al	al	PROPN
easat-3242	53	1	[	[	X
easat-3242	53	2	12	12	NUM
easat-3242	53	3	]	]	PUNCT
easat-3242	53	4	provide	provide	VERB
easat-3242	53	5	a	a	DET
easat-3242	53	6	series	series	NOUN
easat-3242	53	7	of	of	ADP
easat-3242	53	8	simulated	simulate	VERB
easat-3242	53	9	scenarios	scenario	NOUN
easat-3242	53	10	based	base	VERB
easat-3242	53	11	on	on	ADP
easat-3242	53	12	different	different	ADJ
easat-3242	53	13	atmospheric	atmospheric	ADJ
easat-3242	53	14	conditions	condition	NOUN
easat-3242	53	15	and	and	CCONJ
easat-3242	53	16	noise	noise	NOUN
easat-3242	53	17	pollution	pollution	NOUN
easat-3242	53	18	that	that	PRON
easat-3242	53	19	might	might	AUX
easat-3242	53	20	occur	occur	VERB
easat-3242	53	21	in	in	ADP
easat-3242	53	22	any	any	DET
easat-3242	53	23	future	future	ADJ
easat-3242	53	24	imaging	imaging	NOUN
easat-3242	53	25	satellite	satellite	NOUN
easat-3242	53	26	.	.	PUNCT
easat-3242	54	1	in	in	ADP
easat-3242	54	2	this	this	DET
easat-3242	54	3	paper	paper	NOUN
easat-3242	54	4	,	,	PUNCT
easat-3242	54	5	we	we	PRON
easat-3242	54	6	also	also	ADV
easat-3242	54	7	demonstrate	demonstrate	VERB
easat-3242	54	8	the	the	DET
easat-3242	54	9	impact	impact	NOUN
easat-3242	54	10	of	of	ADP
easat-3242	54	11	these	these	DET
easat-3242	54	12	corrections	correction	NOUN
easat-3242	54	13	on	on	ADP
easat-3242	54	14	spectral	spectral	ADJ
easat-3242	54	15	-	-	PUNCT
easat-3242	54	16	spectral	spectral	ADJ
easat-3242	54	17	and	and	CCONJ
easat-3242	54	18	spectral	spectral	ADJ
easat-3242	54	19	-	-	PUNCT
easat-3242	54	20	spatial	spatial	ADJ
easat-3242	54	21	convolutional	convolutional	ADJ
easat-3242	54	22	neural	neural	ADJ
easat-3242	54	23	networks	network	NOUN
easat-3242	54	24	generalization	generalization	NOUN
easat-3242	54	25	performance	performance	NOUN
easat-3242	54	26	in	in	ADP
easat-3242	54	27	hyper	hyper	ADJ
easat-3242	54	28	-	-	ADJ
easat-3242	54	29	spectral	spectral	ADJ
easat-3242	54	30	image	image	NOUN
easat-3242	54	31	segmentation	segmentation	NOUN
easat-3242	54	32	.	.	PUNCT
easat-3242	55	1	thanh	thanh	PROPN
easat-3242	55	2	tam	tam	PROPN
easat-3242	55	3	nguyen	nguyen	PROPN
easat-3242	55	4	et	et	PROPN
easat-3242	55	5	al	al	PROPN
easat-3242	56	1	[	[	X
easat-3242	56	2	13	13	NUM
easat-3242	56	3	]	]	PUNCT
easat-3242	56	4	employed	employ	VERB
easat-3242	56	5	a	a	DET
easat-3242	56	6	spatio	spatio	NOUN
easat-3242	56	7	-	-	PUNCT
easat-3242	56	8	temporal	temporal	ADJ
easat-3242	56	9	-	-	PUNCT
easat-3242	56	10	spectral	spectral	ADJ
easat-3242	56	11	deep	deep	ADJ
easat-3242	56	12	neural	neural	ADJ
easat-3242	56	13	network	network	NOUN
easat-3242	56	14	of	of	ADP
easat-3242	56	15	high	high	ADJ
easat-3242	56	16	spatial	spatial	ADJ
easat-3242	56	17	resolution	resolution	NOUN
easat-3242	56	18	imagery	imagery	NOUN
easat-3242	56	19	at	at	ADP
easat-3242	56	20	the	the	DET
easat-3242	56	21	pixel	pixel	PROPN
easat-3242	56	22	level	level	NOUN
easat-3242	56	23	for	for	ADP
easat-3242	56	24	an	an	DET
easat-3242	56	25	entire	entire	ADJ
easat-3242	56	26	year	year	NOUN
easat-3242	56	27	and	and	CCONJ
easat-3242	56	28	each	each	DET
easat-3242	56	29	event	event	NOUN
easat-3242	56	30	in	in	ADP
easat-3242	56	31	time	time	NOUN
easat-3242	56	32	provides	provide	VERB
easat-3242	56	33	a	a	DET
easat-3242	56	34	novel	novel	ADJ
easat-3242	56	35	multi	multi	ADJ
easat-3242	56	36	-	-	ADJ
easat-3242	56	37	time	time	ADJ
easat-3242	56	38	high	high	ADJ
easat-3242	56	39	-	-	PUNCT
easat-3242	56	40	resolution	resolution	NOUN
easat-3242	56	41	classification	classification	NOUN
easat-3242	56	42	method	method	NOUN
easat-3242	56	43	to	to	PART
easat-3242	56	44	map	map	VERB
easat-3242	56	45	paddy	paddy	NOUN
easat-3242	56	46	fields	field	NOUN
easat-3242	56	47	.	.	PUNCT
easat-3242	57	1	the	the	DET
easat-3242	57	2	landsat	landsat	PROPN
easat-3242	57	3	8	8	NUM
easat-3242	57	4	data	datum	NOUN
easat-3242	57	5	is	be	AUX
easat-3242	57	6	used	use	VERB
easat-3242	57	7	for	for	ADP
easat-3242	57	8	our	our	PRON
easat-3242	57	9	case	case	NOUN
easat-3242	57	10	study	study	NOUN
easat-3242	57	11	as	as	SCONJ
easat-3242	57	12	it	it	PRON
easat-3242	57	13	has	have	VERB
easat-3242	57	14	exceptional	exceptional	ADJ
easat-3242	57	15	spatial	spatial	ADJ
easat-3242	57	16	resolution	resolution	NOUN
easat-3242	57	17	which	which	PRON
easat-3242	57	18	supports	support	VERB
easat-3242	57	19	development	development	NOUN
easat-3242	57	20	of	of	ADP
easat-3242	57	21	the	the	DET
easat-3242	57	22	technique	technique	NOUN
easat-3242	57	23	and	and	CCONJ
easat-3242	57	24	evaluation	evaluation	NOUN
easat-3242	57	25	.	.	PUNCT
easat-3242	58	1	sultan	sultan	PROPN
easat-3242	58	2	daud	daud	PROPN
easat-3242	58	3	khanet	khanet	PROPN
easat-3242	58	4	al	al	PROPN
easat-3242	59	1	[	[	X
easat-3242	59	2	14	14	NUM
easat-3242	59	3	]	]	PUNCT
easat-3242	59	4	designed	design	VERB
easat-3242	59	5	a	a	DET
easat-3242	59	6	novel	novel	ADJ
easat-3242	59	7	hybrid	hybrid	ADJ
easat-3242	59	8	deep	deep	ADJ
easat-3242	59	9	learning	learning	NOUN
easat-3242	59	10	model	model	NOUN
easat-3242	59	11	which	which	PRON
easat-3242	59	12	merges	merge	VERB
easat-3242	59	13	the	the	DET
easat-3242	59	14	characteristics	characteristic	NOUN
easat-3242	59	15	of	of	ADP
easat-3242	59	16	densenet	densenet	NOUN
easat-3242	59	17	and	and	CCONJ
easat-3242	59	18	u	u	NOUN
easat-3242	59	19	-	-	ADJ
easat-3242	59	20	net	net	ADJ
easat-3242	59	21	architectures	architecture	NOUN
easat-3242	59	22	.	.	PUNCT
easat-3242	60	1	this	this	DET
easat-3242	60	2	categorization	categorization	NOUN
easat-3242	60	3	is	be	AUX
easat-3242	60	4	carried	carry	VERB
easat-3242	60	5	out	out	ADP
easat-3242	60	6	to	to	ADP
easat-3242	60	7	pixelwise	pixelwise	ADJ
easat-3242	60	8	cover	cover	NOUN
easat-3242	60	9	land	land	NOUN
easat-3242	60	10	.	.	PUNCT
easat-3242	61	1	by	by	ADP
easat-3242	61	2	using	use	VERB
easat-3242	61	3	long	long	ADJ
easat-3242	61	4	-	-	PUNCT
easat-3242	61	5	range	range	NOUN
easat-3242	61	6	connections	connection	NOUN
easat-3242	61	7	of	of	ADP
easat-3242	61	8	u	u	NOUN
easat-3242	61	9	-	-	NOUN
easat-3242	61	10	net	net	ADJ
easat-3242	61	11	to	to	PART
easat-3242	61	12	concatenate	concatenate	VERB
easat-3242	61	13	the	the	DET
easat-3242	61	14	encoder	encoder	NOUN
easat-3242	61	15	and	and	CCONJ
easat-3242	61	16	decoder	decoder	NOUN
easat-3242	61	17	paths	path	NOUN
easat-3242	61	18	;	;	PUNCT
easat-3242	61	19	keeping	keep	VERB
easat-3242	61	20	some	some	DET
easat-3242	61	21	low	low	ADJ
easat-3242	61	22	-	-	PUNCT
easat-3242	61	23	level	level	NOUN
easat-3242	61	24	features	feature	NOUN
easat-3242	61	25	,	,	PUNCT
easat-3242	61	26	we	we	PRON
easat-3242	61	27	achieve	achieve	VERB
easat-3242	61	28	padding	padding	NOUN
easat-3242	61	29	information	information	NOUN
easat-3242	61	30	detection	detection	NOUN
easat-3242	61	31	results	result	NOUN
easat-3242	61	32	that	that	PRON
easat-3242	61	33	are	be	AUX
easat-3242	61	34	better	well	ADJ
easat-3242	61	35	than	than	ADP
easat-3242	61	36	existing	exist	VERB
easat-3242	61	37	methods	method	NOUN
easat-3242	61	38	.	.	PUNCT
easat-3242	62	1	5709	5709	NUM
easat-3242	62	2	edelweiss	edelweiss	PROPN
easat-3242	62	3	applied	apply	VERB
easat-3242	62	4	science	science	NOUN
easat-3242	62	5	and	and	CCONJ
easat-3242	62	6	technology	technology	NOUN
easat-3242	62	7	issn	issn	PROPN
easat-3242	62	8	:	:	PUNCT
easat-3242	62	9	2576	2576	NUM
easat-3242	62	10	-	-	SYM
easat-3242	62	11	8484	8484	NUM
easat-3242	62	12	vol	vol	NOUN
easat-3242	62	13	.	.	PROPN
easat-3242	62	14	8	8	NUM
easat-3242	62	15	,	,	PUNCT
easat-3242	62	16	no	no	INTJ
easat-3242	62	17	.	.	NOUN
easat-3242	63	1	6	6	NUM
easat-3242	63	2	:	:	SYM
easat-3242	63	3	5707	5707	NUM
easat-3242	63	4	-	-	SYM
easat-3242	63	5	5722	5722	NUM
easat-3242	63	6	,	,	PUNCT
easat-3242	63	7	2024	2024	NUM
easat-3242	63	8	doi	doi	NOUN
easat-3242	63	9	:	:	PUNCT
easat-3242	63	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	63	11	©	©	PROPN
easat-3242	63	12	2024	2024	NUM
easat-3242	63	13	by	by	ADP
easat-3242	63	14	the	the	DET
easat-3242	63	15	authors	author	NOUN
easat-3242	63	16	;	;	PUNCT
easat-3242	63	17	licensee	licensee	PROPN
easat-3242	63	18	learning	learning	NOUN
easat-3242	63	19	gate	gate	PROPN
easat-3242	63	20	bipulneupane	bipulneupane	PROPN
easat-3242	63	21	et	et	PROPN
easat-3242	63	22	al	al	PROPN
easat-3242	64	1	[	[	X
easat-3242	64	2	15	15	NUM
easat-3242	64	3	]	]	PUNCT
easat-3242	64	4	analyzed	analyze	VERB
easat-3242	64	5	and	and	CCONJ
easat-3242	64	6	perform	perform	VERB
easat-3242	64	7	meta	meta	NOUN
easat-3242	64	8	-	-	PUNCT
easat-3242	64	9	analyses	analysis	NOUN
easat-3242	64	10	of	of	ADP
easat-3242	64	11	the	the	DET
easat-3242	64	12	recent	recent	ADJ
easat-3242	64	13	publications	publication	NOUN
easat-3242	64	14	on	on	ADP
easat-3242	64	15	research	research	NOUN
easat-3242	64	16	questions	question	NOUN
easat-3242	64	17	,	,	PUNCT
easat-3242	64	18	data	datum	NOUN
easat-3242	64	19	sources	source	NOUN
easat-3242	64	20	pre	pre	ADJ
easat-3242	64	21	-	-	ADJ
easat-3242	64	22	processing	processing	ADJ
easat-3242	64	23	-	-	PUNCT
easat-3242	64	24	augmentation	augmentation	NOUN
easat-3242	64	25	techniques	technique	NOUN
easat-3242	64	26	training	training	NOUN
easat-3242	64	27	settings	setting	NOUN
easat-3242	64	28	architectures	architecture	NOUN
easat-3242	64	29	,	,	PUNCT
easat-3242	64	30	backbones	backbone	NOUN
easat-3242	64	31	,	,	PUNCT
easat-3242	64	32	frameworks	framework	NOUN
easat-3242	64	33	,	,	PUNCT
easat-3242	64	34	optimizers	optimizer	NOUN
easat-3242	64	35	,	,	PUNCT
easat-3242	64	36	loss	loss	NOUN
easat-3242	64	37	functions	function	NOUN
easat-3242	64	38	,	,	PUNCT
easat-3242	64	39	other	other	ADJ
easat-3242	64	40	hyper	hyper	ADJ
easat-3242	64	41	-	-	ADJ
easat-3242	64	42	parameters	parameter	NOUN
easat-3242	64	43	performance	performance	NOUN
easat-3242	64	44	measurement	measurement	NOUN
easat-3242	64	45	.	.	PUNCT
easat-3242	65	1	arsalan	arsalan	PROPN
easat-3242	65	2	tahir	tahir	PROPN
easat-3242	65	3	et	et	PROPN
easat-3242	65	4	al	al	PROPN
easat-3242	66	1	[	[	X
easat-3242	66	2	16	16	NUM
easat-3242	66	3	]	]	X
easat-3242	66	4	specific	specific	ADJ
easat-3242	66	5	to	to	ADP
easat-3242	66	6	assessing	assess	VERB
easat-3242	66	7	the	the	DET
easat-3242	66	8	state	state	NOUN
easat-3242	66	9	-	-	PUNCT
easat-3242	66	10	of	of	NOUN
easat-3242	66	11	-	-	PUNCT
easat-3242	66	12	play	play	NOUN
easat-3242	66	13	with	with	ADP
easat-3242	66	14	deep	deep	ADJ
easat-3242	66	15	learning	learning	NOUN
easat-3242	66	16	algorithms	algorithm	NOUN
easat-3242	66	17	at	at	ADP
easat-3242	66	18	recognising	recognise	VERB
easat-3242	66	19	objects	object	NOUN
easat-3242	66	20	within	within	ADP
easat-3242	66	21	satellite	satellite	NOUN
easat-3242	66	22	imagery	imagery	NOUN
easat-3242	66	23	.	.	PUNCT
easat-3242	67	1	thus	thus	ADV
easat-3242	67	2	a	a	DET
easat-3242	67	3	dataset	dataset	NOUN
easat-3242	67	4	of	of	ADP
easat-3242	67	5	satellite	satellite	NOUN
easat-3242	67	6	imagery	imagery	NOUN
easat-3242	67	7	is	be	AUX
easat-3242	67	8	made	make	VERB
easat-3242	67	9	using	use	VERB
easat-3242	67	10	the	the	DET
easat-3242	67	11	convolutional	convolutional	ADJ
easat-3242	67	12	neural	neural	ADJ
easat-3242	67	13	network	network	NOUN
easat-3242	67	14	based	base	VERB
easat-3242	67	15	framework	framework	NOUN
easat-3242	67	16	;	;	PUNCT
easat-3242	67	17	faster	fast	ADJ
easat-3242	67	18	rcnn	rcnn	PROPN
easat-3242	67	19	(	(	PUNCT
easat-3242	67	20	faster	fast	ADJ
easat-3242	67	21	region	region	NOUN
easat-3242	67	22	-	-	PUNCT
easat-3242	67	23	based	base	VERB
easat-3242	67	24	convolutional	convolutional	ADJ
easat-3242	67	25	neural	neural	ADJ
easat-3242	67	26	networks	network	NOUN
easat-3242	67	27	)	)	PUNCT
easat-3242	67	28	,	,	PUNCT
easat-3242	67	29	yolo	yolo	PROPN
easat-3242	67	30	,	,	PUNCT
easat-3242	67	31	ssd	ssd	PROPN
easat-3242	67	32	(	(	PUNCT
easat-3242	67	33	single	single	ADJ
easat-3242	67	34	shot	shot	NOUN
easat-3242	67	35	detector	detector	NOUN
easat-3242	67	36	)	)	PUNCT
easat-3242	67	37	and	and	CCONJ
easat-3242	67	38	other	other	ADJ
easat-3242	67	39	frameworks	framework	NOUN
easat-3242	67	40	like	like	ADP
easat-3242	67	41	simrdwn	simrdwn	NOUN
easat-3242	67	42	(	(	PUNCT
easat-3242	67	43	satellite	satellite	NOUN
easat-3242	67	44	imagery	imagery	NOUN
easat-3242	67	45	multiscale	multiscale	ADJ
easat-3242	67	46	rapid	rapid	ADJ
easat-3242	67	47	detection	detection	NOUN
easat-3242	67	48	with	with	ADP
easat-3242	67	49	windowed	windowed	ADJ
easat-3242	67	50	networks	network	NOUN
easat-3242	67	51	)	)	PUNCT
easat-3242	67	52	to	to	PART
easat-3242	67	53	do	do	AUX
easat-3242	67	54	object	object	VERB
easat-3242	67	55	detection	detection	NOUN
easat-3242	67	56	.	.	PUNCT
easat-3242	68	1	in	in	ADP
easat-3242	68	2	addition	addition	NOUN
easat-3242	68	3	,	,	PUNCT
easat-3242	68	4	we	we	PRON
easat-3242	68	5	use	use	VERB
easat-3242	68	6	the	the	DET
easat-3242	68	7	set	set	NOUN
easat-3242	68	8	of	of	ADP
easat-3242	68	9	satellite	satellite	NOUN
easat-3242	68	10	images	image	NOUN
easat-3242	68	11	created	create	VERB
easat-3242	68	12	to	to	PART
easat-3242	68	13	test	test	VERB
easat-3242	68	14	different	different	ADJ
easat-3242	68	15	blasting	blast	VERB
easat-3242	68	16	methods	method	NOUN
easat-3242	68	17	in	in	ADP
easat-3242	68	18	terms	term	NOUN
easat-3242	68	19	of	of	ADP
easat-3242	68	20	quality	quality	NOUN
easat-3242	68	21	and	and	CCONJ
easat-3242	68	22	speed	speed	NOUN
easat-3242	68	23	.	.	PUNCT
easat-3242	69	1	dmitry	dmitry	PROPN
easat-3242	69	2	rashkovetsky	rashkovetsky	PROPN
easat-3242	69	3	et	et	PROPN
easat-3242	69	4	al	al	PROPN
easat-3242	70	1	[	[	X
easat-3242	70	2	17	17	NUM
easat-3242	70	3	]	]	PUNCT
easat-3242	70	4	detailed	detail	VERB
easat-3242	70	5	a	a	DET
easat-3242	70	6	means	means	NOUN
easat-3242	70	7	to	to	PART
easat-3242	70	8	leverage	leverage	VERB
easat-3242	70	9	satellite	satellite	NOUN
easat-3242	70	10	imagery	imagery	NOUN
easat-3242	70	11	acquired	acquire	VERB
easat-3242	70	12	in	in	ADP
easat-3242	70	13	visible	visible	ADJ
easat-3242	70	14	,	,	PUNCT
easat-3242	70	15	infrared	infrared	ADJ
easat-3242	70	16	and	and	CCONJ
easat-3242	70	17	microwave	microwave	NOUN
easat-3242	70	18	regions	region	NOUN
easat-3242	70	19	of	of	ADP
easat-3242	70	20	the	the	DET
easat-3242	70	21	spectrum	spectrum	NOUN
easat-3242	70	22	for	for	ADP
easat-3242	70	23	differentiating	differentiate	VERB
easat-3242	70	24	fire	fire	NOUN
easat-3242	70	25	-	-	PUNCT
easat-3242	70	26	affected	affect	VERB
easat-3242	70	27	areas	area	NOUN
easat-3242	70	28	.	.	PUNCT
easat-3242	71	1	we	we	PRON
easat-3242	71	2	applied	apply	VERB
easat-3242	71	3	this	this	DET
easat-3242	71	4	workflow	workflow	NOUN
easat-3242	71	5	to	to	PART
easat-3242	71	6	assess	assess	VERB
easat-3242	71	7	the	the	DET
easat-3242	71	8	suitability	suitability	NOUN
easat-3242	71	9	for	for	ADP
easat-3242	71	10	fire	fire	NOUN
easat-3242	71	11	detection	detection	NOUN
easat-3242	71	12	in	in	ADP
easat-3242	71	13	four	four	NUM
easat-3242	71	14	open	open	ADJ
easat-3242	71	15	-	-	PUNCT
easat-3242	71	16	access	access	NOUN
easat-3242	71	17	satellite	satellite	NOUN
easat-3242	71	18	data	datum	NOUN
easat-3242	71	19	sources	source	NOUN
easat-3242	71	20	:	:	PUNCT
easat-3242	71	21	sentinel	sentinel	NOUN
easat-3242	71	22	1	1	NUM
easat-3242	71	23	(	(	PUNCT
easat-3242	71	24	c	c	NOUN
easat-3242	71	25	-	-	PUNCT
easat-3242	71	26	sar	sar	NOUN
easat-3242	71	27	)	)	PUNCT
easat-3242	71	28	;	;	PUNCT
easat-3242	71	29	sentinel-2	sentinel-2	NUM
easat-3242	71	30	multispectral	multispectral	ADJ
easat-3242	71	31	;	;	PUNCT
easat-3242	71	32	sentinel	sentinel	NOUN
easat-3242	71	33	-3	-3	PUNCT
easat-3242	71	34	sea	sea	NOUN
easat-3242	71	35	and	and	CCONJ
easat-3242	71	36	land	land	NOUN
easat-3242	71	37	surface	surface	NOUN
easat-3242	71	38	temperature	temperature	NOUN
easat-3242	71	39	;	;	PUNCT
easat-3242	71	40	terra	terra	NOUN
easat-3242	71	41	/	/	SYM
easat-3242	71	42	aqua	aqua	NOUN
easat-3242	71	43	modis	modis	PROPN
easat-3242	71	44	.	.	PUNCT
easat-3242	72	1	he	he	PRON
easat-3242	72	2	retrained	retrain	VERB
easat-3242	72	3	the	the	DET
easat-3242	72	4	same	same	ADJ
easat-3242	72	5	single	single	ADJ
easat-3242	72	6	-	-	PUNCT
easat-3242	72	7	input	input	NOUN
easat-3242	72	8	convolutional	convolutional	ADJ
easat-3242	72	9	neural	neural	ADJ
easat-3242	72	10	network	network	NOUN
easat-3242	72	11	on	on	ADP
easat-3242	72	12	a	a	DET
easat-3242	72	13	new	new	ADJ
easat-3242	72	14	dataset	dataset	NOUN
easat-3242	72	15	(	(	PUNCT
easat-3242	72	16	new	new	ADJ
easat-3242	72	17	data	datum	NOUN
easat-3242	72	18	)	)	PUNCT
easat-3242	72	19	for	for	ADP
easat-3242	72	20	each	each	PRON
easat-3242	72	21	of	of	ADP
easat-3242	72	22	these	these	PRON
easat-3242	72	23	before	before	ADP
easat-3242	72	24	testing	test	VERB
easat-3242	72	25	them	they	PRON
easat-3242	72	26	on	on	ADP
easat-3242	72	27	held	hold	VERB
easat-3242	72	28	-	-	PUNCT
easat-3242	72	29	out	out	ADP
easat-3242	72	30	test	test	NOUN
easat-3242	72	31	samples	sample	NOUN
easat-3242	72	32	.	.	PUNCT
easat-3242	73	1	ekrem	ekrem	VERB
easat-3242	73	2	saralioglu	saralioglu	PROPN
easat-3242	73	3	et	et	PROPN
easat-3242	73	4	al	al	PROPN
easat-3242	74	1	[	[	X
easat-3242	74	2	18	18	NUM
easat-3242	74	3	]	]	PUNCT
easat-3242	74	4	employed	employ	VERB
easat-3242	74	5	a	a	DET
easat-3242	74	6	3d-2d	3d-2d	NUM
easat-3242	74	7	cnn	cnn	PROPN
easat-3242	74	8	network	network	NOUN
easat-3242	74	9	,	,	PUNCT
easat-3242	74	10	which	which	PRON
easat-3242	74	11	incorporates	incorporate	VERB
easat-3242	74	12	both	both	DET
easat-3242	74	13	spectral	spectral	ADJ
easat-3242	74	14	and	and	CCONJ
easat-3242	74	15	spatial	spatial	ADJ
easat-3242	74	16	information	information	NOUN
easat-3242	74	17	to	to	PART
easat-3242	74	18	help	help	VERB
easat-3242	74	19	generate	generate	VERB
easat-3242	74	20	accurate	accurate	ADJ
easat-3242	74	21	land	land	NOUN
easat-3242	74	22	cover	cover	NOUN
easat-3242	74	23	maps	map	NOUN
easat-3242	74	24	out	out	ADP
easat-3242	74	25	of	of	ADP
easat-3242	74	26	gsd	gsd	NOUN
easat-3242	74	27	satellite	satellite	NOUN
easat-3242	74	28	images	image	NOUN
easat-3242	74	29	.	.	PUNCT
easat-3242	75	1	based	base	VERB
easat-3242	75	2	on	on	ADP
easat-3242	75	3	the	the	DET
easat-3242	75	4	sh	sh	PROPN
easat-3242	75	5	model	model	NOUN
easat-3242	75	6	for	for	ADP
easat-3242	75	7	landscape	landscape	NOUN
easat-3242	75	8	,	,	PUNCT
easat-3242	75	9	a	a	DET
easat-3242	75	10	worldview-2	worldview-2	ADJ
easat-3242	75	11	satellite	satellite	NOUN
easat-3242	75	12	image	image	NOUN
easat-3242	75	13	was	be	AUX
easat-3242	75	14	used	use	VERB
easat-3242	75	15	that	that	SCONJ
easat-3242	75	16	incorporated	incorporate	VERB
easat-3242	75	17	9	9	NUM
easat-3242	75	18	land	land	NOUN
easat-3242	75	19	use	use	NOUN
easat-3242	75	20	categories	category	NOUN
easat-3242	75	21	(	(	PUNCT
easat-3242	75	22	houses	house	NOUN
easat-3242	75	23	and	and	CCONJ
easat-3242	75	24	highways	highway	NOUN
easat-3242	75	25	)	)	PUNCT
easat-3242	75	26	in	in	ADP
easat-3242	75	27	addition	addition	NOUN
easat-3242	75	28	to	to	ADP
easat-3242	75	29	orls	orl	NOUN
easat-3242	75	30	depicting	depict	VERB
easat-3242	75	31	product	product	NOUN
easat-3242	75	32	agricultural	agricultural	ADJ
easat-3242	75	33	areas	area	NOUN
easat-3242	75	34	along	along	ADP
easat-3242	75	35	with	with	ADP
easat-3242	75	36	tea	tea	NOUN
easat-3242	75	37	farms	farm	NOUN
easat-3242	75	38	as	as	ADV
easat-3242	75	39	well	well	ADV
easat-3242	75	40	as	as	ADP
easat-3242	75	41	hazelnut	hazelnut	NOUN
easat-3242	75	42	plantations	plantation	NOUN
easat-3242	75	43	.	.	PUNCT
easat-3242	76	1	3	3	X
easat-3242	76	2	.	.	NUM
easat-3242	76	3	proposed	propose	VERB
easat-3242	76	4	method	method	NOUN
easat-3242	76	5	this	this	DET
easat-3242	76	6	proposed	propose	VERB
easat-3242	76	7	methodology	methodology	NOUN
easat-3242	76	8	is	be	AUX
easat-3242	76	9	based	base	VERB
easat-3242	76	10	on	on	ADP
easat-3242	76	11	the	the	DET
easat-3242	76	12	integration	integration	NOUN
easat-3242	76	13	of	of	ADP
easat-3242	76	14	vgg19	vgg19	PROPN
easat-3242	76	15	deep	deep	ADJ
easat-3242	76	16	learning	learning	NOUN
easat-3242	76	17	model	model	NOUN
easat-3242	76	18	with	with	ADP
easat-3242	76	19	unet++	unet++	PROPN
easat-3242	76	20	framework	framework	NOUN
easat-3242	76	21	in	in	ADP
easat-3242	76	22	order	order	NOUN
easat-3242	76	23	to	to	PART
easat-3242	76	24	segment	segment	VERB
easat-3242	76	25	satellite	satellite	NOUN
easat-3242	76	26	images	image	NOUN
easat-3242	76	27	precisely	precisely	ADV
easat-3242	76	28	.	.	PUNCT
easat-3242	77	1	it	it	PRON
easat-3242	77	2	combines	combine	VERB
easat-3242	77	3	vgg19	vgg19	PROPN
easat-3242	77	4	and	and	CCONJ
easat-3242	77	5	unet++	unet++	PROPN
easat-3242	77	6	,	,	PUNCT
easat-3242	77	7	two	two	NUM
easat-3242	77	8	state	state	NOUN
easat-3242	77	9	-	-	PUNCT
easat-3242	77	10	of	of	ADP
easat-3242	77	11	-	-	PUNCT
easat-3242	77	12	the	the	DET
easat-3242	77	13	-	-	PUNCT
easat-3242	77	14	art	art	NOUN
easat-3242	77	15	deep	deep	ADJ
easat-3242	77	16	-	-	PUNCT
easat-3242	77	17	learning	learn	VERB
easat-3242	77	18	frameworks	framework	NOUN
easat-3242	77	19	to	to	PART
easat-3242	77	20	deal	deal	VERB
easat-3242	77	21	with	with	ADP
easat-3242	77	22	the	the	DET
easat-3242	77	23	intricacies	intricacy	NOUN
easat-3242	77	24	of	of	ADP
easat-3242	77	25	satellite	satellite	NOUN
easat-3242	77	26	imagery	imagery	NOUN
easat-3242	77	27	.	.	PUNCT
easat-3242	78	1	the	the	DET
easat-3242	78	2	critical	critical	ADJ
easat-3242	78	3	issues	issue	NOUN
easat-3242	78	4	include	include	VERB
easat-3242	78	5	the	the	DET
easat-3242	78	6	presence	presence	NOUN
easat-3242	78	7	of	of	ADP
easat-3242	78	8	diverse	diverse	ADJ
easat-3242	78	9	lighting	lighting	NOUN
easat-3242	78	10	conditions	condition	NOUN
easat-3242	78	11	,	,	PUNCT
easat-3242	78	12	different	different	ADJ
easat-3242	78	13	terrains	terrain	NOUN
easat-3242	78	14	and	and	CCONJ
easat-3242	78	15	required	require	VERB
easat-3242	78	16	accuracies	accuracy	NOUN
easat-3242	78	17	for	for	ADP
easat-3242	78	18	earth	earth	NOUN
easat-3242	78	19	surface	surface	NOUN
easat-3242	78	20	structures	structure	NOUN
easat-3242	78	21	extractions	extraction	VERB
easat-3242	78	22	.	.	PUNCT
easat-3242	79	1	the	the	DET
easat-3242	79	2	method	method	NOUN
easat-3242	79	3	exploits	exploit	VERB
easat-3242	79	4	these	these	DET
easat-3242	79	5	architectures	architecture	NOUN
easat-3242	79	6	to	to	PART
easat-3242	79	7	improve	improve	VERB
easat-3242	79	8	the	the	DET
easat-3242	79	9	performance	performance	NOUN
easat-3242	79	10	of	of	ADP
easat-3242	79	11	impact	impact	NOUN
easat-3242	79	12	segmentation	segmentation	NOUN
easat-3242	79	13	,	,	PUNCT
easat-3242	79	14	which	which	PRON
easat-3242	79	15	is	be	AUX
easat-3242	79	16	an	an	DET
easat-3242	79	17	early	early	ADJ
easat-3242	79	18	developmental	developmental	ADJ
easat-3242	79	19	stage	stage	NOUN
easat-3242	79	20	that	that	PRON
easat-3242	79	21	has	have	AUX
easat-3242	79	22	so	so	ADV
easat-3242	79	23	far	far	ADV
easat-3242	79	24	proven	prove	VERB
easat-3242	79	25	challenging	challenge	VERB
easat-3242	79	26	as	as	SCONJ
easat-3242	79	27	it	it	PRON
easat-3242	79	28	exhibits	exhibit	VERB
easat-3242	79	29	high	high	ADJ
easat-3242	79	30	detail	detail	NOUN
easat-3242	79	31	in	in	ADP
easat-3242	79	32	satellite	satellite	NOUN
easat-3242	79	33	imagery	imagery	NOUN
easat-3242	79	34	.	.	PUNCT
easat-3242	80	1	3.1	3.1	NUM
easat-3242	80	2	u	u	NOUN
easat-3242	80	3	-	-	PROPN
easat-3242	80	4	net++	net++	VERB
easat-3242	80	5	unet++	unet++	PROPN
easat-3242	80	6	,	,	PUNCT
easat-3242	80	7	also	also	ADV
easat-3242	80	8	called	call	VERB
easat-3242	80	9	unet	unet	NOUN
easat-3242	80	10	+	+	NOUN
easat-3242	80	11	+	+	NUM
easat-3242	80	12	or	or	CCONJ
easat-3242	80	13	nested	nest	VERB
easat-3242	80	14	u	u	NOUN
easat-3242	80	15	-	-	NOUN
easat-3242	80	16	net	net	ADJ
easat-3242	80	17	,	,	PUNCT
easat-3242	80	18	is	be	AUX
easat-3242	80	19	an	an	DET
easat-3242	80	20	improved	improved	ADJ
easat-3242	80	21	version	version	NOUN
easat-3242	80	22	of	of	ADP
easat-3242	80	23	the	the	DET
easat-3242	80	24	original	original	ADJ
easat-3242	80	25	architecture	architecture	NOUN
easat-3242	80	26	which	which	PRON
easat-3242	80	27	uses	use	VERB
easat-3242	80	28	more	more	ADJ
easat-3242	80	29	complex	complex	ADJ
easat-3242	80	30	segmentation	segmentation	NOUN
easat-3242	80	31	tasks	task	NOUN
easat-3242	80	32	.	.	PUNCT
easat-3242	81	1	it	it	PRON
easat-3242	81	2	is	be	AUX
easat-3242	81	3	able	able	ADJ
easat-3242	81	4	to	to	PART
easat-3242	81	5	keep	keep	VERB
easat-3242	81	6	the	the	DET
easat-3242	81	7	basic	basic	ADJ
easat-3242	81	8	ideas	idea	NOUN
easat-3242	81	9	of	of	ADP
easat-3242	81	10	u	u	NOUN
easat-3242	81	11	-	-	NOUN
easat-3242	81	12	net	net	ADJ
easat-3242	81	13	and	and	CCONJ
easat-3242	81	14	improve	improve	VERB
easat-3242	81	15	in	in	ADP
easat-3242	81	16	various	various	ADJ
easat-3242	81	17	forms	form	NOUN
easat-3242	81	18	.	.	PUNCT
easat-3242	82	1	moreover	moreover	ADV
easat-3242	82	2	,	,	PUNCT
easat-3242	82	3	the	the	DET
easat-3242	82	4	proposed	propose	VERB
easat-3242	82	5	u	u	NOUN
easat-3242	82	6	-	-	NOUN
easat-3242	82	7	net++	net++	ADJ
easat-3242	82	8	alterations	alteration	NOUN
easat-3242	82	9	are	be	AUX
easat-3242	82	10	practical	practical	ADJ
easat-3242	82	11	and	and	CCONJ
easat-3242	82	12	can	can	AUX
easat-3242	82	13	tune	tune	NOUN
easat-3242	82	14	-	-	PUNCT
easat-3242	82	15	in	in	ADP
easat-3242	82	16	better	well	ADV
easat-3242	82	17	to	to	AUX
easat-3242	82	18	complex	complex	ADJ
easat-3242	82	19	segmentation	segmentation	NOUN
easat-3242	82	20	problems	problem	NOUN
easat-3242	82	21	especially	especially	ADV
easat-3242	82	22	in	in	ADP
easat-3242	82	23	medical	medical	ADJ
easat-3242	82	24	images	image	NOUN
easat-3242	82	25	type	type	NOUN
easat-3242	82	26	for	for	ADP
easat-3242	82	27	critical	critical	ADJ
easat-3242	82	28	areas	area	NOUN
easat-3242	82	29	.	.	PUNCT
easat-3242	83	1	the	the	DET
easat-3242	83	2	architecture	architecture	NOUN
easat-3242	83	3	now	now	ADV
easat-3242	83	4	includes	include	VERB
easat-3242	83	5	extra	extra	ADJ
easat-3242	83	6	pathways	pathway	NOUN
easat-3242	83	7	and	and	CCONJ
easat-3242	83	8	dense	dense	ADJ
easat-3242	83	9	connections	connection	NOUN
easat-3242	83	10	,	,	PUNCT
easat-3242	83	11	which	which	PRON
easat-3242	83	12	help	help	VERB
easat-3242	83	13	improve	improve	VERB
easat-3242	83	14	feature	feature	NOUN
easat-3242	83	15	extraction	extraction	NOUN
easat-3242	83	16	as	as	ADV
easat-3242	83	17	well	well	ADV
easat-3242	83	18	provide	provide	VERB
easat-3242	83	19	better	well	ADJ
easat-3242	83	20	segmentation	segmentation	NOUN
easat-3242	83	21	results	result	NOUN
easat-3242	83	22	in	in	ADP
easat-3242	83	23	non	non	ADJ
easat-3242	83	24	-	-	ADJ
easat-3242	83	25	homogeneous	homogeneous	ADJ
easat-3242	83	26	space	space	NOUN
easat-3242	83	27	.	.	PUNCT
easat-3242	84	1	5710	5710	NUM
easat-3242	84	2	edelweiss	edelweiss	PROPN
easat-3242	84	3	applied	apply	VERB
easat-3242	84	4	science	science	NOUN
easat-3242	84	5	and	and	CCONJ
easat-3242	84	6	technology	technology	NOUN
easat-3242	84	7	issn	issn	PROPN
easat-3242	84	8	:	:	PUNCT
easat-3242	84	9	2576	2576	NUM
easat-3242	84	10	-	-	SYM
easat-3242	84	11	8484	8484	NUM
easat-3242	84	12	vol	vol	NOUN
easat-3242	84	13	.	.	PROPN
easat-3242	84	14	8	8	NUM
easat-3242	84	15	,	,	PUNCT
easat-3242	84	16	no	no	INTJ
easat-3242	84	17	.	.	NOUN
easat-3242	84	18	6	6	NUM
easat-3242	84	19	:	:	SYM
easat-3242	84	20	5707	5707	NUM
easat-3242	84	21	-	-	SYM
easat-3242	84	22	5722	5722	NUM
easat-3242	84	23	,	,	PUNCT
easat-3242	84	24	2024	2024	NUM
easat-3242	84	25	doi	doi	NOUN
easat-3242	84	26	:	:	PUNCT
easat-3242	84	27	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	84	28	©	©	PROPN
easat-3242	84	29	2024	2024	NUM
easat-3242	84	30	by	by	ADP
easat-3242	84	31	the	the	DET
easat-3242	84	32	authors	author	NOUN
easat-3242	84	33	;	;	PUNCT
easat-3242	84	34	licensee	licensee	PROPN
easat-3242	84	35	learning	learn	VERB
easat-3242	84	36	gate	gate	NOUN
easat-3242	84	37	figure	figure	NOUN
easat-3242	84	38	1	1	NUM
easat-3242	84	39	.	.	PUNCT
easat-3242	84	40	general	general	ADJ
easat-3242	84	41	architecture	architecture	NOUN
easat-3242	84	42	of	of	ADP
easat-3242	84	43	unet++	unet++	PROPN
easat-3242	84	44	.	.	PUNCT
easat-3242	85	1	u	u	X
easat-3242	85	2	-	-	PUNCT
easat-3242	85	3	net++	net++	PROPN
easat-3242	85	4	still	still	ADV
easat-3242	85	5	sticks	stick	VERB
easat-3242	85	6	to	to	ADP
easat-3242	85	7	the	the	DET
easat-3242	85	8	core	core	NOUN
easat-3242	85	9	encoder	encoder	NOUN
easat-3242	85	10	-	-	PUNCT
easat-3242	85	11	decoder	decoder	NOUN
easat-3242	85	12	architecture	architecture	NOUN
easat-3242	85	13	of	of	ADP
easat-3242	85	14	u	u	NOUN
easat-3242	85	15	-	-	NOUN
easat-3242	85	16	net	net	ADJ
easat-3242	85	17	and	and	CCONJ
easat-3242	85	18	makes	make	VERB
easat-3242	85	19	modifications	modification	NOUN
easat-3242	85	20	for	for	ADP
easat-3242	85	21	better	well	ADJ
easat-3242	85	22	segmentation	segmentation	NOUN
easat-3242	85	23	performance	performance	NOUN
easat-3242	85	24	.	.	PUNCT
easat-3242	86	1	to	to	ADP
easat-3242	86	2	this	this	DET
easat-3242	86	3	end	end	NOUN
easat-3242	86	4	,	,	PUNCT
easat-3242	86	5	a	a	DET
easat-3242	86	6	major	major	ADJ
easat-3242	86	7	change	change	NOUN
easat-3242	86	8	is	be	AUX
easat-3242	86	9	the	the	DET
easat-3242	86	10	use	use	NOUN
easat-3242	86	11	of	of	ADP
easat-3242	86	12	multi	multi	ADJ
easat-3242	86	13	-	-	ADJ
easat-3242	86	14	scale	scale	ADJ
easat-3242	86	15	skip	skip	ADJ
easat-3242	86	16	connections	connection	NOUN
easat-3242	86	17	that	that	PRON
easat-3242	86	18	help	help	VERB
easat-3242	86	19	capturing	capture	VERB
easat-3242	86	20	local	local	ADJ
easat-3242	86	21	and	and	CCONJ
easat-3242	86	22	global	global	ADJ
easat-3242	86	23	context	context	NOUN
easat-3242	86	24	more	more	ADV
easat-3242	86	25	robustly	robustly	ADV
easat-3242	86	26	.	.	PUNCT
easat-3242	87	1	this	this	PRON
easat-3242	87	2	enables	enable	VERB
easat-3242	87	3	u	u	NOUN
easat-3242	87	4	-	-	NOUN
easat-3242	87	5	net++	net++	PROPN
easat-3242	87	6	to	to	PART
easat-3242	87	7	carry	carry	VERB
easat-3242	87	8	out	out	ADP
easat-3242	87	9	more	more	ADV
easat-3242	87	10	precise	precise	ADJ
easat-3242	87	11	segmentations	segmentation	NOUN
easat-3242	87	12	by	by	ADP
easat-3242	87	13	taking	take	VERB
easat-3242	87	14	both	both	DET
easat-3242	87	15	fine	fine	ADJ
easat-3242	87	16	details	detail	NOUN
easat-3242	87	17	and	and	CCONJ
easat-3242	87	18	larger	large	ADJ
easat-3242	87	19	imaging	imaging	NOUN
easat-3242	87	20	context	context	NOUN
easat-3242	87	21	into	into	ADP
easat-3242	87	22	account	account	NOUN
easat-3242	87	23	.	.	PUNCT
easat-3242	88	1	the	the	DET
easat-3242	88	2	encoder	encoder	NOUN
easat-3242	88	3	is	be	AUX
easat-3242	88	4	responsible	responsible	ADJ
easat-3242	88	5	for	for	ADP
easat-3242	88	6	down	down	ADV
easat-3242	88	7	-	-	PUNCT
easat-3242	88	8	scaling	scale	VERB
easat-3242	88	9	the	the	DET
easat-3242	88	10	input	input	NOUN
easat-3242	88	11	image	image	NOUN
easat-3242	88	12	using	use	VERB
easat-3242	88	13	convolutional	convolutional	ADJ
easat-3242	88	14	and	and	CCONJ
easat-3242	88	15	pooling	pool	VERB
easat-3242	88	16	layers	layer	NOUN
easat-3242	88	17	to	to	PART
easat-3242	88	18	extract	extract	VERB
easat-3242	88	19	hierarchical	hierarchical	ADJ
easat-3242	88	20	features	feature	NOUN
easat-3242	88	21	across	across	ADP
easat-3242	88	22	various	various	ADJ
easat-3242	88	23	levels	level	NOUN
easat-3242	88	24	.	.	PUNCT
easat-3242	89	1	the	the	DET
easat-3242	89	2	decoder	decoder	NOUN
easat-3242	89	3	then	then	ADV
easat-3242	89	4	upscale	upscale	VERB
easat-3242	89	5	these	these	DET
easat-3242	89	6	feature	feature	NOUN
easat-3242	89	7	maps	map	NOUN
easat-3242	89	8	to	to	PART
easat-3242	89	9	give	give	VERB
easat-3242	89	10	the	the	DET
easat-3242	89	11	final	final	ADJ
easat-3242	89	12	reconstruction	reconstruction	NOUN
easat-3242	89	13	of	of	ADP
easat-3242	89	14	segmentation	segmentation	NOUN
easat-3242	89	15	mask	mask	NOUN
easat-3242	89	16	and	and	CCONJ
easat-3242	89	17	improve	improve	VERB
easat-3242	89	18	it	it	PRON
easat-3242	89	19	further	far	ADV
easat-3242	89	20	upon	upon	SCONJ
easat-3242	89	21	iteration	iteration	NOUN
easat-3242	89	22	.	.	PUNCT
easat-3242	90	1	this	this	DET
easat-3242	90	2	basic	basic	ADJ
easat-3242	90	3	structure	structure	NOUN
easat-3242	90	4	is	be	AUX
easat-3242	90	5	similar	similar	ADJ
easat-3242	90	6	to	to	ADP
easat-3242	90	7	the	the	DET
easat-3242	90	8	u	u	NOUN
easat-3242	90	9	-	-	NOUN
easat-3242	90	10	net	net	ADJ
easat-3242	90	11	,	,	PUNCT
easat-3242	90	12	while	while	SCONJ
easat-3242	90	13	in	in	ADP
easat-3242	90	14	case	case	NOUN
easat-3242	90	15	of	of	ADP
easat-3242	90	16	u	u	NOUN
easat-3242	90	17	-	-	PROPN
easat-3242	90	18	net++	net++	PROPN
easat-3242	90	19	,	,	PUNCT
easat-3242	90	20	there	there	PRON
easat-3242	90	21	are	be	VERB
easat-3242	90	22	additional	additional	ADJ
easat-3242	90	23	connections	connection	NOUN
easat-3242	90	24	providing	provide	VERB
easat-3242	90	25	a	a	DET
easat-3242	90	26	more	more	ADV
easat-3242	90	27	easy	easy	ADJ
easat-3242	90	28	flow	flow	NOUN
easat-3242	90	29	of	of	ADP
easat-3242	90	30	information	information	NOUN
easat-3242	90	31	through	through	ADP
easat-3242	90	32	-	-	PUNCT
easat-3242	90	33	out	out	NOUN
easat-3242	90	34	the	the	DET
easat-3242	90	35	network	network	NOUN
easat-3242	90	36	.	.	PUNCT
easat-3242	91	1	it	it	PRON
easat-3242	91	2	directly	directly	ADV
easat-3242	91	3	leads	lead	VERB
easat-3242	91	4	to	to	ADP
easat-3242	91	5	more	more	ADV
easat-3242	91	6	effective	effective	ADJ
easat-3242	91	7	use	use	NOUN
easat-3242	91	8	of	of	ADP
easat-3242	91	9	data	datum	NOUN
easat-3242	91	10	from	from	ADP
easat-3242	91	11	various	various	ADJ
easat-3242	91	12	image	image	NOUN
easat-3242	91	13	regions	region	NOUN
easat-3242	91	14	and	and	CCONJ
easat-3242	91	15	improves	improve	VERB
easat-3242	91	16	the	the	DET
easat-3242	91	17	segmentation	segmentation	NOUN
easat-3242	91	18	accuracy	accuracy	NOUN
easat-3242	91	19	.	.	PUNCT
easat-3242	92	1	the	the	DET
easat-3242	92	2	key	key	ADJ
easat-3242	92	3	innovation	innovation	NOUN
easat-3242	92	4	in	in	ADP
easat-3242	92	5	u	u	PROPN
easat-3242	92	6	-	-	PROPN
easat-3242	92	7	net++	net++	PROPN
easat-3242	92	8	is	be	AUX
easat-3242	92	9	its	its	PRON
easat-3242	92	10	nested	nested	ADJ
easat-3242	92	11	architecture	architecture	NOUN
easat-3242	92	12	,	,	PUNCT
easat-3242	92	13	where	where	SCONJ
easat-3242	92	14	each	each	DET
easat-3242	92	15	encoder	encoder	NOUN
easat-3242	92	16	-	-	PUNCT
easat-3242	92	17	decoder	decoder	NOUN
easat-3242	92	18	stage	stage	NOUN
easat-3242	92	19	consists	consist	VERB
easat-3242	92	20	of	of	ADP
easat-3242	92	21	stacked	stack	VERB
easat-3242	92	22	unet	unet	NOUN
easat-3242	92	23	modules	module	NOUN
easat-3242	92	24	.	.	PUNCT
easat-3242	93	1	these	these	DET
easat-3242	93	2	multi	multi	ADJ
easat-3242	93	3	-	-	ADJ
easat-3242	93	4	scale	scale	ADJ
easat-3242	93	5	skip	skip	NOUN
easat-3242	93	6	connections	connection	NOUN
easat-3242	93	7	enable	enable	VERB
easat-3242	93	8	the	the	DET
easat-3242	93	9	network	network	NOUN
easat-3242	93	10	to	to	PART
easat-3242	93	11	collect	collect	VERB
easat-3242	93	12	and	and	CCONJ
easat-3242	93	13	aggregate	aggregate	VERB
easat-3242	93	14	information	information	NOUN
easat-3242	93	15	from	from	ADP
easat-3242	93	16	different	different	ADJ
easat-3242	93	17	levels	level	NOUN
easat-3242	93	18	of	of	ADP
easat-3242	93	19	abstraction	abstraction	NOUN
easat-3242	93	20	for	for	ADP
easat-3242	93	21	enhanced	enhanced	ADJ
easat-3242	93	22	segmentation	segmentation	NOUN
easat-3242	93	23	performance	performance	NOUN
easat-3242	93	24	when	when	SCONJ
easat-3242	93	25	objects	object	NOUN
easat-3242	93	26	are	be	AUX
easat-3242	93	27	in	in	ADP
easat-3242	93	28	various	various	ADJ
easat-3242	93	29	scales	scale	NOUN
easat-3242	93	30	.	.	PUNCT
easat-3242	94	1	in	in	ADP
easat-3242	94	2	the	the	DET
easat-3242	94	3	decoding	decode	VERB
easat-3242	94	4	phase	phase	NOUN
easat-3242	94	5	feature	feature	NOUN
easat-3242	94	6	maps	map	NOUN
easat-3242	94	7	,	,	PUNCT
easat-3242	94	8	produced	produce	VERB
easat-3242	94	9	by	by	ADP
easat-3242	94	10	corresponding	correspond	VERB
easat-3242	94	11	encoder	encoder	NOUN
easat-3242	94	12	stages	stage	NOUN
easat-3242	94	13	are	be	AUX
easat-3242	94	14	aggregated	aggregate	VERB
easat-3242	94	15	to	to	PART
easat-3242	94	16	produce	produce	VERB
easat-3242	94	17	more	more	ADJ
easat-3242	94	18	information	information	NOUN
easat-3242	94	19	for	for	ADP
easat-3242	94	20	segmentation	segmentation	NOUN
easat-3242	94	21	.	.	PUNCT
easat-3242	95	1	u	u	X
easat-3242	95	2	-	-	PUNCT
easat-3242	95	3	net++	net++	PROPN
easat-3242	95	4	build	build	VERB
easat-3242	95	5	a	a	DET
easat-3242	95	6	network	network	NOUN
easat-3242	95	7	with	with	ADP
easat-3242	95	8	feature	feature	NOUN
easat-3242	95	9	from	from	ADP
easat-3242	95	10	different	different	ADJ
easat-3242	95	11	scale	scale	NOUN
easat-3242	95	12	concatenated	concatenate	VERB
easat-3242	95	13	such	such	ADJ
easat-3242	95	14	that	that	SCONJ
easat-3242	95	15	it	it	PRON
easat-3242	95	16	understand	understand	VERB
easat-3242	95	17	both	both	CCONJ
easat-3242	95	18	fine	fine	ADJ
easat-3242	95	19	,	,	PUNCT
easat-3242	95	20	detailed	detailed	ADJ
easat-3242	95	21	and	and	CCONJ
easat-3242	95	22	more	more	ADV
easat-3242	95	23	abstract	abstract	ADJ
easat-3242	95	24	stuff	stuff	NOUN
easat-3242	95	25	at	at	ADP
easat-3242	95	26	high	high	ADJ
easat-3242	95	27	-	-	PUNCT
easat-3242	95	28	level	level	NOUN
easat-3242	95	29	.	.	PUNCT
easat-3242	96	1	with	with	ADP
easat-3242	96	2	such	such	ADJ
easat-3242	96	3	layer	layer	NOUN
easat-3242	96	4	-	-	PUNCT
easat-3242	96	5	by	by	ADP
easat-3242	96	6	-	-	PUNCT
easat-3242	96	7	layer	layer	NOUN
easat-3242	96	8	design	design	NOUN
easat-3242	96	9	,	,	PUNCT
easat-3242	96	10	the	the	DET
easat-3242	96	11	network	network	NOUN
easat-3242	96	12	can	can	AUX
easat-3242	96	13	perform	perform	VERB
easat-3242	96	14	even	even	ADV
easat-3242	96	15	better	well	ADV
easat-3242	96	16	in	in	ADP
easat-3242	96	17	segmentation	segmentation	NOUN
easat-3242	96	18	on	on	ADP
easat-3242	96	19	complex	complex	ADJ
easat-3242	96	20	structures	structure	NOUN
easat-3242	96	21	.	.	PUNCT
easat-3242	97	1	overall	overall	ADJ
easat-3242	97	2	,	,	PUNCT
easat-3242	97	3	u	u	NOUN
easat-3242	97	4	-	-	NOUN
easat-3242	97	5	net++	net++	PROPN
easat-3242	97	6	is	be	AUX
easat-3242	97	7	specifically	specifically	ADV
easat-3242	97	8	designed	design	VERB
easat-3242	97	9	to	to	PART
easat-3242	97	10	enhance	enhance	VERB
easat-3242	97	11	the	the	DET
easat-3242	97	12	segmentation	segmentation	NOUN
easat-3242	97	13	potential	potential	NOUN
easat-3242	97	14	of	of	ADP
easat-3242	97	15	neural	neural	ADJ
easat-3242	97	16	networks	network	NOUN
easat-3242	97	17	and	and	CCONJ
easat-3242	97	18	seems	seem	VERB
easat-3242	97	19	also	also	ADV
easat-3242	97	20	well	well	ADV
easat-3242	97	21	suited	suited	ADJ
easat-3242	97	22	for	for	ADP
easat-3242	97	23	tasks	task	NOUN
easat-3242	97	24	involve	involve	VERB
easat-3242	97	25	multi	multi	ADJ
easat-3242	97	26	-	-	ADJ
easat-3242	97	27	scale	scale	ADJ
easat-3242	97	28	feature	feature	NOUN
easat-3242	97	29	integration	integration	NOUN
easat-3242	97	30	that	that	PRON
easat-3242	97	31	require	require	VERB
easat-3242	97	32	precise	precise	ADJ
easat-3242	97	33	boundaries	boundary	NOUN
easat-3242	97	34	.	.	PUNCT
easat-3242	98	1	3.1.1	3.1.1	X
easat-3242	98	2	.	.	PUNCT
easat-3242	98	3	functionality	functionality	NOUN
easat-3242	98	4	in	in	ADP
easat-3242	98	5	straight	straight	ADJ
easat-3242	98	6	words	word	NOUN
easat-3242	98	7	,	,	PUNCT
easat-3242	98	8	u	u	NOUN
easat-3242	98	9	-	-	NOUN
easat-3242	98	10	net++	net++	PROPN
easat-3242	98	11	is	be	AUX
easat-3242	98	12	focusing	focus	VERB
easat-3242	98	13	on	on	ADP
easat-3242	98	14	extracting	extract	VERB
easat-3242	98	15	most	most	ADV
easat-3242	98	16	local	local	ADJ
easat-3242	98	17	and	and	CCONJ
easat-3242	98	18	global	global	ADJ
easat-3242	98	19	guidance	guidance	NOUN
easat-3242	98	20	from	from	ADP
easat-3242	98	21	different	different	ADJ
easat-3242	98	22	regions	region	NOUN
easat-3242	98	23	of	of	ADP
easat-3242	98	24	an	an	DET
easat-3242	98	25	input	input	NOUN
easat-3242	98	26	image	image	NOUN
easat-3242	98	27	.	.	PUNCT
easat-3242	99	1	this	this	PRON
easat-3242	99	2	is	be	AUX
easat-3242	99	3	achieved	achieve	VERB
easat-3242	99	4	by	by	ADP
easat-3242	99	5	using	use	VERB
easat-3242	99	6	multi	multi	ADJ
easat-3242	99	7	-	-	ADJ
easat-3242	99	8	scale	scale	ADJ
easat-3242	99	9	skip	skip	ADJ
easat-3242	99	10	connections	connection	NOUN
easat-3242	99	11	as	as	ADP
easat-3242	99	12	well	well	ADV
easat-3242	99	13	stacking	stack	VERB
easat-3242	99	14	multiple	multiple	ADJ
easat-3242	99	15	u	u	ADJ
easat-3242	99	16	-	-	ADJ
easat-3242	99	17	net	net	ADJ
easat-3242	99	18	modules	module	NOUN
easat-3242	99	19	,	,	PUNCT
easat-3242	99	20	in	in	ADP
easat-3242	99	21	order	order	NOUN
easat-3242	99	22	to	to	PART
easat-3242	99	23	help	help	VERB
easat-3242	99	24	the	the	DET
easat-3242	99	25	model	model	NOUN
easat-3242	99	26	better	well	ADV
easat-3242	99	27	capture	capture	VERB
easat-3242	99	28	spatiotemporal	spatiotemporal	ADJ
easat-3242	99	29	orientation	orientation	NOUN
easat-3242	99	30	and	and	CCONJ
easat-3242	99	31	structure	structure	NOUN
easat-3242	99	32	of	of	ADP
easat-3242	99	33	objects	object	NOUN
easat-3242	99	34	that	that	PRON
easat-3242	99	35	change	change	VERB
easat-3242	99	36	rapidly	rapidly	ADV
easat-3242	99	37	with	with	ADP
easat-3242	99	38	size	size	NOUN
easat-3242	99	39	or	or	CCONJ
easat-3242	99	40	shape	shape	NOUN
easat-3242	99	41	during	during	ADP
easat-3242	99	42	its	its	PRON
easat-3242	99	43	progress	progress	NOUN
easat-3242	99	44	.	.	PUNCT
easat-3242	100	1	with	with	ADP
easat-3242	100	2	the	the	DET
easat-3242	100	3	help	help	NOUN
easat-3242	100	4	of	of	ADP
easat-3242	100	5	these	these	DET
easat-3242	100	6	multi	multi	ADJ
easat-3242	100	7	-	-	ADJ
easat-3242	100	8	level	level	ADJ
easat-3242	100	9	skip	skip	ADJ
easat-3242	100	10	connections	connection	NOUN
easat-3242	100	11	in	in	ADP
easat-3242	100	12	both	both	CCONJ
easat-3242	100	13	up	up	ADV
easat-3242	100	14	-	-	PUNCT
easat-3242	100	15	sampling	sample	VERB
easat-3242	100	16	and	and	CCONJ
easat-3242	100	17	downscaling	downscale	VERB
easat-3242	100	18	processes	process	NOUN
easat-3242	100	19	,	,	PUNCT
easat-3242	100	20	it	it	PRON
easat-3242	100	21	reduces	reduce	VERB
easat-3242	100	22	loss	loss	NOUN
easat-3242	100	23	information	information	NOUN
easat-3242	100	24	by	by	ADP
easat-3242	100	25	maintaining	maintain	VERB
easat-3242	100	26	among	among	ADP
easat-3242	100	27	original	original	ADJ
easat-3242	100	28	image	image	NOUN
easat-3242	100	29	features	feature	NOUN
easat-3242	100	30	at	at	ADP
easat-3242	100	31	different	different	ADJ
easat-3242	100	32	scaling	scaling	NOUN
easat-3242	100	33	leading	lead	VERB
easat-3242	100	34	to	to	ADP
easat-3242	100	35	a	a	DET
easat-3242	100	36	perceptible	perceptible	ADJ
easat-3242	100	37	increase	increase	NOUN
easat-3242	100	38	in	in	ADP
easat-3242	100	39	pixel	pixel	ADJ
easat-3242	100	40	-	-	ADJ
easat-3242	100	41	wise	wise	ADJ
easat-3242	100	42	segmentation	segmentation	NOUN
easat-3242	100	43	accuracy	accuracy	NOUN
easat-3242	100	44	.	.	PUNCT
easat-3242	101	1	5711	5711	NUM
easat-3242	101	2	edelweiss	edelweiss	PROPN
easat-3242	101	3	applied	apply	VERB
easat-3242	101	4	science	science	NOUN
easat-3242	101	5	and	and	CCONJ
easat-3242	101	6	technology	technology	NOUN
easat-3242	101	7	issn	issn	PROPN
easat-3242	101	8	:	:	PUNCT
easat-3242	101	9	2576	2576	NUM
easat-3242	101	10	-	-	SYM
easat-3242	101	11	8484	8484	NUM
easat-3242	101	12	vol	vol	NOUN
easat-3242	101	13	.	.	PROPN
easat-3242	101	14	8	8	NUM
easat-3242	101	15	,	,	PUNCT
easat-3242	101	16	no	no	INTJ
easat-3242	101	17	.	.	NOUN
easat-3242	102	1	6	6	NUM
easat-3242	102	2	:	:	SYM
easat-3242	102	3	5707	5707	NUM
easat-3242	102	4	-	-	SYM
easat-3242	102	5	5722	5722	NUM
easat-3242	102	6	,	,	PUNCT
easat-3242	102	7	2024	2024	NUM
easat-3242	102	8	doi	doi	NOUN
easat-3242	102	9	:	:	PUNCT
easat-3242	102	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	102	11	©	©	PROPN
easat-3242	102	12	2024	2024	NUM
easat-3242	102	13	by	by	ADP
easat-3242	102	14	the	the	DET
easat-3242	102	15	authors	author	NOUN
easat-3242	102	16	;	;	PUNCT
easat-3242	102	17	licensee	licensee	PROPN
easat-3242	102	18	learning	learn	VERB
easat-3242	102	19	gate	gate	NOUN
easat-3242	102	20	u	u	PROPN
easat-3242	102	21	-	-	PROPN
easat-3242	102	22	net++	net++	PROPN
easat-3242	102	23	is	be	AUX
easat-3242	102	24	a	a	DET
easat-3242	102	25	more	more	ADV
easat-3242	102	26	hierarchical	hierarchical	ADJ
easat-3242	102	27	design	design	NOUN
easat-3242	102	28	model	model	NOUN
easat-3242	102	29	and	and	CCONJ
easat-3242	102	30	it	it	PRON
easat-3242	102	31	can	can	AUX
easat-3242	102	32	capture	capture	VERB
easat-3242	102	33	long	long	ADJ
easat-3242	102	34	-	-	PUNCT
easat-3242	102	35	range	range	NOUN
easat-3242	102	36	dependencies	dependency	NOUN
easat-3242	102	37	of	of	ADP
easat-3242	102	38	image	image	NOUN
easat-3242	102	39	information	information	NOUN
easat-3242	102	40	.	.	PUNCT
easat-3242	103	1	thus	thus	ADV
easat-3242	103	2	,	,	PUNCT
easat-3242	103	3	u	u	PROPN
easat-3242	103	4	-	-	NOUN
easat-3242	103	5	net++	net++	PROPN
easat-3242	103	6	can	can	AUX
easat-3242	103	7	capture	capture	VERB
easat-3242	103	8	both	both	CCONJ
easat-3242	103	9	global	global	ADJ
easat-3242	103	10	and	and	CCONJ
easat-3242	103	11	local	local	ADJ
easat-3242	103	12	contexts	context	NOUN
easat-3242	103	13	of	of	ADP
easat-3242	103	14	different	different	ADJ
easat-3242	103	15	spatial	spatial	ADJ
easat-3242	103	16	distributions	distribution	NOUN
easat-3242	103	17	in	in	ADP
easat-3242	103	18	the	the	DET
easat-3242	103	19	images	image	NOUN
easat-3242	103	20	to	to	PART
easat-3242	103	21	adapt	adapt	VERB
easat-3242	103	22	to	to	ADP
easat-3242	103	23	changes	change	NOUN
easat-3242	103	24	such	such	ADJ
easat-3242	103	25	as	as	ADP
easat-3242	103	26	target	target	NOUN
easat-3242	103	27	object	object	NOUN
easat-3242	103	28	size	size	NOUN
easat-3242	103	29	,	,	PUNCT
easat-3242	103	30	location	location	NOUN
easat-3242	103	31	or	or	CCONJ
easat-3242	103	32	appearance	appearance	NOUN
easat-3242	103	33	.	.	PUNCT
easat-3242	104	1	in	in	ADP
easat-3242	104	2	addition	addition	NOUN
easat-3242	104	3	,	,	PUNCT
easat-3242	104	4	this	this	DET
easat-3242	104	5	network	network	NOUN
easat-3242	104	6	has	have	VERB
easat-3242	104	7	multi	multi	ADJ
easat-3242	104	8	-	-	ADJ
easat-3242	104	9	scale	scale	ADJ
easat-3242	104	10	skip	skip	ADJ
easat-3242	104	11	connections	connection	NOUN
easat-3242	104	12	that	that	PRON
easat-3242	104	13	benefit	benefit	VERB
easat-3242	104	14	segmentation	segmentation	NOUN
easat-3242	104	15	tasks	task	NOUN
easat-3242	104	16	involving	involve	VERB
easat-3242	104	17	:	:	PUNCT
easat-3242	104	18	a	a	X
easat-3242	104	19	)	)	PUNCT
easat-3242	104	20	detecting	detect	VERB
easat-3242	104	21	fine	fine	ADJ
easat-3242	104	22	and	and	CCONJ
easat-3242	104	23	coarse	coarse	ADJ
easat-3242	104	24	objects	object	NOUN
easat-3242	104	25	,	,	PUNCT
easat-3242	104	26	and	and	CCONJ
easat-3242	104	27	b	b	X
easat-3242	104	28	)	)	PUNCT
easat-3242	104	29	simple	simple	ADJ
easat-3242	104	30	or	or	CCONJ
easat-3242	104	31	complex	complex	ADJ
easat-3242	104	32	object	object	NOUN
easat-3242	104	33	shapes	shape	NOUN
easat-3242	104	34	.	.	PUNCT
easat-3242	105	1	the	the	DET
easat-3242	105	2	u	u	NOUN
easat-3242	105	3	-	-	NOUN
easat-3242	105	4	net++	net++	PROPN
easat-3242	105	5	is	be	AUX
easat-3242	105	6	also	also	ADV
easat-3242	105	7	shown	show	VERB
easat-3242	105	8	to	to	PART
easat-3242	105	9	perform	perform	VERB
easat-3242	105	10	very	very	ADV
easat-3242	105	11	well	well	ADV
easat-3242	105	12	in	in	ADP
easat-3242	105	13	other	other	ADJ
easat-3242	105	14	biomedical	biomedical	ADJ
easat-3242	105	15	image	image	NOUN
easat-3242	105	16	segmentation	segmentation	NOUN
easat-3242	105	17	tasks	task	NOUN
easat-3242	105	18	and	and	CCONJ
easat-3242	105	19	it	it	PRON
easat-3242	105	20	was	be	AUX
easat-3242	105	21	applied	apply	VERB
easat-3242	105	22	actively	actively	ADV
easat-3242	105	23	not	not	PART
easat-3242	105	24	only	only	ADV
easat-3242	105	25	sematic	sematic	ADJ
easat-3242	105	26	segmentation	segmentation	NOUN
easat-3242	105	27	but	but	CCONJ
easat-3242	105	28	also	also	ADV
easat-3242	105	29	satellite	satellite	NOUN
easat-3242	105	30	imagery	imagery	NOUN
easat-3242	105	31	interpretation	interpretation	NOUN
easat-3242	105	32	,	,	PUNCT
easat-3242	105	33	remote	remote	ADJ
easat-3242	105	34	sensing	sense	VERB
easat-3242	105	35	application	application	NOUN
easat-3242	105	36	or	or	CCONJ
easat-3242	105	37	natural	natural	ADJ
easat-3242	105	38	scene	scene	NOUN
easat-3242	105	39	analysis	analysis	NOUN
easat-3242	105	40	area	area	NOUN
easat-3242	105	41	.	.	PUNCT
easat-3242	106	1	it	it	PRON
easat-3242	106	2	has	have	AUX
easat-3242	106	3	been	be	AUX
easat-3242	106	4	the	the	DET
easat-3242	106	5	tool	tool	NOUN
easat-3242	106	6	of	of	ADP
easat-3242	106	7	choice	choice	NOUN
easat-3242	106	8	for	for	ADP
easat-3242	106	9	segmentation	segmentation	NOUN
easat-3242	106	10	among	among	ADP
easat-3242	106	11	researchers	researcher	NOUN
easat-3242	106	12	and	and	CCONJ
easat-3242	106	13	practitioners	practitioner	NOUN
easat-3242	106	14	due	due	ADJ
easat-3242	106	15	to	to	ADP
easat-3242	106	16	how	how	SCONJ
easat-3242	106	17	flexible	flexible	ADJ
easat-3242	106	18	,	,	PUNCT
easat-3242	106	19	whilst	whilst	SCONJ
easat-3242	106	20	still	still	ADV
easat-3242	106	21	efficient	efficient	ADJ
easat-3242	106	22	it	it	PRON
easat-3242	106	23	is	be	AUX
easat-3242	106	24	.	.	PUNCT
easat-3242	107	1	u	u	X
easat-3242	107	2	-	-	PUNCT
easat-3242	107	3	net++	net++	PROPN
easat-3242	107	4	is	be	AUX
easat-3242	107	5	a	a	DET
easat-3242	107	6	better	well	ADJ
easat-3242	107	7	u	u	ADJ
easat-3242	107	8	-	-	ADJ
easat-3242	107	9	net	net	ADJ
easat-3242	107	10	model	model	NOUN
easat-3242	107	11	than	than	ADP
easat-3242	107	12	the	the	DET
easat-3242	107	13	original	original	ADJ
easat-3242	107	14	one	one	NOUN
easat-3242	107	15	and	and	CCONJ
easat-3242	107	16	resolves	resolve	VERB
easat-3242	107	17	some	some	DET
easat-3242	107	18	limitations	limitation	NOUN
easat-3242	107	19	about	about	ADP
easat-3242	107	20	more	more	ADJ
easat-3242	107	21	context	context	NOUN
easat-3242	107	22	powerful	powerful	ADJ
easat-3242	107	23	/	/	SYM
easat-3242	107	24	better	well	ADJ
easat-3242	107	25	segmentation	segmentation	NOUN
easat-3242	107	26	accuracy	accuracy	NOUN
easat-3242	107	27	.	.	PUNCT
easat-3242	108	1	it	it	PRON
easat-3242	108	2	is	be	AUX
easat-3242	108	3	not	not	PART
easat-3242	108	4	just	just	ADV
easat-3242	108	5	suitable	suitable	ADJ
easat-3242	108	6	for	for	ADP
easat-3242	108	7	computer	computer	NOUN
easat-3242	108	8	vision	vision	NOUN
easat-3242	108	9	tasks	task	NOUN
easat-3242	108	10	but	but	CCONJ
easat-3242	108	11	also	also	ADV
easat-3242	108	12	can	can	AUX
easat-3242	108	13	be	be	AUX
easat-3242	108	14	applied	apply	VERB
easat-3242	108	15	to	to	ADP
easat-3242	108	16	many	many	ADJ
easat-3242	108	17	medical	medical	ADJ
easat-3242	108	18	image	image	NOUN
easat-3242	108	19	domains	domain	NOUN
easat-3242	108	20	supported	support	VERB
easat-3242	108	21	by	by	ADP
easat-3242	108	22	its	its	PRON
easat-3242	108	23	architecture	architecture	NOUN
easat-3242	108	24	.	.	PUNCT
easat-3242	109	1	3.2	3.2	NUM
easat-3242	109	2	.	.	PUNCT
easat-3242	110	1	vgg19	vgg19	VERB
easat-3242	110	2	the	the	DET
easat-3242	110	3	vgg19	vgg19	NOUN
easat-3242	110	4	is	be	AUX
easat-3242	110	5	a	a	DET
easat-3242	110	6	deep	deep	ADJ
easat-3242	110	7	convolutional	convolutional	ADJ
easat-3242	110	8	neural	neural	ADJ
easat-3242	110	9	network	network	NOUN
easat-3242	110	10	(	(	PUNCT
easat-3242	110	11	cnn	cnn	PROPN
easat-3242	110	12	)	)	PUNCT
easat-3242	110	13	from	from	ADP
easat-3242	110	14	the	the	DET
easat-3242	110	15	visual	visual	ADJ
easat-3242	110	16	geometry	geometry	NOUN
easat-3242	110	17	group	group	NOUN
easat-3242	110	18	.	.	PUNCT
easat-3242	111	1	an	an	DET
easat-3242	111	2	extension	extension	NOUN
easat-3242	111	3	of	of	ADP
easat-3242	111	4	the	the	DET
easat-3242	111	5	vgg16	vgg16	NOUN
easat-3242	111	6	architecture	architecture	NOUN
easat-3242	111	7	with	with	ADP
easat-3242	111	8	additional	additional	ADJ
easat-3242	111	9	higher	high	ADJ
easat-3242	111	10	resolution	resolution	NOUN
easat-3242	111	11	layers	layer	NOUN
easat-3242	111	12	and	and	CCONJ
easat-3242	111	13	a	a	DET
easat-3242	111	14	downscaled	downscaled	ADJ
easat-3242	111	15	pooling	pool	VERB
easat-3242	111	16	operation	operation	NOUN
easat-3242	111	17	to	to	PART
easat-3242	111	18	achieve	achieve	VERB
easat-3242	111	19	better	well	ADJ
easat-3242	111	20	performance	performance	NOUN
easat-3242	111	21	for	for	ADP
easat-3242	111	22	image	image	NOUN
easat-3242	111	23	recognition	recognition	NOUN
easat-3242	111	24	,	,	PUNCT
easat-3242	111	25	made	make	VERB
easat-3242	111	26	by	by	ADP
easat-3242	111	27	karen	karen	PROPN
easat-3242	111	28	simonyan	simonyan	PROPN
easat-3242	111	29	and	and	CCONJ
easat-3242	111	30	andrew	andrew	PROPN
easat-3242	111	31	zisserman	zisserman	PROPN
easat-3242	111	32	in	in	ADP
easat-3242	111	33	"	"	PUNCT
easat-3242	111	34	very	very	ADV
easat-3242	111	35	deep	deep	ADJ
easat-3242	111	36	convolutional	convolutional	ADJ
easat-3242	111	37	networks	network	NOUN
easat-3242	111	38	for	for	ADP
easat-3242	111	39	large	large	ADJ
easat-3242	111	40	-	-	PUNCT
easat-3242	111	41	scale	scale	NOUN
easat-3242	111	42	image	image	NOUN
easat-3242	111	43	recognition	recognition	NOUN
easat-3242	111	44	"	"	PUNCT
easat-3242	111	45	(	(	PUNCT
easat-3242	111	46	2014	2014	NUM
easat-3242	111	47	)	)	PUNCT
easat-3242	111	48	.	.	PUNCT
easat-3242	112	1	detailed	detailed	ADJ
easat-3242	112	2	analysis	analysis	NOUN
easat-3242	112	3	of	of	ADP
easat-3242	112	4	vgg19	vgg19	NOUN
easat-3242	112	5	architecture	architecture	NOUN
easat-3242	112	6	is	be	AUX
easat-3242	112	7	given	give	VERB
easat-3242	112	8	here	here	ADV
easat-3242	112	9	:	:	PUNCT
easat-3242	112	10	3.3	3.3	NUM
easat-3242	112	11	.	.	PUNCT
easat-3242	113	1	input	input	NOUN
easat-3242	113	2	layer	layer	NOUN
easat-3242	113	3	the	the	DET
easat-3242	113	4	input	input	NOUN
easat-3242	113	5	layer	layer	NOUN
easat-3242	113	6	of	of	ADP
easat-3242	113	7	vgg19	vgg19	PROPN
easat-3242	113	8	takes	take	VERB
easat-3242	113	9	care	care	NOUN
easat-3242	113	10	to	to	PART
easat-3242	113	11	receive	receive	VERB
easat-3242	113	12	and	and	CCONJ
easat-3242	113	13	process	process	VERB
easat-3242	113	14	the	the	DET
easat-3242	113	15	image	image	NOUN
easat-3242	113	16	data	datum	NOUN
easat-3242	113	17	before	before	SCONJ
easat-3242	113	18	it	it	PRON
easat-3242	113	19	is	be	AUX
easat-3242	113	20	passed	pass	VERB
easat-3242	113	21	through	through	ADP
easat-3242	113	22	the	the	DET
easat-3242	113	23	network	network	NOUN
easat-3242	113	24	.	.	PUNCT
easat-3242	114	1	vgg19	vgg19	PROPN
easat-3242	114	2	model	model	NOUN
easat-3242	114	3	was	be	AUX
easat-3242	114	4	developed	develop	VERB
easat-3242	114	5	with	with	ADP
easat-3242	114	6	the	the	DET
easat-3242	114	7	goal	goal	NOUN
easat-3242	114	8	that	that	SCONJ
easat-3242	114	9	it	it	PRON
easat-3242	114	10	should	should	AUX
easat-3242	114	11	work	work	VERB
easat-3242	114	12	on	on	ADP
easat-3242	114	13	input	input	NOUN
easat-3242	114	14	images	image	NOUN
easat-3242	114	15	of	of	ADP
easat-3242	114	16	size	size	NOUN
easat-3242	114	17	224×224	224×224	NUM
easat-3242	114	18	pixels	pixel	NOUN
easat-3242	114	19	allowing	allow	VERB
easat-3242	114	20	a	a	DET
easat-3242	114	21	more	more	ADV
easat-3242	114	22	consistent	consistent	ADJ
easat-3242	114	23	way	way	NOUN
easat-3242	114	24	to	to	PART
easat-3242	114	25	create	create	VERB
easat-3242	114	26	and	and	CCONJ
easat-3242	114	27	use	use	VERB
easat-3242	114	28	image	image	NOUN
easat-3242	114	29	data	data	NOUN
easat-3242	114	30	sets	set	NOUN
easat-3242	114	31	.	.	PUNCT
easat-3242	115	1	images	image	NOUN
easat-3242	115	2	are	be	AUX
easat-3242	115	3	either	either	CCONJ
easat-3242	115	4	in	in	ADP
easat-3242	115	5	colour	colour	NOUN
easat-3242	115	6	(	(	PUNCT
easat-3242	115	7	rgb	rgb	PROPN
easat-3242	115	8	three	three	NUM
easat-3242	115	9	channels	channel	NOUN
easat-3242	115	10	)	)	PUNCT
easat-3242	115	11	or	or	CCONJ
easat-3242	115	12	gray	gray	ADJ
easat-3242	115	13	scale	scale	NOUN
easat-3242	115	14	(	(	PUNCT
easat-3242	115	15	single	single	ADJ
easat-3242	115	16	channel	channel	NOUN
easat-3242	115	17	)	)	PUNCT
easat-3242	115	18	.	.	PUNCT
easat-3242	116	1	rgb	rgb	PROPN
easat-3242	116	2	images	image	NOUN
easat-3242	116	3	have	have	VERB
easat-3242	116	4	the	the	DET
easat-3242	116	5	shape	shape	NOUN
easat-3242	116	6	224x224x3	224x224x3	NUM
easat-3242	116	7	where	where	SCONJ
easat-3242	116	8	each	each	PRON
easat-3242	116	9	of	of	ADP
easat-3242	116	10	red	red	ADJ
easat-3242	116	11	,	,	PUNCT
easat-3242	116	12	green	green	ADJ
easat-3242	116	13	or	or	CCONJ
easat-3242	116	14	blue	blue	ADJ
easat-3242	116	15	channels	channel	NOUN
easat-3242	116	16	has	have	VERB
easat-3242	116	17	a	a	DET
easat-3242	116	18	dimensionality	dimensionality	NOUN
easat-3242	116	19	for	for	ADP
easat-3242	116	20	rgb	rgb	PROPN
easat-3242	116	21	colour	colour	NOUN
easat-3242	116	22	maps	map	NOUN
easat-3242	116	23	and	and	CCONJ
easat-3242	116	24	gray	gray	ADJ
easat-3242	116	25	scale	scale	NOUN
easat-3242	116	26	is	be	AUX
easat-3242	116	27	in	in	ADP
easat-3242	116	28	form	form	NOUN
easat-3242	116	29	of	of	ADP
easat-3242	116	30	one	one	NUM
easat-3242	116	31	2d	2d	NUM
easat-3242	116	32	array	array	NOUN
easat-3242	116	33	with	with	ADP
easat-3242	116	34	dimensions	dimension	NOUN
easat-3242	116	35	(	(	PUNCT
easat-3242	116	36	also	also	ADV
easat-3242	116	37	flattened	flatten	VERB
easat-3242	116	38	)	)	PUNCT
easat-3242	116	39	to	to	PART
easat-3242	116	40	make	make	VERB
easat-3242	116	41	its	its	PRON
easat-3242	116	42	appearance	appearance	NOUN
easat-3242	116	43	similar	similar	ADJ
easat-3242	116	44	.	.	PUNCT
easat-3242	117	1	we	we	PRON
easat-3242	117	2	use	use	VERB
easat-3242	117	3	the	the	DET
easat-3242	117	4	input	input	NOUN
easat-3242	117	5	layer	layer	NOUN
easat-3242	117	6	so	so	SCONJ
easat-3242	117	7	that	that	SCONJ
easat-3242	117	8	image	image	NOUN
easat-3242	117	9	size	size	NOUN
easat-3242	117	10	is	be	AUX
easat-3242	117	11	always	always	ADV
easat-3242	117	12	fixed	fix	VERB
easat-3242	117	13	when	when	SCONJ
easat-3242	117	14	every	every	DET
easat-3242	117	15	input	input	NOUN
easat-3242	117	16	goes	go	VERB
easat-3242	117	17	through	through	ADP
easat-3242	117	18	these	these	PRON
easat-3242	117	19	;	;	PUNCT
easat-3242	117	20	it	it	PRON
easat-3242	117	21	makes	make	VERB
easat-3242	117	22	training	training	NOUN
easat-3242	117	23	as	as	ADV
easat-3242	117	24	well	well	ADV
easat-3242	117	25	inference	inference	NOUN
easat-3242	117	26	simple	simple	ADJ
easat-3242	117	27	.	.	PUNCT
easat-3242	118	1	after	after	ADP
easat-3242	118	2	resizing	resize	VERB
easat-3242	118	3	the	the	DET
easat-3242	118	4	input	input	NOUN
easat-3242	118	5	images	image	NOUN
easat-3242	118	6	,	,	PUNCT
easat-3242	118	7	rgb	rgb	PROPN
easat-3242	118	8	pixel	pixel	PROPN
easat-3242	118	9	values	value	NOUN
easat-3242	118	10	are	be	AUX
easat-3242	118	11	normalized	normalize	VERB
easat-3242	118	12	and	and	CCONJ
easat-3242	118	13	turned	turn	VERB
easat-3242	118	14	into	into	ADP
easat-3242	118	15	numerical	numerical	ADJ
easat-3242	118	16	tensors	tensor	NOUN
easat-3242	118	17	.	.	PUNCT
easat-3242	119	1	basically	basically	ADV
easat-3242	119	2	each	each	DET
easat-3242	119	3	pixel	pixel	NOUN
easat-3242	119	4	in	in	ADP
easat-3242	119	5	the	the	DET
easat-3242	119	6	image	image	NOUN
easat-3242	119	7	becomes	become	VERB
easat-3242	119	8	a	a	DET
easat-3242	119	9	input	input	NOUN
easat-3242	119	10	feature	feature	NOUN
easat-3242	119	11	,	,	PUNCT
easat-3242	119	12	where	where	SCONJ
easat-3242	119	13	its	its	PRON
easat-3242	119	14	intensity	intensity	NOUN
easat-3242	119	15	level	level	NOUN
easat-3242	119	16	being	be	AUX
easat-3242	119	17	treated	treat	VERB
easat-3242	119	18	as	as	ADP
easat-3242	119	19	value	value	NOUN
easat-3242	119	20	of	of	ADP
easat-3242	119	21	that	that	DET
easat-3242	119	22	feature	feature	NOUN
easat-3242	119	23	.	.	PUNCT
easat-3242	120	1	this	this	DET
easat-3242	120	2	input	input	NOUN
easat-3242	120	3	tensor	tensor	NOUN
easat-3242	120	4	is	be	AUX
easat-3242	120	5	then	then	ADV
easat-3242	120	6	flowed	flow	VERB
easat-3242	120	7	through	through	ADP
easat-3242	120	8	the	the	DET
easat-3242	120	9	convolutional	convolutional	ADJ
easat-3242	120	10	layers	layer	NOUN
easat-3242	120	11	which	which	PRON
easat-3242	120	12	initiates	initiate	VERB
easat-3242	120	13	feature	feature	NOUN
easat-3242	120	14	extraction	extraction	NOUN
easat-3242	120	15	.	.	PUNCT
easat-3242	121	1	this	this	PRON
easat-3242	121	2	is	be	AUX
easat-3242	121	3	an	an	DET
easat-3242	121	4	important	important	ADJ
easat-3242	121	5	contribution	contribution	NOUN
easat-3242	121	6	to	to	ADP
easat-3242	121	7	the	the	DET
easat-3242	121	8	model	model	NOUN
easat-3242	121	9	since	since	SCONJ
easat-3242	121	10	it	it	PRON
easat-3242	121	11	guarantees	guarantee	VERB
easat-3242	121	12	that	that	SCONJ
easat-3242	121	13	all	all	DET
easat-3242	121	14	images	image	NOUN
easat-3242	121	15	are	be	AUX
easat-3242	121	16	processed	process	VERB
easat-3242	121	17	equally	equally	ADV
easat-3242	121	18	,	,	PUNCT
easat-3242	121	19	promoting	promote	VERB
easat-3242	121	20	thus	thus	ADV
easat-3242	121	21	some	some	DET
easat-3242	121	22	concretion	concretion	NOUN
easat-3242	121	23	with	with	ADP
easat-3242	121	24	those	those	DET
easat-3242	121	25	images	image	NOUN
easat-3242	121	26	which	which	PRON
easat-3242	121	27	come	come	VERB
easat-3242	121	28	from	from	ADP
easat-3242	121	29	a	a	DET
easat-3242	121	30	different	different	ADJ
easat-3242	121	31	dataset	dataset	NOUN
easat-3242	121	32	.	.	PUNCT
easat-3242	122	1	vgg19	vgg19	PROPN
easat-3242	122	2	is	be	AUX
easat-3242	122	3	able	able	ADJ
easat-3242	122	4	to	to	PART
easat-3242	122	5	outperform	outperform	VERB
easat-3242	122	6	across	across	ADP
easat-3242	122	7	a	a	DET
easat-3242	122	8	spectrum	spectrum	NOUN
easat-3242	122	9	of	of	ADP
easat-3242	122	10	visual	visual	ADJ
easat-3242	122	11	recognition	recognition	NOUN
easat-3242	122	12	tasks	task	NOUN
easat-3242	122	13	due	due	ADJ
easat-3242	122	14	the	the	DET
easat-3242	122	15	fact	fact	NOUN
easat-3242	122	16	its	its	PRON
easat-3242	122	17	input	input	NOUN
easat-3242	122	18	size	size	NOUN
easat-3242	122	19	and	and	CCONJ
easat-3242	122	20	format	format	NOUN
easat-3242	122	21	are	be	AUX
easat-3242	122	22	standardised	standardise	VERB
easat-3242	122	23	,	,	PUNCT
easat-3242	122	24	freeing	free	VERB
easat-3242	122	25	up	up	ADP
easat-3242	122	26	resources	resource	NOUN
easat-3242	122	27	in	in	ADP
easat-3242	122	28	training	training	NOUN
easat-3242	122	29	for	for	ADP
easat-3242	122	30	learning	learn	VERB
easat-3242	122	31	patterns	pattern	NOUN
easat-3242	122	32	opposed	oppose	VERB
easat-3242	122	33	to	to	ADP
easat-3242	122	34	adjusting	adjust	VERB
easat-3242	122	35	weights	weight	NOUN
easat-3242	122	36	at	at	ADP
easat-3242	122	37	each	each	DET
easat-3242	122	38	layer	layer	NOUN
easat-3242	122	39	.	.	PUNCT
easat-3242	123	1	3.3.1	3.3.1	X
easat-3242	123	2	.	.	PUNCT
easat-3242	123	3	convolutional	convolutional	ADJ
easat-3242	123	4	layers	layer	NOUN
easat-3242	123	5	the	the	DET
easat-3242	123	6	convolutional	convolutional	ADJ
easat-3242	123	7	layers	layer	NOUN
easat-3242	123	8	form	form	VERB
easat-3242	123	9	the	the	DET
easat-3242	123	10	base	base	NOUN
easat-3242	123	11	of	of	ADP
easat-3242	123	12	vgg19	vgg19	PROPN
easat-3242	123	13	and	and	CCONJ
easat-3242	123	14	take	take	VERB
easat-3242	123	15	care	care	NOUN
easat-3242	123	16	for	for	ADP
easat-3242	123	17	feature	feature	NOUN
easat-3242	123	18	extraction	extraction	NOUN
easat-3242	123	19	from	from	ADP
easat-3242	123	20	input	input	NOUN
easat-3242	123	21	images	image	NOUN
easat-3242	123	22	.	.	PUNCT
easat-3242	124	1	convolutional	convolutional	ADJ
easat-3242	124	2	layers	layer	NOUN
easat-3242	124	3	hold	hold	VERB
easat-3242	124	4	learnable	learnable	ADJ
easat-3242	124	5	weights	weight	NOUN
easat-3242	124	6	by	by	ADP
easat-3242	124	7	executing	execute	VERB
easat-3242	124	8	several	several	ADJ
easat-3242	124	9	filters	filter	NOUN
easat-3242	124	10	or	or	CCONJ
easat-3242	124	11	kernels	kernel	NOUN
easat-3242	124	12	on	on	ADP
easat-3242	124	13	each	each	DET
easat-3242	124	14	image	image	NOUN
easat-3242	124	15	which	which	PRON
easat-3242	124	16	in	in	ADP
easat-3242	124	17	the	the	DET
easat-3242	124	18	process	process	NOUN
easat-3242	124	19	will	will	AUX
easat-3242	124	20	learn	learn	VERB
easat-3242	124	21	2d	2d	NOUN
easat-3242	124	22	patterns	pattern	NOUN
easat-3242	124	23	(	(	PUNCT
easat-3242	124	24	like	like	ADP
easat-3242	124	25	edges	edge	NOUN
easat-3242	124	26	,	,	PUNCT
easat-3242	124	27	gradients	gradient	NOUN
easat-3242	124	28	)	)	PUNCT
easat-3242	124	29	and	and	CCONJ
easat-3242	124	30	so	so	ADV
easat-3242	124	31	forth	forth	ADV
easat-3242	124	32	.	.	PUNCT
easat-3242	125	1	vgg19	vgg19	PROPN
easat-3242	125	2	uses	use	VERB
easat-3242	125	3	small	small	ADJ
easat-3242	125	4	3×3	3×3	NUM
easat-3242	125	5	size	size	NOUN
easat-3242	125	6	filters	filter	NOUN
easat-3242	125	7	which	which	PRON
easat-3242	125	8	allow	allow	VERB
easat-3242	125	9	them	they	PRON
easat-3242	125	10	to	to	PART
easat-3242	125	11	capture	capture	VERB
easat-3242	125	12	the	the	DET
easat-3242	125	13	spatial	spatial	ADJ
easat-3242	125	14	pattern	pattern	NOUN
easat-3242	125	15	within	within	ADP
easat-3242	125	16	an	an	DET
easat-3242	125	17	image	image	NOUN
easat-3242	125	18	.	.	PUNCT
easat-3242	126	1	these	these	DET
easat-3242	126	2	filters	filter	NOUN
easat-3242	126	3	are	be	AUX
easat-3242	126	4	then	then	ADV
easat-3242	126	5	drawn	draw	VERB
easat-3242	126	6	across	across	ADP
easat-3242	126	7	the	the	DET
easat-3242	126	8	entire	entire	ADJ
easat-3242	126	9	image	image	NOUN
easat-3242	126	10	one	one	NUM
easat-3242	126	11	by	by	ADP
easat-3242	126	12	one	one	NUM
easat-3242	126	13	,	,	PUNCT
easat-3242	126	14	where	where	SCONJ
easat-3242	126	15	they	they	PRON
easat-3242	126	16	create	create	VERB
easat-3242	126	17	feature	feature	NOUN
easat-3242	126	18	maps	map	NOUN
easat-3242	126	19	that	that	PRON
easat-3242	126	20	give	give	VERB
easat-3242	126	21	high	high	ADJ
easat-3242	126	22	importance	importance	NOUN
easat-3242	126	23	to	to	ADP
easat-3242	126	24	a	a	DET
easat-3242	126	25	particular	particular	ADJ
easat-3242	126	26	structure	structure	NOUN
easat-3242	126	27	such	such	ADJ
easat-3242	126	28	as	as	ADP
easat-3242	126	29	an	an	DET
easat-3242	126	30	edge	edge	NOUN
easat-3242	126	31	or	or	CCONJ
easat-3242	126	32	curve	curve	NOUN
easat-3242	126	33	.	.	PUNCT
easat-3242	127	1	convolutional	convolutional	ADJ
easat-3242	127	2	layers	layer	NOUN
easat-3242	127	3	operate	operate	VERB
easat-3242	127	4	in	in	ADP
easat-3242	127	5	blocks	block	NOUN
easat-3242	127	6	.	.	PUNCT
easat-3242	128	1	there	there	PRON
easat-3242	128	2	can	can	AUX
easat-3242	128	3	be	be	AUX
easat-3242	128	4	any	any	DET
easat-3242	128	5	number	number	NOUN
easat-3242	128	6	of	of	ADP
easat-3242	128	7	convolutions	convolution	NOUN
easat-3242	128	8	stacked	stack	VERB
easat-3242	128	9	together	together	ADV
easat-3242	128	10	forming	form	VERB
easat-3242	128	11	a	a	DET
easat-3242	128	12	block	block	NOUN
easat-3242	128	13	.	.	PUNCT
easat-3242	129	1	these	these	DET
easat-3242	129	2	layers	layer	NOUN
easat-3242	129	3	capture	capture	VERB
easat-3242	129	4	more	more	ADV
easat-3242	129	5	complex	complex	ADJ
easat-3242	129	6	features	feature	NOUN
easat-3242	129	7	as	as	SCONJ
easat-3242	129	8	they	they	PRON
easat-3242	129	9	attune	attune	VERB
easat-3242	129	10	to	to	ADP
easat-3242	129	11	the	the	DET
easat-3242	129	12	progression	progression	NOUN
easat-3242	129	13	,	,	PUNCT
easat-3242	129	14	capturing	capture	VERB
easat-3242	129	15	finer	fine	ADJ
easat-3242	129	16	and	and	CCONJ
easat-3242	129	17	finer	fine	ADJ
easat-3242	129	18	patterns	pattern	NOUN
easat-3242	129	19	ranging	range	VERB
easat-3242	129	20	from	from	ADP
easat-3242	129	21	simple	simple	ADJ
easat-3242	129	22	edges	edge	NOUN
easat-3242	129	23	at	at	ADP
easat-3242	129	24	very	very	ADV
easat-3242	129	25	early	early	ADJ
easat-3242	129	26	levels	level	NOUN
easat-3242	129	27	of	of	ADP
easat-3242	129	28	layer	layer	NOUN
easat-3242	129	29	1	1	NUM
easat-3242	129	30	up	up	ADP
easat-3242	129	31	to	to	ADP
easat-3242	129	32	parts	part	NOUN
easat-3242	129	33	of	of	ADP
easat-3242	129	34	objects	object	NOUN
easat-3242	129	35	in	in	ADP
easat-3242	129	36	higher	high	ADJ
easat-3242	129	37	level	level	NOUN
easat-3242	129	38	concepts	concept	NOUN
easat-3242	129	39	.	.	PUNCT
easat-3242	130	1	convolutional	convolutional	ADJ
easat-3242	130	2	layers	layer	NOUN
easat-3242	130	3	in	in	ADP
easat-3242	130	4	the	the	DET
easat-3242	130	5	vgg19	vgg19	NOUN
easat-3242	130	6	model	model	NOUN
easat-3242	130	7	are	be	AUX
easat-3242	130	8	accompanied	accompany	VERB
easat-3242	130	9	by	by	ADP
easat-3242	130	10	an	an	DET
easat-3242	130	11	activation	activation	NOUN
easat-3242	130	12	function	function	NOUN
easat-3242	130	13	,	,	PUNCT
easat-3242	130	14	rectified	rectify	VERB
easat-3242	130	15	linear	linear	NOUN
easat-3242	130	16	unit	unit	NOUN
easat-3242	130	17	(	(	PUNCT
easat-3242	130	18	relu	relu	NOUN
easat-3242	130	19	)	)	PUNCT
easat-3242	130	20	,	,	PUNCT
easat-3242	130	21	which	which	PRON
easat-3242	130	22	injects	inject	VERB
easat-3242	130	23	non	non	ADJ
easat-3242	130	24	-	-	NOUN
easat-3242	130	25	linearity	linearity	NOUN
easat-3242	130	26	into	into	ADP
easat-3242	130	27	our	our	PRON
easat-3242	130	28	architecture	architecture	NOUN
easat-3242	130	29	.	.	PUNCT
easat-3242	131	1	this	this	PRON
easat-3242	131	2	non	non	ADJ
easat-3242	131	3	-	-	ADJ
easat-3242	131	4	linearity	linearity	NOUN
easat-3242	131	5	is	be	AUX
easat-3242	131	6	essential	essential	ADJ
easat-3242	131	7	for	for	SCONJ
easat-3242	131	8	the	the	DET
easat-3242	131	9	neural	neural	ADJ
easat-3242	131	10	network	network	NOUN
easat-3242	131	11	to	to	PART
easat-3242	131	12	learn	learn	VERB
easat-3242	131	13	intricate	intricate	ADJ
easat-3242	131	14	and	and	CCONJ
easat-3242	131	15	abstract	abstract	ADJ
easat-3242	131	16	patterns	pattern	NOUN
easat-3242	131	17	which	which	PRON
easat-3242	131	18	would	would	AUX
easat-3242	131	19	not	not	PART
easat-3242	131	20	be	be	AUX
easat-3242	131	21	able	able	ADJ
easat-3242	131	22	without	without	ADP
easat-3242	131	23	adding	add	VERB
easat-3242	131	24	up	up	ADP
easat-3242	131	25	all	all	PRON
easat-3242	131	26	of	of	ADP
easat-3242	131	27	those	those	DET
easat-3242	131	28	linear	linear	ADJ
easat-3242	131	29	transformations	transformation	NOUN
easat-3242	131	30	.	.	PUNCT
easat-3242	132	1	due	due	ADP
easat-3242	132	2	to	to	ADP
easat-3242	132	3	the	the	DET
easat-3242	132	4	presence	presence	NOUN
easat-3242	132	5	of	of	ADP
easat-3242	132	6	several	several	ADJ
easat-3242	132	7	convolutional	convolutional	ADJ
easat-3242	132	8	layers	layer	NOUN
easat-3242	132	9	and	and	CCONJ
easat-3242	132	10	relu	relu	NOUN
easat-3242	132	11	5712	5712	NUM
easat-3242	132	12	edelweiss	edelweiss	PROPN
easat-3242	132	13	applied	apply	VERB
easat-3242	132	14	science	science	NOUN
easat-3242	132	15	and	and	CCONJ
easat-3242	132	16	technology	technology	NOUN
easat-3242	132	17	issn	issn	PROPN
easat-3242	132	18	:	:	PUNCT
easat-3242	132	19	2576	2576	NUM
easat-3242	132	20	-	-	SYM
easat-3242	132	21	8484	8484	NUM
easat-3242	132	22	vol	vol	NOUN
easat-3242	132	23	.	.	PROPN
easat-3242	132	24	8	8	NUM
easat-3242	132	25	,	,	PUNCT
easat-3242	132	26	no	no	INTJ
easat-3242	132	27	.	.	NOUN
easat-3242	133	1	6	6	NUM
easat-3242	133	2	:	:	SYM
easat-3242	133	3	5707	5707	NUM
easat-3242	133	4	-	-	SYM
easat-3242	133	5	5722	5722	NUM
easat-3242	133	6	,	,	PUNCT
easat-3242	133	7	2024	2024	NUM
easat-3242	133	8	doi	doi	NOUN
easat-3242	133	9	:	:	PUNCT
easat-3242	133	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	133	11	©	©	PROPN
easat-3242	133	12	2024	2024	NUM
easat-3242	133	13	by	by	ADP
easat-3242	133	14	the	the	DET
easat-3242	133	15	authors	author	NOUN
easat-3242	133	16	;	;	PUNCT
easat-3242	133	17	licensee	licensee	PROPN
easat-3242	133	18	learning	learn	VERB
easat-3242	133	19	gate	gate	NOUN
easat-3242	133	20	activation	activation	NOUN
easat-3242	133	21	,	,	PUNCT
easat-3242	133	22	vgg19	vgg19	PROPN
easat-3242	133	23	is	be	AUX
easat-3242	133	24	able	able	ADJ
easat-3242	133	25	to	to	PART
easat-3242	133	26	automatically	automatically	ADV
easat-3242	133	27	learn	learn	VERB
easat-3242	133	28	a	a	DET
easat-3242	133	29	hierarchy	hierarchy	NOUN
easat-3242	133	30	of	of	ADP
easat-3242	133	31	features	feature	NOUN
easat-3242	133	32	from	from	ADP
easat-3242	133	33	simple	simple	ADJ
easat-3242	133	34	edge	edge	NOUN
easat-3242	133	35	patterns	pattern	NOUN
easat-3242	133	36	at	at	ADP
easat-3242	133	37	lower	low	ADJ
easat-3242	133	38	levels	level	NOUN
easat-3242	133	39	to	to	ADP
easat-3242	133	40	more	more	ADV
easat-3242	133	41	sophisticated	sophisticated	ADJ
easat-3242	133	42	texture	texture	ADJ
easat-3242	133	43	representations	representation	NOUN
easat-3242	133	44	at	at	ADP
easat-3242	133	45	deeper	deep	ADJ
easat-3242	133	46	layers	layer	NOUN
easat-3242	133	47	.	.	PUNCT
easat-3242	134	1	each	each	PRON
easat-3242	134	2	after	after	ADP
easat-3242	134	3	each	each	DET
easat-3242	134	4	convolutional	convolutional	ADJ
easat-3242	134	5	block	block	NOUN
easat-3242	134	6	,	,	PUNCT
easat-3242	134	7	we	we	PRON
easat-3242	134	8	have	have	VERB
easat-3242	134	9	a	a	DET
easat-3242	134	10	max	max	ADV
easat-3242	134	11	-	-	PUNCT
easat-3242	134	12	pooling	pool	VERB
easat-3242	134	13	layer	layer	NOUN
easat-3242	134	14	that	that	PRON
easat-3242	134	15	is	be	AUX
easat-3242	134	16	used	use	VERB
easat-3242	134	17	to	to	PART
easat-3242	134	18	reduce	reduce	VERB
easat-3242	134	19	the	the	DET
easat-3242	134	20	dimensionality	dimensionality	NOUN
easat-3242	134	21	of	of	ADP
easat-3242	134	22	the	the	DET
easat-3242	134	23	feature	feature	NOUN
easat-3242	134	24	maps	map	NOUN
easat-3242	134	25	but	but	CCONJ
easat-3242	134	26	retaining	retain	VERB
easat-3242	134	27	relevant	relevant	ADJ
easat-3242	134	28	information	information	NOUN
easat-3242	134	29	.	.	PUNCT
easat-3242	135	1	this	this	DET
easat-3242	135	2	architecture	architecture	NOUN
easat-3242	135	3	makes	make	VERB
easat-3242	135	4	vgg19	vgg19	PRON
easat-3242	135	5	very	very	ADV
easat-3242	135	6	good	good	ADJ
easat-3242	135	7	at	at	ADP
easat-3242	135	8	image	image	NOUN
easat-3242	135	9	recognition	recognition	NOUN
easat-3242	135	10	and	and	CCONJ
easat-3242	135	11	is	be	AUX
easat-3242	135	12	widely	widely	ADV
easat-3242	135	13	used	use	VERB
easat-3242	135	14	in	in	ADP
easat-3242	135	15	tasks	task	NOUN
easat-3242	135	16	like	like	ADP
easat-3242	135	17	object	object	NOUN
easat-3242	135	18	detection	detection	NOUN
easat-3242	135	19	,	,	PUNCT
easat-3242	135	20	image	image	NOUN
easat-3242	135	21	classification	classification	NOUN
easat-3242	135	22	etc	etc	NOUN
easat-3242	135	23	.	.	X
easat-3242	136	1	3.3.2	3.3.2	X
easat-3242	136	2	.	.	X
easat-3242	136	3	maxpooling	maxpoole	VERB
easat-3242	136	4	layers	layer	NOUN
easat-3242	136	5	maxpooling	maxpoole	VERB
easat-3242	136	6	maxpooling	maxpoole	VERB
easat-3242	136	7	layers	layer	NOUN
easat-3242	136	8	are	be	AUX
easat-3242	136	9	an	an	DET
easat-3242	136	10	essential	essential	ADJ
easat-3242	136	11	ingredient	ingredient	NOUN
easat-3242	136	12	of	of	ADP
easat-3242	136	13	the	the	DET
easat-3242	136	14	vgg19	vgg19	NOUN
easat-3242	136	15	architecture	architecture	NOUN
easat-3242	136	16	,	,	PUNCT
easat-3242	136	17	responsible	responsible	ADJ
easat-3242	136	18	for	for	ADP
easat-3242	136	19	down	down	ADV
easat-3242	136	20	-	-	PUNCT
easat-3242	136	21	sampling	sample	VERB
easat-3242	136	22	feature	feature	NOUN
easat-3242	136	23	maps	map	NOUN
easat-3242	136	24	along	along	ADP
easat-3242	136	25	their	their	PRON
easat-3242	136	26	spatial	spatial	ADJ
easat-3242	136	27	dimensions	dimension	NOUN
easat-3242	136	28	while	while	SCONJ
easat-3242	136	29	maintaining	maintain	VERB
easat-3242	136	30	only	only	ADV
easat-3242	136	31	their	their	PRON
easat-3242	136	32	most	most	ADV
easat-3242	136	33	prominent	prominent	ADJ
easat-3242	136	34	features	feature	NOUN
easat-3242	136	35	.	.	PUNCT
easat-3242	137	1	max	max	PROPN
easat-3242	137	2	-	-	ADJ
easat-3242	137	3	polling	polling	NOUN
easat-3242	137	4	layers	layer	NOUN
easat-3242	137	5	are	be	AUX
easat-3242	137	6	used	use	VERB
easat-3242	137	7	for	for	ADP
easat-3242	137	8	sub	sub	NOUN
easat-3242	137	9	-	-	ADJ
easat-3242	137	10	sampling	sample	VERB
easat-3242	137	11	the	the	DET
easat-3242	137	12	feature	feature	NOUN
easat-3242	137	13	maps	map	NOUN
easat-3242	137	14	after	after	ADP
easat-3242	137	15	every	every	DET
easat-3242	137	16	convolutional	convolutional	ADJ
easat-3242	137	17	block	block	NOUN
easat-3242	137	18	.	.	PUNCT
easat-3242	138	1	this	this	DET
easat-3242	138	2	layer	layer	NOUN
easat-3242	138	3	works	work	VERB
easat-3242	138	4	by	by	ADP
easat-3242	138	5	taking	take	VERB
easat-3242	138	6	the	the	DET
easat-3242	138	7	max	max	PROPN
easat-3242	138	8	value	value	NOUN
easat-3242	138	9	from	from	ADP
easat-3242	138	10	a	a	DET
easat-3242	138	11	2×2	2×2	NUM
easat-3242	138	12	region	region	NOUN
easat-3242	138	13	in	in	ADP
easat-3242	138	14	the	the	DET
easat-3242	138	15	feature	feature	NOUN
easat-3242	138	16	map	map	NOUN
easat-3242	138	17	hence	hence	ADV
easat-3242	138	18	reducing	reduce	VERB
easat-3242	138	19	its	its	PRON
easat-3242	138	20	height	height	NOUN
easat-3242	138	21	and	and	CCONJ
easat-3242	138	22	weight	weight	NOUN
easat-3242	138	23	to	to	ADP
easat-3242	138	24	half	half	NOUN
easat-3242	138	25	.	.	PUNCT
easat-3242	139	1	the	the	DET
easat-3242	139	2	max	max	PROPN
easat-3242	139	3	-	-	PUNCT
easat-3242	139	4	pooling	pool	VERB
easat-3242	139	5	layer	layer	NOUN
easat-3242	139	6	,	,	PUNCT
easat-3242	139	7	by	by	ADP
easat-3242	139	8	downsampling	downsample	VERB
easat-3242	139	9	the	the	DET
easat-3242	139	10	feature	feature	NOUN
easat-3242	139	11	maps	map	NOUN
easat-3242	139	12	for	for	ADP
easat-3242	139	13	us	we	PRON
easat-3242	139	14	,	,	PUNCT
easat-3242	139	15	greatly	greatly	ADV
easat-3242	139	16	reduces	reduce	VERB
easat-3242	139	17	the	the	DET
easat-3242	139	18	computational	computational	ADJ
easat-3242	139	19	load	load	NOUN
easat-3242	139	20	on	on	ADP
easat-3242	139	21	our	our	PRON
easat-3242	139	22	network	network	NOUN
easat-3242	139	23	and	and	CCONJ
easat-3242	139	24	at	at	ADP
easat-3242	139	25	the	the	DET
easat-3242	139	26	same	same	ADJ
easat-3242	139	27	time	time	NOUN
easat-3242	139	28	helps	help	VERB
easat-3242	139	29	in	in	ADP
easat-3242	139	30	retaining	retain	VERB
easat-3242	139	31	key	key	ADJ
easat-3242	139	32	features	feature	NOUN
easat-3242	139	33	required	require	VERB
easat-3242	139	34	to	to	PART
easat-3242	139	35	classify	classify	VERB
easat-3242	139	36	any	any	DET
easat-3242	139	37	image	image	NOUN
easat-3242	139	38	correctly	correctly	ADV
easat-3242	139	39	.	.	PUNCT
easat-3242	140	1	this	this	DET
easat-3242	140	2	process	process	NOUN
easat-3242	140	3	helps	help	VERB
easat-3242	140	4	by	by	ADP
easat-3242	140	5	making	make	VERB
easat-3242	140	6	the	the	DET
easat-3242	140	7	network	network	NOUN
easat-3242	140	8	learn	learn	VERB
easat-3242	140	9	spatial	spatial	ADJ
easat-3242	140	10	hierarchies	hierarchy	NOUN
easat-3242	140	11	again	again	ADV
easat-3242	140	12	and	and	CCONJ
easat-3242	140	13	makes	make	VERB
easat-3242	140	14	it	it	PRON
easat-3242	140	15	generalize	generalize	VERB
easat-3242	140	16	well	well	ADV
easat-3242	140	17	across	across	ADP
easat-3242	140	18	different	different	ADJ
easat-3242	140	19	image	image	NOUN
easat-3242	140	20	scales	scale	NOUN
easat-3242	140	21	.	.	PUNCT
easat-3242	141	1	using	use	VERB
easat-3242	141	2	max	max	PROPN
easat-3242	141	3	-	-	PUNCT
easat-3242	141	4	pooling	pooling	NOUN
easat-3242	141	5	also	also	ADV
easat-3242	141	6	results	result	VERB
easat-3242	141	7	in	in	ADP
easat-3242	141	8	more	more	ADJ
easat-3242	141	9	translation	translation	NOUN
easat-3242	141	10	invariance	invariance	NOUN
easat-3242	141	11	,	,	PUNCT
easat-3242	141	12	i.e.	i.e.	X
easat-3242	141	13	,	,	PUNCT
easat-3242	141	14	the	the	DET
easat-3242	141	15	network	network	NOUN
easat-3242	141	16	becomes	become	VERB
easat-3242	141	17	less	less	ADV
easat-3242	141	18	sensitive	sensitive	ADJ
easat-3242	141	19	to	to	ADP
easat-3242	141	20	where	where	SCONJ
easat-3242	141	21	objects	object	NOUN
easat-3242	141	22	are	be	AUX
easat-3242	141	23	located	locate	VERB
easat-3242	141	24	(	(	PUNCT
easat-3242	141	25	but	but	CCONJ
easat-3242	141	26	it	it	PRON
easat-3242	141	27	still	still	ADV
easat-3242	141	28	has	have	VERB
easat-3242	141	29	no	no	DET
easat-3242	141	30	idea	idea	NOUN
easat-3242	141	31	what	what	DET
easat-3242	141	32	object	object	NOUN
easat-3242	141	33	is	be	AUX
easat-3242	141	34	there	there	ADV
easat-3242	141	35	)	)	PUNCT
easat-3242	141	36	.	.	PUNCT
easat-3242	142	1	so	so	ADV
easat-3242	142	2	max	max	PROPN
easat-3242	142	3	-	-	PUNCT
easat-3242	142	4	pooling	pooling	NOUN
easat-3242	142	5	helps	help	VERB
easat-3242	142	6	to	to	PART
easat-3242	142	7	make	make	VERB
easat-3242	142	8	our	our	PRON
easat-3242	142	9	feature	feature	NOUN
easat-3242	142	10	robust	robust	ADJ
easat-3242	142	11	and	and	CCONJ
easat-3242	142	12	discriminatory	discriminatory	ADJ
easat-3242	142	13	by	by	ADP
easat-3242	142	14	only	only	ADV
easat-3242	142	15	focusing	focus	VERB
easat-3242	142	16	on	on	ADP
easat-3242	142	17	like	like	INTJ
easat-3242	142	18	important	important	ADJ
easat-3242	142	19	features	feature	NOUN
easat-3242	142	20	:	:	PUNCT
easat-3242	142	21	things	thing	NOUN
easat-3242	142	22	such	such	ADJ
easat-3242	142	23	as	as	ADP
easat-3242	142	24	edges	edge	NOUN
easat-3242	142	25	or	or	CCONJ
easat-3242	142	26	shapes	shape	NOUN
easat-3242	142	27	etc	etc	X
easat-3242	142	28	.	.	X
easat-3242	143	1	vgg19	vgg19	PROPN
easat-3242	143	2	utilised	utilise	VERB
easat-3242	143	3	max	max	PROPN
easat-3242	143	4	-	-	PUNCT
easat-3242	143	5	pooling	pool	VERB
easat-3242	143	6	layers	layer	NOUN
easat-3242	143	7	between	between	ADP
easat-3242	143	8	convolutional	convolutional	ADJ
easat-3242	143	9	blocks	block	NOUN
easat-3242	143	10	to	to	PART
easat-3242	143	11	enable	enable	VERB
easat-3242	143	12	the	the	DET
easat-3242	143	13	network	network	NOUN
easat-3242	143	14	to	to	PART
easat-3242	143	15	induce	induce	VERB
easat-3242	143	16	spatial	spatial	ADJ
easat-3242	143	17	invariance	invariance	NOUN
easat-3242	143	18	over	over	ADP
easat-3242	143	19	higher	high	ADJ
easat-3242	143	20	and	and	CCONJ
easat-3242	143	21	deeper	deep	ADJ
easat-3242	143	22	feature	feature	NOUN
easat-3242	143	23	space	space	NOUN
easat-3242	143	24	i	i	PRON
easat-3242	143	25	formation	formation	VERB
easat-3242	143	26	.	.	PUNCT
easat-3242	144	1	this	this	DET
easat-3242	144	2	balance	balance	NOUN
easat-3242	144	3	of	of	ADP
easat-3242	144	4	dimensionality	dimensionality	NOUN
easat-3242	144	5	reduction	reduction	NOUN
easat-3242	144	6	and	and	CCONJ
easat-3242	144	7	feature	feature	NOUN
easat-3242	144	8	extraction	extraction	NOUN
easat-3242	144	9	allows	allow	VERB
easat-3242	144	10	vgg19	vgg19	PRON
easat-3242	144	11	to	to	PART
easat-3242	144	12	work	work	VERB
easat-3242	144	13	well	well	ADV
easat-3242	144	14	with	with	ADP
easat-3242	144	15	large	large	ADJ
easat-3242	144	16	,	,	PUNCT
easat-3242	144	17	highdimensional	highdimensional	ADJ
easat-3242	144	18	data	datum	NOUN
easat-3242	144	19	sets	set	NOUN
easat-3242	144	20	which	which	PRON
easat-3242	144	21	we	we	PRON
easat-3242	144	22	have	have	VERB
easat-3242	144	23	in	in	ADP
easat-3242	144	24	visual	visual	ADJ
easat-3242	144	25	tasks	task	NOUN
easat-3242	144	26	.	.	PUNCT
easat-3242	145	1	3.3.3	3.3.3	X
easat-3242	145	2	.	.	PUNCT
easat-3242	145	3	fully	fully	ADV
easat-3242	145	4	connected	connected	ADJ
easat-3242	145	5	layers	layer	NOUN
easat-3242	145	6	the	the	DET
easat-3242	145	7	fully	fully	ADV
easat-3242	145	8	connected	connected	ADJ
easat-3242	145	9	layers	layer	NOUN
easat-3242	145	10	in	in	ADP
easat-3242	145	11	vgg19	vgg19	PROPN
easat-3242	145	12	are	be	AUX
easat-3242	145	13	located	locate	VERB
easat-3242	145	14	at	at	ADP
easat-3242	145	15	the	the	DET
easat-3242	145	16	last	last	ADJ
easat-3242	145	17	stage	stage	NOUN
easat-3242	145	18	of	of	ADP
easat-3242	145	19	the	the	DET
easat-3242	145	20	neural	neural	ADJ
easat-3242	145	21	network	network	NOUN
easat-3242	145	22	,	,	PUNCT
easat-3242	145	23	and	and	CCONJ
easat-3242	145	24	they	they	PRON
easat-3242	145	25	utilize	utilize	VERB
easat-3242	145	26	features	feature	NOUN
easat-3242	145	27	which	which	PRON
easat-3242	145	28	were	be	AUX
easat-3242	145	29	extracted	extract	VERB
easat-3242	145	30	by	by	ADP
easat-3242	145	31	convolutional	convolutional	ADJ
easat-3242	145	32	stages	stage	NOUN
easat-3242	145	33	to	to	PART
easat-3242	145	34	make	make	VERB
easat-3242	145	35	decisions	decision	NOUN
easat-3242	145	36	on	on	ADP
easat-3242	145	37	if	if	SCONJ
easat-3242	145	38	a	a	DET
easat-3242	145	39	region	region	NOUN
easat-3242	145	40	belongs	belong	VERB
easat-3242	145	41	to	to	ADP
easat-3242	145	42	one	one	NUM
easat-3242	145	43	class	class	NOUN
easat-3242	145	44	or	or	CCONJ
easat-3242	145	45	not	not	PART
easat-3242	145	46	.	.	PUNCT
easat-3242	146	1	these	these	DET
easat-3242	146	2	layers	layer	NOUN
easat-3242	146	3	are	be	AUX
easat-3242	146	4	fully	fully	ADV
easat-3242	146	5	connected	connect	VERB
easat-3242	146	6	,	,	PUNCT
easat-3242	146	7	which	which	PRON
easat-3242	146	8	means	mean	VERB
easat-3242	146	9	there	there	PRON
easat-3242	146	10	is	be	VERB
easat-3242	146	11	a	a	DET
easat-3242	146	12	connection	connection	NOUN
easat-3242	146	13	from	from	ADP
easat-3242	146	14	each	each	DET
easat-3242	146	15	neuron	neuron	NOUN
easat-3242	146	16	in	in	ADP
easat-3242	146	17	one	one	NUM
easat-3242	146	18	layer	layer	NOUN
easat-3242	146	19	to	to	ADP
easat-3242	146	20	all	all	DET
easat-3242	146	21	neurons	neuron	NOUN
easat-3242	146	22	on	on	ADP
easat-3242	146	23	the	the	DET
easat-3242	146	24	next	next	ADJ
easat-3242	146	25	.	.	PUNCT
easat-3242	147	1	this	this	DET
easat-3242	147	2	dense	dense	ADJ
easat-3242	147	3	connectivity	connectivity	NOUN
easat-3242	147	4	helps	help	VERB
easat-3242	147	5	the	the	DET
easat-3242	147	6	network	network	NOUN
easat-3242	147	7	to	to	PART
easat-3242	147	8	aggregate	aggregate	VERB
easat-3242	147	9	and	and	CCONJ
easat-3242	147	10	make	make	VERB
easat-3242	147	11	sense	sense	NOUN
easat-3242	147	12	of	of	ADP
easat-3242	147	13	hierarchical	hierarchical	ADJ
easat-3242	147	14	representations	representation	NOUN
easat-3242	147	15	learned	learn	VERB
easat-3242	147	16	by	by	ADP
easat-3242	147	17	convolutional	convolutional	ADJ
easat-3242	147	18	layers	layer	NOUN
easat-3242	147	19	.	.	PUNCT
easat-3242	148	1	the	the	DET
easat-3242	148	2	first	first	ADJ
easat-3242	148	3	two	two	NUM
easat-3242	148	4	fully	fully	ADV
easat-3242	148	5	connected	connected	ADJ
easat-3242	148	6	layers	layer	NOUN
easat-3242	148	7	in	in	ADP
easat-3242	148	8	vgg19	vgg19	PROPN
easat-3242	148	9	have	have	VERB
easat-3242	148	10	4096	4096	NUM
easat-3242	148	11	neurons	neuron	NOUN
easat-3242	148	12	each	each	PRON
easat-3242	148	13	,	,	PUNCT
easat-3242	148	14	and	and	CCONJ
easat-3242	148	15	act	act	VERB
easat-3242	148	16	to	to	PART
easat-3242	148	17	reduce	reduce	VERB
easat-3242	148	18	the	the	DET
easat-3242	148	19	high	high	ADJ
easat-3242	148	20	-	-	PUNCT
easat-3242	148	21	dimensional	dimensional	ADJ
easat-3242	148	22	feature	feature	NOUN
easat-3242	148	23	representations	representation	NOUN
easat-3242	148	24	into	into	ADP
easat-3242	148	25	a	a	DET
easat-3242	148	26	more	more	ADV
easat-3242	148	27	compressed	compress	VERB
easat-3242	148	28	form	form	NOUN
easat-3242	148	29	.	.	PUNCT
easat-3242	149	1	this	this	PRON
easat-3242	149	2	will	will	AUX
easat-3242	149	3	help	help	VERB
easat-3242	149	4	the	the	DET
easat-3242	149	5	network	network	NOUN
easat-3242	149	6	to	to	PART
easat-3242	149	7	learn	learn	VERB
easat-3242	149	8	patterns	pattern	NOUN
easat-3242	149	9	,	,	PUNCT
easat-3242	149	10	interactions	interaction	NOUN
easat-3242	149	11	between	between	ADP
easat-3242	149	12	features	feature	NOUN
easat-3242	149	13	which	which	PRON
easat-3242	149	14	are	be	AUX
easat-3242	149	15	important	important	ADJ
easat-3242	149	16	for	for	ADP
easat-3242	149	17	accurate	accurate	ADJ
easat-3242	149	18	prediction	prediction	NOUN
easat-3242	149	19	.	.	PUNCT
easat-3242	150	1	the	the	DET
easat-3242	150	2	final	final	ADJ
easat-3242	150	3	fully	fully	ADV
easat-3242	150	4	connected	connected	ADJ
easat-3242	150	5	layer	layer	NOUN
easat-3242	150	6	of	of	ADP
easat-3242	150	7	vgg	vgg	PROPN
easat-3242	150	8	19	19	NUM
easat-3242	150	9	is	be	AUX
easat-3242	150	10	the	the	DET
easat-3242	150	11	1000	1000	NUM
easat-3242	150	12	neurons	neuron	NOUN
easat-3242	150	13	,	,	PUNCT
easat-3242	150	14	which	which	PRON
easat-3242	150	15	corresponds	correspond	VERB
easat-3242	150	16	to	to	ADP
easat-3242	150	17	its	its	PRON
easat-3242	150	18	output	output	NOUN
easat-3242	150	19	image	image	NOUN
easat-3242	150	20	classes	class	NOUN
easat-3242	150	21	when	when	SCONJ
easat-3242	150	22	it	it	PRON
easat-3242	150	23	was	be	AUX
easat-3242	150	24	trained	train	VERB
easat-3242	150	25	with	with	ADP
easat-3242	150	26	imagenet	imagenet	NOUN
easat-3242	150	27	dataset	dataset	NOUN
easat-3242	150	28	.	.	PUNCT
easat-3242	151	1	the	the	DET
easat-3242	151	2	resulting	result	VERB
easat-3242	151	3	value	value	NOUN
easat-3242	151	4	is	be	AUX
easat-3242	151	5	a	a	DET
easat-3242	151	6	probability	probability	NOUN
easat-3242	151	7	distribution	distribution	NOUN
easat-3242	151	8	represented	represent	VERB
easat-3242	151	9	by	by	ADP
easat-3242	151	10	softmax	softmax	NOUN
easat-3242	151	11	function	function	NOUN
easat-3242	151	12	and	and	CCONJ
easat-3242	151	13	the	the	DET
easat-3242	151	14	sum	sum	NOUN
easat-3242	151	15	of	of	ADP
easat-3242	151	16	their	their	PRON
easat-3242	151	17	probabilities	probability	NOUN
easat-3242	151	18	will	will	AUX
easat-3242	151	19	be	be	AUX
easat-3242	151	20	1	1	NUM
easat-3242	151	21	.	.	PUNCT
easat-3242	152	1	it	it	PRON
easat-3242	152	2	helps	help	VERB
easat-3242	152	3	the	the	DET
easat-3242	152	4	model	model	NOUN
easat-3242	152	5	to	to	PART
easat-3242	152	6	give	give	VERB
easat-3242	152	7	a	a	DET
easat-3242	152	8	probability	probability	NOUN
easat-3242	152	9	score	score	NOUN
easat-3242	152	10	regarding	regard	VERB
easat-3242	152	11	each	each	DET
easat-3242	152	12	class	class	NOUN
easat-3242	152	13	that	that	SCONJ
easat-3242	152	14	how	how	SCONJ
easat-3242	152	15	much	much	ADJ
easat-3242	152	16	chance	chance	NOUN
easat-3242	152	17	is	be	AUX
easat-3242	152	18	there	there	PRON
easat-3242	152	19	for	for	ADP
easat-3242	152	20	an	an	DET
easat-3242	152	21	input	input	NOUN
easat-3242	152	22	image	image	NOUN
easat-3242	152	23	of	of	ADP
easat-3242	152	24	certain	certain	ADJ
easat-3242	152	25	category	category	NOUN
easat-3242	152	26	.	.	PUNCT
easat-3242	153	1	those	those	DET
easat-3242	153	2	fully	fully	ADV
easat-3242	153	3	connected	connected	ADJ
easat-3242	153	4	layers	layer	NOUN
easat-3242	153	5	are	be	AUX
easat-3242	153	6	essential	essential	ADJ
easat-3242	153	7	and	and	CCONJ
easat-3242	153	8	take	take	VERB
easat-3242	153	9	all	all	DET
easat-3242	153	10	the	the	DET
easat-3242	153	11	learned	learn	VERB
easat-3242	153	12	features	feature	NOUN
easat-3242	153	13	convert	convert	VERB
easat-3242	153	14	them	they	PRON
easat-3242	153	15	into	into	ADP
easat-3242	153	16	one	one	NUM
easat-3242	153	17	of	of	ADP
easat-3242	153	18	most	most	ADJ
easat-3242	153	19	significant	significant	ADJ
easat-3242	153	20	predictions	prediction	NOUN
easat-3242	153	21	,	,	PUNCT
easat-3242	153	22	allowing	allow	VERB
easat-3242	153	23	vgg19	vgg19	PRON
easat-3242	153	24	to	to	PART
easat-3242	153	25	perform	perform	VERB
easat-3242	153	26	well	well	ADV
easat-3242	153	27	in	in	ADP
easat-3242	153	28	tasks	task	NOUN
easat-3242	153	29	like	like	ADP
easat-3242	153	30	image	image	NOUN
easat-3242	153	31	classification	classification	NOUN
easat-3242	153	32	or	or	CCONJ
easat-3242	153	33	object	object	NOUN
easat-3242	153	34	detection	detection	NOUN
easat-3242	153	35	.	.	PUNCT
easat-3242	154	1	3.3.4	3.3.4	X
easat-3242	154	2	.	.	PUNCT
easat-3242	154	3	activation	activation	NOUN
easat-3242	154	4	function	function	NOUN
easat-3242	154	5	(	(	PUNCT
easat-3242	154	6	relu	relu	NOUN
easat-3242	154	7	)	)	PUNCT
easat-3242	154	8	vgg19	vgg19	PROPN
easat-3242	154	9	uses	use	VERB
easat-3242	154	10	relu	relu	NOUN
easat-3242	154	11	(	(	PUNCT
easat-3242	154	12	rectified	rectified	ADJ
easat-3242	154	13	linear	linear	NOUN
easat-3242	154	14	unit	unit	NOUN
easat-3242	154	15	)	)	PUNCT
easat-3242	154	16	as	as	ADP
easat-3242	154	17	an	an	DET
easat-3242	154	18	activation	activation	NOUN
easat-3242	154	19	function	function	NOUN
easat-3242	154	20	in	in	ADP
easat-3242	154	21	each	each	DET
easat-3242	154	22	layer	layer	NOUN
easat-3242	154	23	of	of	ADP
easat-3242	154	24	the	the	DET
easat-3242	154	25	convolutional	convolutional	ADJ
easat-3242	154	26	and	and	CCONJ
easat-3242	154	27	fully	fully	ADV
easat-3242	154	28	connected	connected	ADJ
easat-3242	154	29	layers	layer	NOUN
easat-3242	154	30	excluding	exclude	VERB
easat-3242	154	31	output	output	NOUN
easat-3242	154	32	.	.	PUNCT
easat-3242	155	1	the	the	DET
easat-3242	155	2	relu	relu	NOUN
easat-3242	155	3	introduces	introduce	VERB
easat-3242	155	4	non	non	ADJ
easat-3242	155	5	-	-	NOUN
easat-3242	155	6	linearity	linearity	NOUN
easat-3242	155	7	in	in	ADP
easat-3242	155	8	the	the	DET
easat-3242	155	9	network	network	NOUN
easat-3242	155	10	which	which	PRON
easat-3242	155	11	is	be	AUX
easat-3242	155	12	crucial	crucial	ADJ
easat-3242	155	13	while	while	SCONJ
easat-3242	155	14	learning	learn	VERB
easat-3242	155	15	complex	complex	ADJ
easat-3242	155	16	patterns	pattern	NOUN
easat-3242	155	17	from	from	ADP
easat-3242	155	18	data	datum	NOUN
easat-3242	155	19	.	.	PUNCT
easat-3242	156	1	it	it	PRON
easat-3242	156	2	is	be	AUX
easat-3242	156	3	defined	define	VERB
easat-3242	156	4	as	as	ADP
easat-3242	156	5	f(x	f(x	PROPN
easat-3242	156	6	)	)	PUNCT
easat-3242	157	1	=	=	SYM
easat-3242	157	2	max	max	PROPN
easat-3242	157	3	(	(	PUNCT
easat-3242	157	4	0,x	0,x	PROPN
easat-3242	157	5	)	)	PUNCT
easat-3242	157	6	,	,	PUNCT
easat-3242	157	7	which	which	PRON
easat-3242	157	8	simply	simply	ADV
easat-3242	157	9	sets	set	VERB
easat-3242	157	10	any	any	DET
easat-3242	157	11	positive	positive	ADJ
easat-3242	157	12	values	value	NOUN
easat-3242	157	13	to	to	PART
easat-3242	157	14	directly	directly	ADV
easat-3242	157	15	be	be	AUX
easat-3242	157	16	that	that	SCONJ
easat-3242	157	17	'	'	PUNCT
easat-3242	157	18	x	x	X
easat-3242	157	19	'	'	X
easat-3242	157	20	,	,	PUNCT
easat-3242	157	21	or	or	CCONJ
easat-3242	157	22	zero	zero	NUM
easat-3242	157	23	otherwise	otherwise	ADV
easat-3242	157	24	.	.	PUNCT
easat-3242	158	1	this	this	DET
easat-3242	158	2	complex	complex	ADJ
easat-3242	158	3	yet	yet	CCONJ
easat-3242	158	4	intuitive	intuitive	ADJ
easat-3242	158	5	concept	concept	NOUN
easat-3242	158	6	prevents	prevent	VERB
easat-3242	158	7	the	the	DET
easat-3242	158	8	vanishing	vanish	VERB
easat-3242	158	9	gradient	gradient	NOUN
easat-3242	158	10	problem	problem	NOUN
easat-3242	158	11	,	,	PUNCT
easat-3242	158	12	which	which	PRON
easat-3242	158	13	will	will	AUX
easat-3242	158	14	not	not	PART
easat-3242	158	15	be	be	AUX
easat-3242	158	16	discussed	discuss	VERB
easat-3242	158	17	here	here	ADV
easat-3242	158	18	;	;	PUNCT
easat-3242	158	19	in	in	ADP
easat-3242	158	20	short	short	ADJ
easat-3242	158	21	it	it	PRON
easat-3242	158	22	is	be	AUX
easat-3242	158	23	just	just	ADV
easat-3242	158	24	that	that	SCONJ
easat-3242	158	25	we	we	PRON
easat-3242	158	26	would	would	AUX
easat-3242	158	27	like	like	VERB
easat-3242	158	28	to	to	PART
easat-3242	158	29	avoid	avoid	VERB
easat-3242	158	30	cases	case	NOUN
easat-3242	158	31	where	where	SCONJ
easat-3242	158	32	gradients	gradient	NOUN
easat-3242	158	33	become	become	VERB
easat-3242	158	34	too	too	ADV
easat-3242	158	35	small	small	ADJ
easat-3242	158	36	values	value	NOUN
easat-3242	158	37	or	or	CCONJ
easat-3242	158	38	infinitesimally	infinitesimally	ADV
easat-3242	158	39	close	close	ADJ
easat-3242	158	40	such	such	ADJ
easat-3242	158	41	that	that	SCONJ
easat-3242	158	42	they	they	PRON
easat-3242	158	43	have	have	VERB
easat-3242	158	44	no	no	DET
easat-3242	158	45	effect	effect	NOUN
easat-3242	158	46	when	when	SCONJ
easat-3242	158	47	updating	update	VERB
easat-3242	158	48	weights	weight	NOUN
easat-3242	158	49	happens	happen	VERB
easat-3242	158	50	during	during	ADP
easat-3242	158	51	back	back	ADJ
easat-3242	158	52	propagation	propagation	NOUN
easat-3242	158	53	.	.	PUNCT
easat-3242	159	1	the	the	DET
easat-3242	159	2	use	use	NOUN
easat-3242	159	3	of	of	ADP
easat-3242	159	4	relu	relu	NOUN
easat-3242	159	5	in	in	ADP
easat-3242	159	6	vgg19	vgg19	PROPN
easat-3242	159	7	speeds	speed	VERB
easat-3242	159	8	up	up	ADP
easat-3242	159	9	the	the	DET
easat-3242	159	10	convergence	convergence	NOUN
easat-3242	159	11	during	during	ADP
easat-3242	159	12	training	training	NOUN
easat-3242	159	13	as	as	SCONJ
easat-3242	159	14	it	it	PRON
easat-3242	159	15	helps	helps	AUX
easat-3242	159	16	learn	learn	VERB
easat-3242	159	17	faster	fast	ADV
easat-3242	159	18	.	.	PUNCT
easat-3242	160	1	unlike	unlike	ADP
easat-3242	160	2	sigmoid	sigmoid	NOUN
easat-3242	160	3	or	or	CCONJ
easat-3242	160	4	tanh	tanh	NOUN
easat-3242	160	5	,	,	PUNCT
easat-3242	160	6	relu	relu	NOUN
easat-3242	160	7	does	do	AUX
easat-3242	160	8	not	not	PART
easat-3242	160	9	saturate	saturate	VERB
easat-3242	160	10	which	which	PRON
easat-3242	160	11	in	in	ADP
easat-3242	160	12	other	other	ADJ
easat-3242	160	13	words	word	NOUN
easat-3242	160	14	mean	mean	VERB
easat-3242	160	15	that	that	SCONJ
easat-3242	160	16	the	the	DET
easat-3242	160	17	gradients	gradient	NOUN
easat-3242	160	18	can	can	AUX
easat-3242	160	19	flow	flow	VERB
easat-3242	160	20	through	through	ADP
easat-3242	160	21	neurons	neuron	NOUN
easat-3242	160	22	even	even	ADV
easat-3242	160	23	when	when	SCONJ
easat-3242	160	24	using	use	VERB
easat-3242	160	25	large	large	ADJ
easat-3242	160	26	inputs	input	NOUN
easat-3242	160	27	.	.	PUNCT
easat-3242	161	1	this	this	PRON
easat-3242	161	2	makes	make	VERB
easat-3242	161	3	the	the	DET
easat-3242	161	4	training	training	NOUN
easat-3242	161	5	more	more	ADV
easat-3242	161	6	efficient	efficient	ADJ
easat-3242	161	7	and	and	CCONJ
easat-3242	161	8	less	less	ADV
easat-3242	161	9	susceptible	susceptible	ADJ
easat-3242	161	10	to	to	ADP
easat-3242	161	11	getting	getting	AUX
easat-3242	161	12	stuck	stick	VERB
easat-3242	161	13	.	.	PUNCT
easat-3242	162	1	relu	relu	NOUN
easat-3242	162	2	is	be	AUX
easat-3242	162	3	now	now	ADV
easat-3242	162	4	one	one	NUM
easat-3242	162	5	of	of	ADP
easat-3242	162	6	the	the	DET
easat-3242	162	7	most	most	ADV
easat-3242	162	8	used	used	ADJ
easat-3242	162	9	bits	bit	NOUN
easat-3242	162	10	and	and	CCONJ
easat-3242	162	11	pieces	piece	NOUN
easat-3242	162	12	in	in	ADP
easat-3242	162	13	this	this	DET
easat-3242	162	14	jigsaw	jigsaw	NOUN
easat-3242	162	15	,	,	PUNCT
easat-3242	162	16	its	its	PRON
easat-3242	162	17	5713	5713	NUM
easat-3242	162	18	edelweiss	edelweiss	PROPN
easat-3242	162	19	applied	apply	VERB
easat-3242	162	20	science	science	NOUN
easat-3242	162	21	and	and	CCONJ
easat-3242	162	22	technology	technology	NOUN
easat-3242	162	23	issn	issn	PROPN
easat-3242	162	24	:	:	PUNCT
easat-3242	162	25	2576	2576	NUM
easat-3242	162	26	-	-	SYM
easat-3242	162	27	8484	8484	NUM
easat-3242	162	28	vol	vol	NOUN
easat-3242	162	29	.	.	PROPN
easat-3242	163	1	8	8	NUM
easat-3242	163	2	,	,	PUNCT
easat-3242	163	3	no	no	INTJ
easat-3242	163	4	.	.	NOUN
easat-3242	164	1	6	6	NUM
easat-3242	164	2	:	:	SYM
easat-3242	164	3	5707	5707	NUM
easat-3242	164	4	-	-	SYM
easat-3242	164	5	5722	5722	NUM
easat-3242	164	6	,	,	PUNCT
easat-3242	164	7	2024	2024	NUM
easat-3242	164	8	doi	doi	NOUN
easat-3242	164	9	:	:	PUNCT
easat-3242	164	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	164	11	©	©	PROPN
easat-3242	164	12	2024	2024	NUM
easat-3242	164	13	by	by	ADP
easat-3242	164	14	the	the	DET
easat-3242	164	15	authors	author	NOUN
easat-3242	164	16	;	;	PUNCT
easat-3242	164	17	licensee	licensee	PROPN
easat-3242	164	18	learning	learning	NOUN
easat-3242	164	19	gate	gate	PROPN
easat-3242	164	20	simplicity	simplicity	NOUN
easat-3242	164	21	and	and	CCONJ
easat-3242	164	22	effectiveness	effectiveness	NOUN
easat-3242	164	23	can	can	AUX
easat-3242	164	24	be	be	AUX
easat-3242	164	25	understood	understand	VERB
easat-3242	164	26	through	through	ADP
easat-3242	164	27	vgg19	vgg19	PRON
easat-3242	164	28	where	where	SCONJ
easat-3242	164	29	it	it	PRON
easat-3242	164	30	has	have	VERB
easat-3242	164	31	tremendous	tremendous	ADJ
easat-3242	164	32	success	success	NOUN
easat-3242	164	33	when	when	SCONJ
easat-3242	164	34	applied	apply	VERB
easat-3242	164	35	to	to	ADP
easat-3242	164	36	image	image	NOUN
easat-3242	164	37	recognition	recognition	NOUN
easat-3242	164	38	/	/	SYM
easat-3242	164	39	classification	classification	NOUN
easat-3242	164	40	etc	etc	X
easat-3242	164	41	.	.	PUNCT
easat-3242	164	42	relu	relu	NOUN
easat-3242	164	43	,	,	PUNCT
easat-3242	164	44	by	by	ADP
easat-3242	164	45	injecting	inject	VERB
easat-3242	164	46	the	the	DET
easat-3242	164	47	layer	layer	NOUN
easat-3242	164	48	with	with	ADP
easat-3242	164	49	non	non	ADJ
easat-3242	164	50	-	-	ADJ
easat-3242	164	51	linearity	linearity	ADJ
easat-3242	164	52	helps	help	VERB
easat-3242	164	53	in	in	ADP
easat-3242	164	54	functionality	functionality	NOUN
easat-3242	164	55	making	make	VERB
easat-3242	164	56	it	it	PRON
easat-3242	164	57	easier	easy	ADJ
easat-3242	164	58	for	for	ADP
easat-3242	164	59	modelling	modelling	NOUN
easat-3242	164	60	to	to	PART
easat-3242	164	61	learn	learn	VERB
easat-3242	164	62	and	and	CCONJ
easat-3242	164	63	represent	represent	VERB
easat-3242	164	64	complex	complex	ADJ
easat-3242	164	65	features	feature	NOUN
easat-3242	164	66	of	of	ADP
easat-3242	164	67	data	datum	NOUN
easat-3242	164	68	.	.	PUNCT
easat-3242	165	1	3.3.5	3.3.5	X
easat-3242	165	2	.	.	X
easat-3242	165	3	dropout	dropout	NOUN
easat-3242	165	4	layer	layer	NOUN
easat-3242	165	5	vgg19	vgg19	PROPN
easat-3242	165	6	uses	use	VERB
easat-3242	165	7	a	a	DET
easat-3242	165	8	dropout	dropout	NOUN
easat-3242	165	9	to	to	PART
easat-3242	165	10	prevent	prevent	VERB
easat-3242	165	11	overfitting	overfitting	NOUN
easat-3242	165	12	,	,	PUNCT
easat-3242	165	13	which	which	PRON
easat-3242	165	14	means	mean	VERB
easat-3242	165	15	the	the	DET
easat-3242	165	16	network	network	NOUN
easat-3242	165	17	is	be	AUX
easat-3242	165	18	good	good	ADJ
easat-3242	165	19	in	in	ADP
easat-3242	165	20	training	training	NOUN
easat-3242	165	21	data	datum	NOUN
easat-3242	165	22	but	but	CCONJ
easat-3242	165	23	bad	bad	ADJ
easat-3242	165	24	on	on	ADP
easat-3242	165	25	new	new	ADJ
easat-3242	165	26	unseen	unseen	ADJ
easat-3242	165	27	test	test	NOUN
easat-3242	165	28	data	datum	NOUN
easat-3242	165	29	.	.	PUNCT
easat-3242	166	1	vgg19	vgg19	PROPN
easat-3242	166	2	uses	use	VERB
easat-3242	166	3	dropout	dropout	NOUN
easat-3242	166	4	between	between	ADP
easat-3242	166	5	the	the	DET
easat-3242	166	6	fully	fully	ADV
easat-3242	166	7	connected	connected	ADJ
easat-3242	166	8	layers	layer	NOUN
easat-3242	166	9	during	during	ADP
easat-3242	166	10	training	training	NOUN
easat-3242	166	11	.	.	PUNCT
easat-3242	167	1	randomly	randomly	ADV
easat-3242	167	2	"	"	PUNCT
easat-3242	167	3	drops	drop	VERB
easat-3242	167	4	out	out	ADP
easat-3242	167	5	"	"	PUNCT
easat-3242	167	6	a	a	DET
easat-3242	167	7	percentage	percentage	NOUN
easat-3242	167	8	of	of	ADP
easat-3242	167	9	neurons	neuron	NOUN
easat-3242	167	10	in	in	ADP
easat-3242	167	11	layer	layer	NOUN
easat-3242	167	12	to	to	PART
easat-3242	167	13	allow	allow	VERB
easat-3242	167	14	the	the	DET
easat-3242	167	15	network	network	NOUN
easat-3242	167	16	to	to	PART
easat-3242	167	17	converge	converge	VERB
easat-3242	167	18	on	on	ADP
easat-3242	167	19	more	more	ADJ
easat-3242	167	20	robust	robust	ADJ
easat-3242	167	21	features	feature	NOUN
easat-3242	167	22	and	and	CCONJ
easat-3242	167	23	not	not	PART
easat-3242	167	24	rely	rely	VERB
easat-3242	167	25	too	too	ADV
easat-3242	167	26	heavily	heavily	ADV
easat-3242	167	27	on	on	ADP
easat-3242	167	28	specific	specific	ADJ
easat-3242	167	29	neurons	neuron	NOUN
easat-3242	167	30	.	.	PUNCT
easat-3242	168	1	this	this	PRON
easat-3242	168	2	makes	make	VERB
easat-3242	168	3	the	the	DET
easat-3242	168	4	model	model	NOUN
easat-3242	168	5	generalize	generalize	VERB
easat-3242	168	6	more	more	ADV
easat-3242	168	7	well	well	ADV
easat-3242	168	8	on	on	ADP
easat-3242	168	9	new	new	ADJ
easat-3242	168	10	data	datum	NOUN
easat-3242	168	11	as	as	SCONJ
easat-3242	168	12	it	it	PRON
easat-3242	168	13	can	can	AUX
easat-3242	168	14	not	not	PART
easat-3242	168	15	rely	rely	VERB
easat-3242	168	16	too	too	ADV
easat-3242	168	17	much	much	ADJ
easat-3242	168	18	on	on	ADP
easat-3242	168	19	a	a	DET
easat-3242	168	20	single	single	ADJ
easat-3242	168	21	neuron	neuron	NOUN
easat-3242	168	22	or	or	CCONJ
easat-3242	168	23	group	group	NOUN
easat-3242	168	24	of	of	ADP
easat-3242	168	25	neurons	neuron	NOUN
easat-3242	168	26	during	during	ADP
easat-3242	168	27	training	training	NOUN
easat-3242	168	28	.	.	PUNCT
easat-3242	169	1	dropout	dropout	NOUN
easat-3242	169	2	,	,	PUNCT
easat-3242	169	3	as	as	ADP
easat-3242	169	4	with	with	ADP
easat-3242	169	5	vgg19	vgg19	NOUN
easat-3242	169	6	here	here	ADV
easat-3242	169	7	,	,	PUNCT
easat-3242	169	8	has	have	AUX
easat-3242	169	9	increased	increase	VERB
easat-3242	169	10	in	in	ADP
easat-3242	169	11	importance	importance	NOUN
easat-3242	169	12	due	due	ADP
easat-3242	169	13	to	to	ADP
easat-3242	169	14	its	its	PRON
easat-3242	169	15	ability	ability	NOUN
easat-3242	169	16	to	to	PART
easat-3242	169	17	enhance	enhance	VERB
easat-3242	169	18	generalization	generalization	NOUN
easat-3242	169	19	performances	performance	NOUN
easat-3242	169	20	(	(	PUNCT
easat-3242	169	21	especially	especially	ADV
easat-3242	169	22	for	for	ADP
easat-3242	169	23	large	large	ADJ
easat-3242	169	24	datasets	dataset	NOUN
easat-3242	169	25	and	and	CCONJ
easat-3242	169	26	difficult	difficult	ADJ
easat-3242	169	27	patterns	pattern	NOUN
easat-3242	169	28	)	)	PUNCT
easat-3242	169	29	.	.	PUNCT
easat-3242	170	1	the	the	DET
easat-3242	170	2	dropout	dropout	NOUN
easat-3242	170	3	rate	rate	NOUN
easat-3242	170	4	(	(	PUNCT
easat-3242	170	5	usually	usually	ADV
easat-3242	170	6	between	between	ADP
easat-3242	170	7	0.2	0.2	NUM
easat-3242	170	8	and	and	CCONJ
easat-3242	170	9	0.5	0.5	NUM
easat-3242	170	10	)	)	PUNCT
easat-3242	170	11	specifies	specifie	NOUN
easat-3242	170	12	which	which	PRON
easat-3242	170	13	neurons	neuron	NOUN
easat-3242	170	14	to	to	PART
easat-3242	170	15	turn	turn	VERB
easat-3242	170	16	off	off	ADP
easat-3242	170	17	as	as	ADP
easat-3242	170	18	a	a	DET
easat-3242	170	19	fraction	fraction	NOUN
easat-3242	170	20	of	of	ADP
easat-3242	170	21	the	the	DET
easat-3242	170	22	total	total	ADJ
easat-3242	170	23	number	number	NOUN
easat-3242	170	24	of	of	ADP
easat-3242	170	25	that	that	DET
easat-3242	170	26	layer	layer	NOUN
easat-3242	170	27	’s	’s	PART
easat-3242	170	28	input	input	PROPN
easat-3242	170	29	neuron	neuron	PROPN
easat-3242	170	30	count	count	PROPN
easat-3242	170	31	.	.	PUNCT
easat-3242	171	1	dropout	dropout	NOUN
easat-3242	171	2	is	be	AUX
easat-3242	171	3	used	use	VERB
easat-3242	171	4	in	in	ADP
easat-3242	171	5	training	training	NOUN
easat-3242	171	6	mode	mode	NOUN
easat-3242	171	7	because	because	SCONJ
easat-3242	171	8	it	it	PRON
easat-3242	171	9	forces	force	VERB
easat-3242	171	10	the	the	DET
easat-3242	171	11	network	network	NOUN
easat-3242	171	12	to	to	PART
easat-3242	171	13	learn	learn	VERB
easat-3242	171	14	from	from	ADP
easat-3242	171	15	a	a	DET
easat-3242	171	16	broader	broad	ADJ
easat-3242	171	17	set	set	NOUN
easat-3242	171	18	of	of	ADP
easat-3242	171	19	features	feature	NOUN
easat-3242	171	20	mitigating	mitigate	VERB
easat-3242	171	21	overfitting	overfitting	NOUN
easat-3242	171	22	.	.	PUNCT
easat-3242	172	1	however	however	ADV
easat-3242	172	2	,	,	PUNCT
easat-3242	172	3	during	during	ADP
easat-3242	172	4	inference	inference	NOUN
easat-3242	172	5	(	(	PUNCT
easat-3242	172	6	when	when	SCONJ
easat-3242	172	7	the	the	DET
easat-3242	172	8	model	model	NOUN
easat-3242	172	9	is	be	AUX
easat-3242	172	10	being	be	AUX
easat-3242	172	11	used	use	VERB
easat-3242	172	12	for	for	ADP
easat-3242	172	13	predictions	prediction	NOUN
easat-3242	172	14	)	)	PUNCT
easat-3242	172	15	all	all	DET
easat-3242	172	16	neurons	neuron	NOUN
easat-3242	172	17	are	be	AUX
easat-3242	172	18	active	active	ADJ
easat-3242	172	19	and	and	CCONJ
easat-3242	172	20	so	so	ADV
easat-3242	172	21	it	it	PRON
easat-3242	172	22	can	can	AUX
easat-3242	172	23	make	make	VERB
easat-3242	172	24	full	full	ADJ
easat-3242	172	25	use	use	NOUN
easat-3242	172	26	of	of	ADP
easat-3242	172	27	its	its	PRON
easat-3242	172	28	capacity	capacity	NOUN
easat-3242	172	29	.	.	PUNCT
easat-3242	173	1	this	this	PRON
easat-3242	173	2	ensures	ensure	VERB
easat-3242	173	3	that	that	SCONJ
easat-3242	173	4	activations	activation	NOUN
easat-3242	173	5	takes	take	VERB
easat-3242	173	6	place	place	NOUN
easat-3242	173	7	much	much	ADV
easat-3242	173	8	faster	fast	ADV
easat-3242	173	9	and	and	CCONJ
easat-3242	173	10	is	be	AUX
easat-3242	173	11	also	also	ADV
easat-3242	173	12	more	more	ADV
easat-3242	173	13	robust	robust	ADJ
easat-3242	173	14	,	,	PUNCT
easat-3242	173	15	which	which	PRON
easat-3242	173	16	ultimately	ultimately	ADV
easat-3242	173	17	makes	make	VERB
easat-3242	173	18	vgg19	vgg19	PRON
easat-3242	173	19	a	a	DET
easat-3242	173	20	better	well	ADJ
easat-3242	173	21	performer	performer	NOUN
easat-3242	173	22	especially	especially	ADV
easat-3242	173	23	in	in	ADP
easat-3242	173	24	large	large	ADJ
easat-3242	173	25	image	image	NOUN
easat-3242	173	26	classification	classification	NOUN
easat-3242	173	27	problems	problem	NOUN
easat-3242	173	28	.	.	PUNCT
easat-3242	174	1	3.3.6	3.3.6	X
easat-3242	174	2	.	.	PUNCT
easat-3242	174	3	output	output	NOUN
easat-3242	174	4	layer	layer	NOUN
easat-3242	174	5	in	in	ADP
easat-3242	174	6	the	the	DET
easat-3242	174	7	vgg19	vgg19	PROPN
easat-3242	174	8	model	model	NOUN
easat-3242	174	9	,	,	PUNCT
easat-3242	174	10	this	this	PRON
easat-3242	174	11	is	be	AUX
easat-3242	174	12	called	call	VERB
easat-3242	174	13	the	the	DET
easat-3242	174	14	output	output	NOUN
easat-3242	174	15	layer	layer	NOUN
easat-3242	174	16	;	;	PUNCT
easat-3242	174	17	it	it	PRON
easat-3242	174	18	performs	perform	VERB
easat-3242	174	19	logic	logic	NOUN
easat-3242	174	20	to	to	PART
easat-3242	174	21	assess	assess	VERB
easat-3242	174	22	predictions	prediction	NOUN
easat-3242	174	23	(	(	PUNCT
easat-3242	174	24	features	feature	NOUN
easat-3242	174	25	)	)	PUNCT
easat-3242	174	26	extracted	extract	VERB
easat-3242	174	27	by	by	ADP
easat-3242	174	28	previous	previous	ADJ
easat-3242	174	29	layers	layer	NOUN
easat-3242	174	30	.	.	PUNCT
easat-3242	175	1	1000	1000	NUM
easat-3242	175	2	is	be	AUX
easat-3242	175	3	that	that	SCONJ
easat-3242	175	4	corresponding	correspond	VERB
easat-3242	175	5	to	to	ADP
easat-3242	175	6	1000	1000	NUM
easat-3242	175	7	classes	class	NOUN
easat-3242	175	8	(	(	PUNCT
easat-3242	175	9	i.e.	i.e.	X
easat-3242	175	10	,	,	PUNCT
easat-3242	175	11	imagenet	imagenet	ADJ
easat-3242	175	12	training	training	NOUN
easat-3242	175	13	)	)	PUNCT
easat-3242	175	14	,	,	PUNCT
easat-3242	175	15	because	because	SCONJ
easat-3242	175	16	vgg19	vgg19	PROPN
easat-3242	175	17	was	be	AUX
easat-3242	175	18	trained	train	VERB
easat-3242	175	19	on	on	ADP
easat-3242	175	20	the	the	DET
easat-3242	175	21	imagenet	imagenet	NOUN
easat-3242	175	22	dataset	dataset	NOUN
easat-3242	175	23	.	.	PUNCT
easat-3242	176	1	the	the	DET
easat-3242	176	2	output	output	NOUN
easat-3242	176	3	layer	layer	NOUN
easat-3242	176	4	consists	consist	VERB
easat-3242	176	5	of	of	ADP
easat-3242	176	6	a	a	DET
easat-3242	176	7	neuron	neuron	NOUN
easat-3242	176	8	per	per	ADP
easat-3242	176	9	class	class	NOUN
easat-3242	176	10	,	,	PUNCT
easat-3242	176	11	and	and	CCONJ
easat-3242	176	12	the	the	DET
easat-3242	176	13	network	network	NOUN
easat-3242	176	14	predicts	predict	VERB
easat-3242	176	15	into	into	ADP
easat-3242	176	16	which	which	DET
easat-3242	176	17	class	class	NOUN
easat-3242	176	18	an	an	DET
easat-3242	176	19	input	input	NOUN
easat-3242	176	20	image	image	NOUN
easat-3242	176	21	belongs	belong	VERB
easat-3242	176	22	based	base	VERB
easat-3242	176	23	on	on	ADP
easat-3242	176	24	the	the	DET
easat-3242	176	25	highest	high	ADJ
easat-3242	176	26	activation	activation	NOUN
easat-3242	176	27	.	.	PUNCT
easat-3242	177	1	the	the	DET
easat-3242	177	2	output	output	NOUN
easat-3242	177	3	layer	layer	NOUN
easat-3242	177	4	uses	use	VERB
easat-3242	177	5	a	a	DET
easat-3242	177	6	softmax	softmax	ADJ
easat-3242	177	7	activation	activation	NOUN
easat-3242	177	8	function	function	NOUN
easat-3242	177	9	that	that	PRON
easat-3242	177	10	obtains	obtain	VERB
easat-3242	177	11	the	the	DET
easat-3242	177	12	probability	probability	NOUN
easat-3242	177	13	distribution	distribution	NOUN
easat-3242	177	14	over	over	ADP
easat-3242	177	15	all	all	DET
easat-3242	177	16	classes	class	NOUN
easat-3242	177	17	from	from	ADP
easat-3242	177	18	the	the	DET
easat-3242	177	19	raw	raw	ADJ
easat-3242	177	20	output	output	NOUN
easat-3242	177	21	.	.	PUNCT
easat-3242	178	1	in	in	ADP
easat-3242	178	2	other	other	ADJ
easat-3242	178	3	words	word	NOUN
easat-3242	178	4	,	,	PUNCT
easat-3242	178	5	it	it	PRON
easat-3242	178	6	helps	help	VERB
easat-3242	178	7	the	the	DET
easat-3242	178	8	model	model	NOUN
easat-3242	178	9	understand	understand	VERB
easat-3242	178	10	the	the	DET
easat-3242	178	11	possible	possible	ADJ
easat-3242	178	12	likelihood	likelihood	NOUN
easat-3242	178	13	of	of	ADP
easat-3242	178	14	each	each	DET
easat-3242	178	15	class	class	NOUN
easat-3242	178	16	and	and	CCONJ
easat-3242	178	17	selected	select	VERB
easat-3242	178	18	the	the	DET
easat-3242	178	19	most	most	ADV
easat-3242	178	20	probable	probable	ADJ
easat-3242	178	21	one	one	NUM
easat-3242	178	22	.	.	PUNCT
easat-3242	179	1	in	in	ADP
easat-3242	179	2	the	the	DET
easat-3242	179	3	case	case	NOUN
easat-3242	179	4	of	of	ADP
easat-3242	179	5	predictions	prediction	NOUN
easat-3242	179	6	,	,	PUNCT
easat-3242	179	7	the	the	DET
easat-3242	179	8	input	input	NOUN
easat-3242	179	9	image	image	NOUN
easat-3242	179	10	,	,	PUNCT
easat-3242	179	11	after	after	ADP
easat-3242	179	12	passing	pass	VERB
easat-3242	179	13	through	through	ADP
easat-3242	179	14	input	input	NOUN
easat-3242	179	15	,	,	PUNCT
easat-3242	179	16	convolutional	convolutional	ADJ
easat-3242	179	17	,	,	PUNCT
easat-3242	179	18	maxpooling	maxpooling	NOUN
easat-3242	179	19	,	,	PUNCT
easat-3242	179	20	and	and	CCONJ
easat-3242	179	21	fully	fully	ADV
easat-3242	179	22	connected	connect	VERB
easat-3242	179	23	,	,	PUNCT
easat-3242	179	24	goes	go	VERB
easat-3242	179	25	to	to	ADP
easat-3242	179	26	the	the	DET
easat-3242	179	27	output	output	NOUN
easat-3242	179	28	layer	layer	NOUN
easat-3242	179	29	,	,	PUNCT
easat-3242	179	30	where	where	SCONJ
easat-3242	179	31	the	the	DET
easat-3242	179	32	final	final	ADJ
easat-3242	179	33	prediction	prediction	NOUN
easat-3242	179	34	is	be	AUX
easat-3242	179	35	generated	generate	VERB
easat-3242	179	36	.	.	PUNCT
easat-3242	180	1	this	this	PRON
easat-3242	180	2	is	be	AUX
easat-3242	180	3	called	call	VERB
easat-3242	180	4	the	the	DET
easat-3242	180	5	softmax	softmax	NOUN
easat-3242	180	6	function	function	NOUN
easat-3242	180	7	and	and	CCONJ
easat-3242	180	8	it	it	PRON
easat-3242	180	9	implies	imply	VERB
easat-3242	180	10	that	that	DET
easat-3242	180	11	sum	sum	NOUN
easat-3242	180	12	of	of	ADP
easat-3242	180	13	all	all	DET
easat-3242	180	14	probabilities	probability	NOUN
easat-3242	180	15	equals	equal	VERB
easat-3242	180	16	to1	to1	PROPN
easat-3242	180	17	,	,	PUNCT
easat-3242	180	18	which	which	PRON
easat-3242	180	19	relaxes	relax	VERB
easat-3242	180	20	calculation	calculation	NOUN
easat-3242	180	21	a	a	DET
easat-3242	180	22	bit	bit	NOUN
easat-3242	180	23	easier	easy	ADJ
easat-3242	180	24	in	in	ADP
easat-3242	180	25	terms	term	NOUN
easat-3242	180	26	interpreting	interpret	VERB
easat-3242	180	27	model	model	NOUN
easat-3242	180	28	prediction	prediction	NOUN
easat-3242	180	29	.	.	PUNCT
easat-3242	181	1	the	the	DET
easat-3242	181	2	class	class	NOUN
easat-3242	181	3	having	have	VERB
easat-3242	181	4	highest	high	ADJ
easat-3242	181	5	probability	probability	NOUN
easat-3242	181	6	is	be	AUX
easat-3242	181	7	then	then	ADV
easat-3242	181	8	selected	select	VERB
easat-3242	181	9	which	which	PRON
easat-3242	181	10	becomes	become	VERB
easat-3242	181	11	the	the	DET
easat-3242	181	12	predicted	predict	VERB
easat-3242	181	13	label	label	NOUN
easat-3242	181	14	of	of	ADP
easat-3242	181	15	input	input	NOUN
easat-3242	181	16	image	image	NOUN
easat-3242	181	17	.	.	PUNCT
easat-3242	182	1	the	the	DET
easat-3242	182	2	correct	correct	ADJ
easat-3242	182	3	inference	inference	NOUN
easat-3242	182	4	of	of	ADP
easat-3242	182	5	these	these	DET
easat-3242	182	6	meaningful	meaningful	ADJ
easat-3242	182	7	predictions	prediction	NOUN
easat-3242	182	8	and	and	CCONJ
easat-3242	182	9	the	the	DET
easat-3242	182	10	high	high	ADJ
easat-3242	182	11	-	-	PUNCT
easat-3242	182	12	level	level	NOUN
easat-3242	182	13	features	feature	NOUN
easat-3242	182	14	that	that	PRON
easat-3242	182	15	are	be	AUX
easat-3242	182	16	learned	learn	VERB
easat-3242	182	17	in	in	ADP
easat-3242	182	18	a	a	DET
easat-3242	182	19	hierarchical	hierarchical	ADJ
easat-3242	182	20	manner	manner	NOUN
easat-3242	182	21	,	,	PUNCT
easat-3242	182	22	at	at	ADP
easat-3242	182	23	each	each	DET
easat-3242	182	24	layer	layer	NOUN
easat-3242	182	25	,	,	PUNCT
easat-3242	182	26	provide	provide	VERB
easat-3242	182	27	a	a	DET
easat-3242	182	28	unique	unique	ADJ
easat-3242	182	29	capability	capability	NOUN
easat-3242	182	30	—	—	PUNCT
easat-3242	182	31	class	class	NOUN
easat-3242	182	32	-	-	PUNCT
easat-3242	182	33	specific	specific	ADJ
easat-3242	182	34	information	information	NOUN
easat-3242	182	35	is	be	AUX
easat-3242	182	36	distilled	distil	VERB
easat-3242	182	37	into	into	ADP
easat-3242	182	38	taking	take	VERB
easat-3242	182	39	decisions	decision	NOUN
easat-3242	182	40	from	from	ADP
easat-3242	182	41	their	their	PRON
easat-3242	182	42	local	local	ADJ
easat-3242	182	43	receptive	receptive	ADJ
easat-3242	182	44	fields	field	NOUN
easat-3242	182	45	to	to	ADP
easat-3242	182	46	global	global	ADJ
easat-3242	182	47	correlations	correlation	NOUN
easat-3242	182	48	.	.	PUNCT
easat-3242	183	1	3.3.7	3.3.7	X
easat-3242	183	2	.	.	PUNCT
easat-3242	184	1	functionality	functionality	VERB
easat-3242	184	2	the	the	DET
easat-3242	184	3	main	main	ADJ
easat-3242	184	4	application	application	NOUN
easat-3242	184	5	to	to	PART
easat-3242	184	6	use	use	VERB
easat-3242	184	7	vgg19	vgg19	PROPN
easat-3242	184	8	is	be	AUX
easat-3242	184	9	in	in	ADP
easat-3242	184	10	feature	feature	NOUN
easat-3242	184	11	extraction	extraction	NOUN
easat-3242	184	12	from	from	ADP
easat-3242	184	13	input	input	NOUN
easat-3242	184	14	images	image	NOUN
easat-3242	184	15	.	.	PUNCT
easat-3242	185	1	its	its	PRON
easat-3242	185	2	deep	deep	ADJ
easat-3242	185	3	network	network	NOUN
easat-3242	185	4	architecture	architecture	NOUN
easat-3242	185	5	enable	enable	VERB
easat-3242	185	6	hierarchical	hierarchical	ADJ
easat-3242	185	7	representations	representation	NOUN
easat-3242	185	8	of	of	ADP
easat-3242	185	9	visual	visual	ADJ
easat-3242	185	10	data	datum	NOUN
easat-3242	185	11	,	,	PUNCT
easat-3242	185	12	moving	move	VERB
easat-3242	185	13	from	from	ADP
easat-3242	185	14	differential	differential	ADJ
easat-3242	185	15	gradients	gradient	NOUN
easat-3242	185	16	and	and	CCONJ
easat-3242	185	17	textures	texture	NOUN
easat-3242	185	18	in	in	ADP
easat-3242	185	19	the	the	DET
easat-3242	185	20	early	early	ADJ
easat-3242	185	21	layers	layer	NOUN
easat-3242	185	22	(	(	PUNCT
easat-3242	185	23	closer	close	ADJ
easat-3242	185	24	to	to	PART
easat-3242	185	25	input	input	VERB
easat-3242	185	26	)	)	PUNCT
easat-3242	185	27	towards	towards	ADP
easat-3242	185	28	more	more	ADJ
easat-3242	185	29	abstract	abstract	ADJ
easat-3242	185	30	shapes	shape	NOUN
easat-3242	185	31	or	or	CCONJ
easat-3242	185	32	objects	object	NOUN
easat-3242	185	33	as	as	SCONJ
easat-3242	185	34	they	they	PRON
easat-3242	185	35	proceed	proceed	VERB
easat-3242	185	36	deeper	deep	ADJ
easat-3242	185	37	.	.	PUNCT
easat-3242	186	1	this	this	DET
easat-3242	186	2	property	property	NOUN
easat-3242	186	3	makes	make	VERB
easat-3242	186	4	it	it	PRON
easat-3242	186	5	a	a	DET
easat-3242	186	6	versatile	versatile	ADJ
easat-3242	186	7	classifier	classifier	NOUN
easat-3242	186	8	for	for	ADP
easat-3242	186	9	image	image	NOUN
easat-3242	186	10	processing	processing	NOUN
easat-3242	186	11	problems	problem	NOUN
easat-3242	186	12	.	.	PUNCT
easat-3242	187	1	one	one	NUM
easat-3242	187	2	of	of	ADP
easat-3242	187	3	its	its	PRON
easat-3242	187	4	popular	popular	ADJ
easat-3242	187	5	use	use	NOUN
easat-3242	187	6	cases	case	NOUN
easat-3242	187	7	is	be	AUX
easat-3242	187	8	a	a	DET
easat-3242	187	9	pre	pre	ADJ
easat-3242	187	10	-	-	ADJ
easat-3242	187	11	trained	train	VERB
easat-3242	187	12	model	model	NOUN
easat-3242	187	13	for	for	ADP
easat-3242	187	14	transfer	transfer	NOUN
easat-3242	187	15	learning	learning	NOUN
easat-3242	187	16	.	.	PUNCT
easat-3242	188	1	the	the	DET
easat-3242	188	2	model	model	NOUN
easat-3242	188	3	's	's	PART
easat-3242	188	4	weight	weight	NOUN
easat-3242	188	5	values	value	NOUN
easat-3242	188	6	are	be	AUX
easat-3242	188	7	initialized	initialize	VERB
easat-3242	188	8	using	use	VERB
easat-3242	188	9	a	a	DET
easat-3242	188	10	pre	pre	ADJ
easat-3242	188	11	-	-	ADJ
easat-3242	188	12	trained	train	VERB
easat-3242	188	13	model	model	NOUN
easat-3242	188	14	(	(	PUNCT
easat-3242	188	15	e.g.	e.g.	ADJ
easat-3242	188	16	,	,	PUNCT
easat-3242	188	17	imagenet	imagenet	NOUN
easat-3242	188	18	)	)	PUNCT
easat-3242	188	19	and	and	CCONJ
easat-3242	188	20	the	the	DET
easat-3242	188	21	network	network	NOUN
easat-3242	188	22	is	be	AUX
easat-3242	188	23	fine	fine	ADV
easat-3242	188	24	-	-	PUNCT
easat-3242	188	25	tuned	tune	VERB
easat-3242	188	26	for	for	ADP
easat-3242	188	27	small	small	ADJ
easat-3242	188	28	datasets	dataset	NOUN
easat-3242	188	29	,	,	PUNCT
easat-3242	188	30	allowing	allow	VERB
easat-3242	188	31	easy	easy	ADJ
easat-3242	188	32	learning	learning	NOUN
easat-3242	188	33	with	with	ADP
easat-3242	188	34	less	less	ADJ
easat-3242	188	35	data	datum	NOUN
easat-3242	188	36	.	.	PUNCT
easat-3242	189	1	they	they	PRON
easat-3242	189	2	benefit	benefit	VERB
easat-3242	189	3	from	from	ADP
easat-3242	189	4	the	the	DET
easat-3242	189	5	good	good	ADJ
easat-3242	189	6	generalisation	generalisation	NOUN
easat-3242	189	7	of	of	ADP
easat-3242	189	8	the	the	DET
easat-3242	189	9	model	model	NOUN
easat-3242	189	10	,	,	PUNCT
easat-3242	189	11	so	so	SCONJ
easat-3242	189	12	it	it	PRON
easat-3242	189	13	works	work	VERB
easat-3242	189	14	well	well	ADV
easat-3242	189	15	in	in	ADP
easat-3242	189	16	contexts	context	NOUN
easat-3242	189	17	with	with	ADP
easat-3242	189	18	limited	limited	ADJ
easat-3242	189	19	training	training	NOUN
easat-3242	189	20	data	datum	NOUN
easat-3242	189	21	.	.	PUNCT
easat-3242	190	1	one	one	NUM
easat-3242	190	2	of	of	ADP
easat-3242	190	3	the	the	DET
easat-3242	190	4	other	other	ADJ
easat-3242	190	5	main	main	ADJ
easat-3242	190	6	work	work	NOUN
easat-3242	190	7	for	for	ADP
easat-3242	190	8	which	which	PRON
easat-3242	190	9	vgg19	vgg19	PRON
easat-3242	190	10	is	be	AUX
easat-3242	190	11	known	know	VERB
easat-3242	190	12	named	name	VERB
easat-3242	190	13	as	as	ADP
easat-3242	190	14	image	image	NOUN
easat-3242	190	15	classification	classification	NOUN
easat-3242	190	16	.	.	PUNCT
easat-3242	191	1	in	in	ADP
easat-3242	191	2	this	this	DET
easat-3242	191	3	process	process	NOUN
easat-3242	191	4	,	,	PUNCT
easat-3242	191	5	for	for	ADP
easat-3242	191	6	the	the	DET
easat-3242	191	7	input	input	NOUN
easat-3242	191	8	images	image	NOUN
easat-3242	191	9	it	it	PRON
easat-3242	191	10	is	be	AUX
easat-3242	191	11	trained	train	VERB
easat-3242	191	12	and	and	CCONJ
easat-3242	191	13	fot	fot	VERB
easat-3242	191	14	these	these	PRON
easat-3242	191	15	it	it	PRON
easat-3242	191	16	will	will	AUX
easat-3242	191	17	predict	predict	VERB
easat-3242	191	18	to	to	PART
easat-3242	191	19	which	which	DET
easat-3242	191	20	class	class	NOUN
easat-3242	191	21	label	label	NOUN
easat-3242	191	22	they	they	PRON
easat-3242	191	23	belongs	belong	VERB
easat-3242	191	24	using	use	VERB
easat-3242	191	25	features	feature	NOUN
easat-3242	191	26	learned	learn	VERB
easat-3242	191	27	from	from	ADP
easat-3242	191	28	training	training	NOUN
easat-3242	191	29	.	.	PUNCT
easat-3242	192	1	the	the	DET
easat-3242	192	2	last	last	ADJ
easat-3242	192	3	layer	layer	NOUN
easat-3242	192	4	of	of	ADP
easat-3242	192	5	the	the	DET
easat-3242	192	6	network	network	NOUN
easat-3242	192	7	is	be	AUX
easat-3242	192	8	a	a	DET
easat-3242	192	9	softmax	softmax	NOUN
easat-3242	192	10	and	and	CCONJ
easat-3242	192	11	hence	hence	ADV
easat-3242	192	12	outputs	output	VERB
easat-3242	192	13	probabilities	probability	NOUN
easat-3242	192	14	to	to	PART
easat-3242	192	15	put	put	VERB
easat-3242	192	16	one	one	NUM
easat-3242	192	17	particular	particular	ADJ
easat-3242	192	18	class	class	NOUN
easat-3242	192	19	as	as	ADP
easat-3242	192	20	label	label	NOUN
easat-3242	192	21	having	have	VERB
easat-3242	192	22	highest	high	ADJ
easat-3242	192	23	probability	probability	NOUN
easat-3242	192	24	value	value	NOUN
easat-3242	192	25	.	.	PUNCT
easat-3242	193	1	despite	despite	SCONJ
easat-3242	193	2	the	the	DET
easat-3242	193	3	fact	fact	NOUN
easat-3242	193	4	that	that	SCONJ
easat-3242	193	5	vgg19	vgg19	PROPN
easat-3242	193	6	is	be	AUX
easat-3242	193	7	not	not	PART
easat-3242	193	8	built	build	VERB
easat-3242	193	9	for	for	ADP
easat-3242	193	10	object	object	NOUN
easat-3242	193	11	detection	detection	NOUN
easat-3242	193	12	,	,	PUNCT
easat-3242	193	13	its	its	PRON
easat-3242	193	14	property	property	NOUN
easat-3242	193	15	to	to	PART
easat-3242	193	16	learn	learn	VERB
easat-3242	193	17	general	general	ADJ
easat-3242	193	18	features	feature	NOUN
easat-3242	193	19	renders	render	VERB
easat-3242	193	20	it	it	PRON
easat-3242	193	21	compatible	compatible	ADJ
easat-3242	193	22	with	with	ADP
easat-3242	193	23	object	object	NOUN
easat-3242	193	24	-	-	PUNCT
easat-3242	193	25	detection	detection	NOUN
easat-3242	193	26	architectures	architecture	NOUN
easat-3242	193	27	.	.	PUNCT
easat-3242	194	1	for	for	ADP
easat-3242	194	2	example	example	NOUN
easat-3242	194	3	,	,	PUNCT
easat-3242	194	4	the	the	DET
easat-3242	194	5	feature	feature	NOUN
easat-3242	194	6	maps	map	NOUN
easat-3242	194	7	produced	produce	VERB
easat-3242	194	8	by	by	ADP
easat-3242	194	9	vgg19	vgg19	PROPN
easat-3242	194	10	may	may	AUX
easat-3242	194	11	serve	serve	VERB
easat-3242	194	12	as	as	ADP
easat-3242	194	13	input	input	NOUN
easat-3242	194	14	to	to	ADP
easat-3242	194	15	5714	5714	NUM
easat-3242	194	16	edelweiss	edelweiss	PROPN
easat-3242	194	17	applied	apply	VERB
easat-3242	194	18	science	science	NOUN
easat-3242	194	19	and	and	CCONJ
easat-3242	194	20	technology	technology	NOUN
easat-3242	194	21	issn	issn	PROPN
easat-3242	194	22	:	:	PUNCT
easat-3242	194	23	2576	2576	NUM
easat-3242	194	24	-	-	SYM
easat-3242	194	25	8484	8484	NUM
easat-3242	194	26	vol	vol	NOUN
easat-3242	194	27	.	.	PROPN
easat-3242	195	1	8	8	NUM
easat-3242	195	2	,	,	PUNCT
easat-3242	195	3	no	no	INTJ
easat-3242	195	4	.	.	NOUN
easat-3242	196	1	6	6	NUM
easat-3242	196	2	:	:	SYM
easat-3242	196	3	5707	5707	NUM
easat-3242	196	4	-	-	SYM
easat-3242	196	5	5722	5722	NUM
easat-3242	196	6	,	,	PUNCT
easat-3242	196	7	2024	2024	NUM
easat-3242	196	8	doi	doi	NOUN
easat-3242	196	9	:	:	PUNCT
easat-3242	196	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	196	11	©	©	PROPN
easat-3242	196	12	2024	2024	NUM
easat-3242	196	13	by	by	ADP
easat-3242	196	14	the	the	DET
easat-3242	196	15	authors	author	NOUN
easat-3242	196	16	;	;	PUNCT
easat-3242	196	17	licensee	licensee	PROPN
easat-3242	196	18	learning	learn	VERB
easat-3242	196	19	gate	gate	NOUN
easat-3242	196	20	object	object	NOUN
easat-3242	196	21	detection	detection	NOUN
easat-3242	196	22	models	model	NOUN
easat-3242	196	23	like	like	ADP
easat-3242	196	24	region	region	NOUN
easat-3242	196	25	convolutional	convolutional	ADJ
easat-3242	196	26	neural	neural	ADJ
easat-3242	196	27	networks	network	NOUN
easat-3242	196	28	(	(	PUNCT
easat-3242	196	29	r	r	NOUN
easat-3242	196	30	-	-	PUNCT
easat-3242	196	31	cnns	cnns	ADJ
easat-3242	196	32	)	)	PUNCT
easat-3242	196	33	or	or	CCONJ
easat-3242	196	34	single	single	ADJ
easat-3242	196	35	shot	shot	NOUN
easat-3242	196	36	detectors	detector	NOUN
easat-3242	196	37	(	(	PUNCT
easat-3242	196	38	ssds	ssds	NOUN
easat-3242	196	39	)	)	PUNCT
easat-3242	196	40	,	,	PUNCT
easat-3242	196	41	thereby	thereby	ADV
easat-3242	196	42	enabling	enable	VERB
easat-3242	196	43	easy	easy	ADJ
easat-3242	196	44	identification	identification	NOUN
easat-3242	196	45	of	of	ADP
easat-3242	196	46	objects	object	NOUN
easat-3242	196	47	.	.	PUNCT
easat-3242	197	1	the	the	DET
easat-3242	197	2	vgg19	vgg19	PROPN
easat-3242	197	3	model	model	NOUN
easat-3242	197	4	is	be	AUX
easat-3242	197	5	extremely	extremely	ADV
easat-3242	197	6	useful	useful	ADJ
easat-3242	197	7	,	,	PUNCT
easat-3242	197	8	not	not	PART
easat-3242	197	9	only	only	ADV
easat-3242	197	10	in	in	ADP
easat-3242	197	11	classification	classification	NOUN
easat-3242	197	12	and	and	CCONJ
easat-3242	197	13	object	object	NOUN
easat-3242	197	14	detection	detection	NOUN
easat-3242	197	15	but	but	CCONJ
easat-3242	197	16	also	also	ADV
easat-3242	197	17	all	all	DET
easat-3242	197	18	types	type	NOUN
easat-3242	197	19	of	of	ADP
easat-3242	197	20	semantic	semantic	ADJ
easat-3242	197	21	segmentation	segmentation	NOUN
easat-3242	197	22	tasks	task	NOUN
easat-3242	197	23	.	.	PUNCT
easat-3242	198	1	satellite	satellite	NOUN
easat-3242	198	2	image	image	NOUN
easat-3242	198	3	recognition	recognition	NOUN
easat-3242	198	4	on	on	ADP
easat-3242	198	5	the	the	DET
easat-3242	198	6	other	other	ADJ
easat-3242	198	7	hand	hand	NOUN
easat-3242	198	8	represents	represent	VERB
easat-3242	198	9	an	an	DET
easat-3242	198	10	even	even	ADV
easat-3242	198	11	more	more	ADV
easat-3242	198	12	difficult	difficult	ADJ
easat-3242	198	13	task	task	NOUN
easat-3242	198	14	for	for	ADP
easat-3242	198	15	which	which	PRON
easat-3242	198	16	standard	standard	ADJ
easat-3242	198	17	image	image	NOUN
easat-3242	198	18	classification	classification	NOUN
easat-3242	198	19	methods	method	NOUN
easat-3242	198	20	are	be	AUX
easat-3242	198	21	not	not	PART
easat-3242	198	22	sufficient	sufficient	ADJ
easat-3242	198	23	given	give	VERB
easat-3242	198	24	that	that	PRON
easat-3242	198	25	,	,	PUNCT
easat-3242	198	26	in	in	ADP
easat-3242	198	27	addition	addition	NOUN
easat-3242	198	28	to	to	ADP
easat-3242	198	29	classifying	classify	VERB
easat-3242	198	30	and	and	CCONJ
easat-3242	198	31	recognizing	recognize	VERB
easat-3242	198	32	images	image	NOUN
easat-3242	198	33	you	you	PRON
easat-3242	198	34	have	have	AUX
easat-3242	198	35	do	do	VERB
easat-3242	198	36	it	it	PRON
easat-3242	198	37	“	"	PUNCT
easat-3242	198	38	filling	fill	VERB
easat-3242	198	39	”	"	PUNCT
easat-3242	198	40	with	with	ADP
easat-3242	198	41	recognition	recognition	NOUN
easat-3242	198	42	each	each	DET
easat-3242	198	43	pixel	pixel	NOUN
easat-3242	198	44	of	of	ADP
easat-3242	198	45	the	the	DET
easat-3242	198	46	input	input	NOUN
easat-3242	198	47	satellite	satellite	NOUN
easat-3242	198	48	photo	photo	NOUN
easat-3242	198	49	.	.	PUNCT
easat-3242	199	1	the	the	DET
easat-3242	199	2	feature	feature	NOUN
easat-3242	199	3	maps	map	NOUN
easat-3242	199	4	from	from	ADP
easat-3242	199	5	vgg19	vgg19	PROPN
easat-3242	199	6	can	can	AUX
easat-3242	199	7	be	be	AUX
easat-3242	199	8	used	use	VERB
easat-3242	199	9	with	with	ADP
easat-3242	199	10	decoder	decoder	NOUN
easat-3242	199	11	networks	network	NOUN
easat-3242	199	12	to	to	PART
easat-3242	199	13	get	get	VERB
easat-3242	199	14	segmentation	segmentation	NOUN
easat-3242	199	15	masks	mask	NOUN
easat-3242	199	16	that	that	PRON
easat-3242	199	17	give	give	VERB
easat-3242	199	18	where	where	SCONJ
easat-3242	199	19	different	different	ADJ
easat-3242	199	20	objects	object	NOUN
easat-3242	199	21	in	in	ADP
easat-3242	199	22	an	an	DET
easat-3242	199	23	image	image	NOUN
easat-3242	199	24	.	.	PUNCT
easat-3242	200	1	the	the	DET
easat-3242	200	2	network	network	NOUN
easat-3242	200	3	produces	produce	VERB
easat-3242	200	4	feature	feature	NOUN
easat-3242	200	5	maps	map	NOUN
easat-3242	200	6	which	which	PRON
easat-3242	200	7	qualify	qualify	VERB
easat-3242	200	8	as	as	ADP
easat-3242	200	9	the	the	DET
easat-3242	200	10	input	input	NOUN
easat-3242	200	11	for	for	ADP
easat-3242	200	12	constructing	construct	VERB
easat-3242	200	13	these	these	DET
easat-3242	200	14	masks	mask	NOUN
easat-3242	200	15	,	,	PUNCT
easat-3242	200	16	indicating	indicate	VERB
easat-3242	200	17	vgg19	vgg19	PROPN
easat-3242	200	18	's	's	PART
easat-3242	200	19	versatility	versatility	NOUN
easat-3242	200	20	in	in	ADP
easat-3242	200	21	processing	process	VERB
easat-3242	200	22	pixel	pixel	NOUN
easat-3242	200	23	level	level	NOUN
easat-3242	200	24	tasks	task	NOUN
easat-3242	200	25	.	.	PUNCT
easat-3242	201	1	deep	deep	ADJ
easat-3242	201	2	feature	feature	NOUN
easat-3242	201	3	extraction	extraction	NOUN
easat-3242	201	4	from	from	ADP
easat-3242	201	5	vgg-16	vgg-16	NOUN
easat-3242	201	6	,	,	PUNCT
easat-3242	201	7	one	one	NUM
easat-3242	201	8	of	of	ADP
easat-3242	201	9	the	the	DET
easat-3242	201	10	most	most	ADV
easat-3242	201	11	famous	famous	ADJ
easat-3242	201	12	models	model	NOUN
easat-3242	201	13	used	use	VERB
easat-3242	201	14	in	in	ADP
easat-3242	201	15	computer	computer	NOUN
easat-3242	201	16	vision	vision	NOUN
easat-3242	201	17	projects	project	NOUN
easat-3242	201	18	for	for	ADP
easat-3242	201	19	semantic	semantic	ADJ
easat-3242	201	20	segmentation	segmentation	NOUN
easat-3242	201	21	.	.	PUNCT
easat-3242	202	1	in	in	ADP
easat-3242	202	2	the	the	DET
easat-3242	202	3	end	end	NOUN
easat-3242	202	4	,	,	PUNCT
easat-3242	202	5	by	by	ADP
easat-3242	202	6	accommodating	accommodate	VERB
easat-3242	202	7	all	all	PRON
easat-3242	202	8	of	of	ADP
easat-3242	202	9	these	these	DET
easat-3242	202	10	basic	basic	ADJ
easat-3242	202	11	details	detail	NOUN
easat-3242	202	12	,	,	PUNCT
easat-3242	202	13	vgg19	vgg19	PROPN
easat-3242	202	14	is	be	AUX
easat-3242	202	15	one	one	NUM
easat-3242	202	16	among	among	ADP
easat-3242	202	17	very	very	ADV
easat-3242	202	18	strong	strong	ADJ
easat-3242	202	19	and	and	CCONJ
easat-3242	202	20	powerful	powerful	ADJ
easat-3242	202	21	convolutional	convolutional	ADJ
easat-3242	202	22	neural	neural	ADJ
easat-3242	202	23	network	network	NOUN
easat-3242	202	24	(	(	PUNCT
easat-3242	202	25	cnn	cnn	PROPN
easat-3242	202	26	)	)	PUNCT
easat-3242	202	27	structure	structure	NOUN
easat-3242	202	28	that	that	PRON
easat-3242	202	29	may	may	AUX
easat-3242	202	30	be	be	AUX
easat-3242	202	31	widely	widely	ADV
easat-3242	202	32	practiced	practice	VERB
easat-3242	202	33	for	for	ADP
easat-3242	202	34	photo	photo	NOUN
easat-3242	202	35	classification	classification	NOUN
easat-3242	202	36	to	to	PART
easat-3242	202	37	object	object	VERB
easat-3242	202	38	detection	detection	NOUN
easat-3242	202	39	/	/	SYM
easat-3242	202	40	recognition	recognition	NOUN
easat-3242	202	41	along	along	ADP
easat-3242	202	42	with	with	ADP
easat-3242	202	43	semantic	semantic	ADJ
easat-3242	202	44	segmentation	segmentation	NOUN
easat-3242	202	45	.	.	PUNCT
easat-3242	203	1	with	with	ADP
easat-3242	203	2	this	this	DET
easat-3242	203	3	ability	ability	NOUN
easat-3242	203	4	to	to	PART
easat-3242	203	5	generate	generate	VERB
easat-3242	203	6	hierarchical	hierarchical	ADJ
easat-3242	203	7	feature	feature	NOUN
easat-3242	203	8	representations	representation	NOUN
easat-3242	203	9	,	,	PUNCT
easat-3242	203	10	it	it	PRON
easat-3242	203	11	is	be	AUX
easat-3242	203	12	useful	useful	ADJ
easat-3242	203	13	for	for	ADP
easat-3242	203	14	object	object	NOUN
easat-3242	203	15	detection	detection	NOUN
easat-3242	203	16	or	or	CCONJ
easat-3242	203	17	similar	similar	ADJ
easat-3242	203	18	task	task	NOUN
easat-3242	203	19	of	of	ADP
easat-3242	203	20	computer	computer	NOUN
easat-3242	203	21	vision	vision	NOUN
easat-3242	203	22	;	;	PUNCT
easat-3242	203	23	because	because	SCONJ
easat-3242	203	24	an	an	DET
easat-3242	203	25	effective	effective	ADJ
easat-3242	203	26	detector	detector	NOUN
easat-3242	203	27	should	should	AUX
easat-3242	203	28	output	output	VERB
easat-3242	203	29	both	both	DET
easat-3242	203	30	the	the	DET
easat-3242	203	31	regions	region	NOUN
easat-3242	203	32	that	that	PRON
easat-3242	203	33	contain	contain	VERB
easat-3242	203	34	objects	object	NOUN
easat-3242	203	35	and	and	CCONJ
easat-3242	203	36	there	there	PRON
easat-3242	203	37	categories	category	NOUN
easat-3242	203	38	in	in	ADP
easat-3242	203	39	a	a	DET
easat-3242	203	40	region	region	NOUN
easat-3242	203	41	.	.	PUNCT
easat-3242	204	1	vgg19	vgg19	PROPN
easat-3242	204	2	is	be	AUX
easat-3242	204	3	among	among	ADP
easat-3242	204	4	the	the	DET
easat-3242	204	5	most	most	ADV
easat-3242	204	6	widely	widely	ADV
easat-3242	204	7	used	use	VERB
easat-3242	204	8	architectures	architecture	NOUN
easat-3242	204	9	in	in	ADP
easat-3242	204	10	deep	deep	ADJ
easat-3242	204	11	learning	learning	NOUN
easat-3242	204	12	,	,	PUNCT
easat-3242	204	13	popular	popular	ADJ
easat-3242	204	14	amongst	amongst	ADP
easat-3242	204	15	researchers	researcher	NOUN
easat-3242	204	16	and	and	CCONJ
easat-3242	204	17	practitioners	practitioner	NOUN
easat-3242	204	18	because	because	SCONJ
easat-3242	204	19	it	it	PRON
easat-3242	204	20	has	have	AUX
easat-3242	204	21	proven	prove	VERB
easat-3242	204	22	to	to	PART
easat-3242	204	23	be	be	AUX
easat-3242	204	24	very	very	ADV
easat-3242	204	25	effective	effective	ADJ
easat-3242	204	26	at	at	ADP
easat-3242	204	27	solving	solve	VERB
easat-3242	204	28	a	a	DET
easat-3242	204	29	lot	lot	NOUN
easat-3242	204	30	of	of	ADP
easat-3242	204	31	different	different	ADJ
easat-3242	204	32	image	image	NOUN
easat-3242	204	33	challenges	challenge	NOUN
easat-3242	204	34	.	.	PUNCT
easat-3242	205	1	therefore	therefore	ADV
easat-3242	205	2	,	,	PUNCT
easat-3242	205	3	it	it	PRON
easat-3242	205	4	has	have	AUX
easat-3242	205	5	become	become	VERB
easat-3242	205	6	one	one	NUM
easat-3242	205	7	of	of	ADP
easat-3242	205	8	the	the	DET
easat-3242	205	9	vital	vital	ADJ
easat-3242	205	10	tools	tool	NOUN
easat-3242	205	11	for	for	ADP
easat-3242	205	12	various	various	ADJ
easat-3242	205	13	professionals	professional	NOUN
easat-3242	205	14	working	work	VERB
easat-3242	205	15	on	on	ADP
easat-3242	205	16	visual	visual	ADJ
easat-3242	205	17	data	datum	NOUN
easat-3242	205	18	processing	processing	NOUN
easat-3242	205	19	.	.	PUNCT
easat-3242	206	1	3.4	3.4	NUM
easat-3242	206	2	.	.	PUNCT
easat-3242	206	3	proposed	propose	VERB
easat-3242	206	4	unet++vgg19	unet++vgg19	PROPN
easat-3242	206	5	model	model	NOUN
easat-3242	206	6	the	the	DET
easat-3242	206	7	unet++vgg19	unet++vgg19	PROPN
easat-3242	206	8	model	model	NOUN
easat-3242	206	9	is	be	AUX
easat-3242	206	10	a	a	DET
easat-3242	206	11	hybrid	hybrid	ADJ
easat-3242	206	12	architecture	architecture	NOUN
easat-3242	206	13	that	that	PRON
easat-3242	206	14	fuses	fuse	VERB
easat-3242	206	15	the	the	DET
easat-3242	206	16	advantages	advantage	NOUN
easat-3242	206	17	of	of	ADP
easat-3242	206	18	both	both	DET
easat-3242	206	19	models	model	NOUN
easat-3242	206	20	,	,	PUNCT
easat-3242	206	21	aiming	aim	VERB
easat-3242	206	22	to	to	PART
easat-3242	206	23	improve	improve	VERB
easat-3242	206	24	satellite	satellite	NOUN
easat-3242	206	25	image	image	NOUN
easat-3242	206	26	segmentation	segmentation	NOUN
easat-3242	206	27	.	.	PUNCT
easat-3242	207	1	this	this	PRON
easat-3242	207	2	was	be	AUX
easat-3242	207	3	done	do	VERB
easat-3242	207	4	by	by	ADP
easat-3242	207	5	marrying	marry	VERB
easat-3242	207	6	the	the	DET
easat-3242	207	7	stacked	stack	VERB
easat-3242	207	8	skip	skip	ADJ
easat-3242	207	9	connection	connection	NOUN
easat-3242	207	10	of	of	ADP
easat-3242	207	11	unet++	unet++	NOUN
easat-3242	207	12	with	with	ADP
easat-3242	207	13	the	the	DET
easat-3242	207	14	very	very	ADV
easat-3242	207	15	powerful	powerful	ADJ
easat-3242	207	16	feature	feature	NOUN
easat-3242	207	17	extraction	extraction	NOUN
easat-3242	207	18	ability	ability	NOUN
easat-3242	207	19	of	of	ADP
easat-3242	207	20	vgg19	vgg19	PRON
easat-3242	207	21	to	to	PART
easat-3242	207	22	achieve	achieve	VERB
easat-3242	207	23	a	a	DET
easat-3242	207	24	desired	desire	VERB
easat-3242	207	25	balance	balance	NOUN
easat-3242	207	26	between	between	ADP
easat-3242	207	27	clear	clear	ADJ
easat-3242	207	28	segmentation	segmentation	NOUN
easat-3242	207	29	and	and	CCONJ
easat-3242	207	30	detailed	detailed	ADJ
easat-3242	207	31	features	feature	NOUN
easat-3242	207	32	.	.	PUNCT
easat-3242	208	1	an	an	DET
easat-3242	208	2	improved	improved	ADJ
easat-3242	208	3	version	version	NOUN
easat-3242	208	4	of	of	ADP
easat-3242	208	5	the	the	DET
easat-3242	208	6	unet	unet	NOUN
easat-3242	208	7	architecture	architecture	NOUN
easat-3242	208	8	for	for	ADP
easat-3242	208	9	biological	biological	ADJ
easat-3242	208	10	image	image	NOUN
easat-3242	208	11	segmentation	segmentation	NOUN
easat-3242	208	12	tasks	task	NOUN
easat-3242	208	13	;	;	PUNCT
easat-3242	208	14	referred	refer	VERB
easat-3242	208	15	to	to	ADP
easat-3242	208	16	as	as	ADP
easat-3242	208	17	an	an	DET
easat-3242	208	18	evolution	evolution	NOUN
easat-3242	208	19	on	on	ADP
easat-3242	208	20	unet	unet	NOUN
easat-3242	208	21	,	,	PUNCT
easat-3242	208	22	naming	name	VERB
easat-3242	208	23	it	it	PRON
easat-3242	208	24	unet++	unet++	ADJ
easat-3242	208	25	.	.	PUNCT
easat-3242	209	1	it	it	PRON
easat-3242	209	2	has	have	AUX
easat-3242	209	3	nested	nest	VERB
easat-3242	209	4	,	,	PUNCT
easat-3242	209	5	tightly	tightly	ADV
easat-3242	209	6	coupled	couple	VERB
easat-3242	209	7	skip	skip	ADJ
easat-3242	209	8	connections	connection	NOUN
easat-3242	209	9	to	to	PART
easat-3242	209	10	enable	enable	VERB
easat-3242	209	11	it	it	PRON
easat-3242	209	12	to	to	PART
easat-3242	209	13	address	address	VERB
easat-3242	209	14	the	the	DET
easat-3242	209	15	shortcomings	shortcoming	NOUN
easat-3242	209	16	of	of	ADP
easat-3242	209	17	original	original	ADJ
easat-3242	209	18	unet	unet	NOUN
easat-3242	209	19	model	model	NOUN
easat-3242	209	20	by	by	ADP
easat-3242	209	21	enhancing	enhance	VERB
easat-3242	209	22	feature	feature	NOUN
easat-3242	209	23	propagation	propagation	NOUN
easat-3242	209	24	and	and	CCONJ
easat-3242	209	25	alleviating	alleviate	VERB
easat-3242	209	26	semantic	semantic	ADJ
easat-3242	209	27	gap	gap	NOUN
easat-3242	209	28	between	between	ADP
easat-3242	209	29	encoder	encoder	NOUN
easat-3242	209	30	layers	layer	NOUN
easat-3242	209	31	and	and	CCONJ
easat-3242	209	32	decoder	decoder	NOUN
easat-3242	209	33	side	side	NOUN
easat-3242	209	34	.	.	PUNCT
easat-3242	210	1	vgg19	vgg19	PROPN
easat-3242	210	2	,	,	PUNCT
easat-3242	210	3	on	on	ADP
easat-3242	210	4	the	the	DET
easat-3242	210	5	other	other	ADJ
easat-3242	210	6	hand	hand	NOUN
easat-3242	210	7	is	be	AUX
easat-3242	210	8	state	state	NOUN
easat-3242	210	9	-	-	PUNCT
easat-3242	210	10	of	of	ADP
easat-3242	210	11	-	-	PUNCT
easat-3242	210	12	the	the	DET
easat-3242	210	13	-	-	PUNCT
easat-3242	210	14	art	art	NOUN
easat-3242	210	15	convolutional	convolutional	ADJ
easat-3242	210	16	neural	neural	ADJ
easat-3242	210	17	network	network	NOUN
easat-3242	210	18	(	(	PUNCT
easat-3242	210	19	cnn	cnn	PROPN
easat-3242	210	20	)	)	PUNCT
easat-3242	210	21	which	which	PRON
easat-3242	210	22	pretrained	pretraine	VERB
easat-3242	210	23	with	with	ADP
easat-3242	210	24	high	high	ADJ
easat-3242	210	25	-	-	PUNCT
easat-3242	210	26	resolution	resolution	NOUN
easat-3242	210	27	natural	natural	ADJ
easat-3242	210	28	images	image	NOUN
easat-3242	210	29	and	and	CCONJ
easat-3242	210	30	their	their	PRON
easat-3242	210	31	ability	ability	NOUN
easat-3242	210	32	to	to	PART
easat-3242	210	33	learn	learn	VERB
easat-3242	210	34	very	very	ADV
easat-3242	210	35	complex	complex	ADJ
easat-3242	210	36	hierarchical	hierarchical	ADJ
easat-3242	210	37	features	feature	NOUN
easat-3242	210	38	from	from	ADP
easat-3242	210	39	visual	visual	ADJ
easat-3242	210	40	data	datum	NOUN
easat-3242	210	41	makes	make	VERB
easat-3242	210	42	them	they	PRON
easat-3242	210	43	more	more	ADV
easat-3242	210	44	specialized	specialized	ADJ
easat-3242	210	45	for	for	ADP
easat-3242	210	46	pattern	pattern	NOUN
easat-3242	210	47	recognition	recognition	NOUN
easat-3242	210	48	in	in	ADP
easat-3242	210	49	satellite	satellite	NOUN
easat-3242	210	50	image	image	NOUN
easat-3242	210	51	.	.	PUNCT
easat-3242	211	1	the	the	DET
easat-3242	211	2	unet++vgg19	unet++vgg19	PROPN
easat-3242	211	3	model	model	NOUN
easat-3242	211	4	uses	use	VERB
easat-3242	211	5	vgg19	vgg19	PROPN
easat-3242	211	6	's	's	PART
easat-3242	211	7	pre	pre	ADJ
easat-3242	211	8	-	-	ADJ
easat-3242	211	9	trained	train	VERB
easat-3242	211	10	weights	weight	NOUN
easat-3242	211	11	typically	typically	ADV
easat-3242	211	12	trained	train	VERB
easat-3242	211	13	on	on	ADP
easat-3242	211	14	a	a	DET
easat-3242	211	15	large	large	ADJ
easat-3242	211	16	dataset	dataset	NOUN
easat-3242	211	17	like	like	ADP
easat-3242	211	18	imagenet	imagenet	NOUN
easat-3242	211	19	which	which	PRON
easat-3242	211	20	results	result	VERB
easat-3242	211	21	in	in	ADP
easat-3242	211	22	faster	fast	ADJ
easat-3242	211	23	convergence	convergence	NOUN
easat-3242	211	24	and	and	CCONJ
easat-3242	211	25	more	more	ADJ
easat-3242	211	26	generalisation	generalisation	NOUN
easat-3242	211	27	when	when	SCONJ
easat-3242	211	28	performing	perform	VERB
easat-3242	211	29	satellite	satellite	NOUN
easat-3242	211	30	image	image	NOUN
easat-3242	211	31	segmentation	segmentation	NOUN
easat-3242	211	32	.	.	PUNCT
easat-3242	212	1	however	however	ADV
easat-3242	212	2	,	,	PUNCT
easat-3242	212	3	pre	pre	ADJ
easat-3242	212	4	-	-	ADJ
easat-3242	212	5	training	training	ADJ
easat-3242	212	6	vgg19	vgg19	NOUN
easat-3242	212	7	captures	capture	VERB
easat-3242	212	8	important	important	ADJ
easat-3242	212	9	low	low	ADJ
easat-3242	212	10	-	-	PUNCT
easat-3242	212	11	level	level	NOUN
easat-3242	212	12	features	feature	NOUN
easat-3242	212	13	like	like	ADP
easat-3242	212	14	edges	edge	NOUN
easat-3242	212	15	or	or	CCONJ
easat-3242	212	16	textures	texture	NOUN
easat-3242	212	17	(	(	PUNCT
easat-3242	212	18	or	or	CCONJ
easat-3242	212	19	shapes	shape	NOUN
easat-3242	212	20	)	)	PUNCT
easat-3242	212	21	,	,	PUNCT
easat-3242	212	22	that	that	PRON
easat-3242	212	23	are	be	AUX
easat-3242	212	24	essential	essential	ADJ
easat-3242	212	25	to	to	PART
easat-3242	212	26	differentiate	differentiate	VERB
easat-3242	212	27	between	between	ADP
easat-3242	212	28	different	different	ADJ
easat-3242	212	29	land	land	NOUN
easat-3242	212	30	cover	cover	NOUN
easat-3242	212	31	classes	class	NOUN
easat-3242	212	32	in	in	ADP
easat-3242	212	33	the	the	DET
easat-3242	212	34	satellite	satellite	NOUN
easat-3242	212	35	images	image	NOUN
easat-3242	212	36	.	.	PUNCT
easat-3242	213	1	this	this	PRON
easat-3242	213	2	is	be	AUX
easat-3242	213	3	particularly	particularly	ADV
easat-3242	213	4	valuable	valuable	ADJ
easat-3242	213	5	in	in	ADP
easat-3242	213	6	the	the	DET
easat-3242	213	7	task	task	NOUN
easat-3242	213	8	of	of	ADP
easat-3242	213	9	segmenting	segment	VERB
easat-3242	213	10	satellite	satellite	NOUN
easat-3242	213	11	images	image	NOUN
easat-3242	213	12	,	,	PUNCT
easat-3242	213	13	where	where	SCONJ
easat-3242	213	14	objects	object	NOUN
easat-3242	213	15	can	can	AUX
easat-3242	213	16	come	come	VERB
easat-3242	213	17	with	with	ADP
easat-3242	213	18	many	many	ADJ
easat-3242	213	19	unique	unique	ADJ
easat-3242	213	20	sizes	size	NOUN
easat-3242	213	21	and	and	CCONJ
easat-3242	213	22	shapes	shape	NOUN
easat-3242	213	23	,	,	PUNCT
easat-3242	213	24	and	and	CCONJ
easat-3242	213	25	be	be	AUX
easat-3242	213	26	placed	place	VERB
easat-3242	213	27	into	into	ADP
easat-3242	213	28	complex	complex	ADJ
easat-3242	213	29	and	and	CCONJ
easat-3242	213	30	noisy	noisy	ADJ
easat-3242	213	31	surroundings	surrounding	NOUN
easat-3242	213	32	.	.	PUNCT
easat-3242	214	1	the	the	DET
easat-3242	214	2	pre	pre	ADJ
easat-3242	214	3	-	-	ADJ
easat-3242	214	4	trained	trained	ADJ
easat-3242	214	5	weights	weight	NOUN
easat-3242	214	6	allow	allow	VERB
easat-3242	214	7	the	the	DET
easat-3242	214	8	model	model	NOUN
easat-3242	214	9	to	to	PART
easat-3242	214	10	begin	begin	VERB
easat-3242	214	11	with	with	ADP
easat-3242	214	12	some	some	PRON
easat-3242	214	13	started	start	VERB
easat-3242	214	14	learning	learn	VERB
easat-3242	214	15	features	feature	NOUN
easat-3242	214	16	,	,	PUNCT
easat-3242	214	17	speeding	speed	VERB
easat-3242	214	18	up	up	ADP
easat-3242	214	19	training	training	NOUN
easat-3242	214	20	and	and	CCONJ
easat-3242	214	21	boosting	boost	VERB
easat-3242	214	22	performance	performance	NOUN
easat-3242	214	23	in	in	ADP
easat-3242	214	24	satellite	satellite	NOUN
easat-3242	214	25	image	image	NOUN
easat-3242	214	26	segmentation	segmentation	NOUN
easat-3242	214	27	.	.	PUNCT
easat-3242	215	1	by	by	ADP
easat-3242	215	2	grabbing	grab	VERB
easat-3242	215	3	on	on	ADP
easat-3242	215	4	vgg19	vgg19	NOUN
easat-3242	215	5	this	this	DET
easat-3242	215	6	multi	multi	ADJ
easat-3242	215	7	-	-	ADJ
easat-3242	215	8	scale	scale	ADJ
easat-3242	215	9	information	information	NOUN
easat-3242	215	10	,	,	PUNCT
easat-3242	215	11	can	can	AUX
easat-3242	215	12	help	help	VERB
easat-3242	215	13	unet++	unet++	PRON
easat-3242	215	14	to	to	PART
easat-3242	215	15	draw	draw	VERB
easat-3242	215	16	better	well	ADJ
easat-3242	215	17	sharp	sharp	ADJ
easat-3242	215	18	boundaries	boundary	NOUN
easat-3242	215	19	for	for	ADP
easat-3242	215	20	satellite	satellite	NOUN
easat-3242	215	21	images	image	NOUN
easat-3242	215	22	.	.	PUNCT
easat-3242	216	1	another	another	DET
easat-3242	216	2	reason	reason	NOUN
easat-3242	216	3	lies	lie	VERB
easat-3242	216	4	in	in	ADP
easat-3242	216	5	vgg19	vgg19	PROPN
easat-3242	216	6	's	's	PART
easat-3242	216	7	wider	wide	ADJ
easat-3242	216	8	convolutional	convolutional	ADJ
easat-3242	216	9	layers	layer	NOUN
easat-3242	216	10	that	that	PRON
easat-3242	216	11	include	include	VERB
easat-3242	216	12	small	small	ADJ
easat-3242	216	13	receptive	receptive	ADJ
easat-3242	216	14	fields	field	NOUN
easat-3242	216	15	which	which	PRON
easat-3242	216	16	help	help	VERB
easat-3242	216	17	capture	capture	VERB
easat-3242	216	18	certain	certain	ADJ
easat-3242	216	19	subtle	subtle	ADJ
easat-3242	216	20	details	detail	NOUN
easat-3242	216	21	(	(	PUNCT
easat-3242	216	22	microscopic	microscopic	ADJ
easat-3242	216	23	patterns	pattern	NOUN
easat-3242	216	24	)	)	PUNCT
easat-3242	216	25	harder	hard	ADV
easat-3242	216	26	to	to	PART
easat-3242	216	27	notice	notice	VERB
easat-3242	216	28	for	for	ADP
easat-3242	216	29	some	some	PRON
easat-3242	216	30	deeper	deep	ADJ
easat-3242	216	31	,	,	PUNCT
easat-3242	216	32	additional	additional	ADJ
easat-3242	216	33	architecture	architecture	NOUN
easat-3242	216	34	.	.	PUNCT
easat-3242	217	1	this	this	DET
easat-3242	217	2	mechanism	mechanism	NOUN
easat-3242	217	3	of	of	ADP
easat-3242	217	4	extracting	extract	VERB
easat-3242	217	5	better	well	ADJ
easat-3242	217	6	features	feature	NOUN
easat-3242	217	7	and	and	CCONJ
easat-3242	217	8	the	the	DET
easat-3242	217	9	use	use	NOUN
easat-3242	217	10	of	of	ADP
easat-3242	217	11	nested	nested	ADJ
easat-3242	217	12	skip	skip	ADJ
easat-3242	217	13	connections	connection	NOUN
easat-3242	217	14	in	in	ADP
easat-3242	217	15	unet++	unet++	PROPN
easat-3242	217	16	,	,	PUNCT
easat-3242	217	17	allows	allow	VERB
easat-3242	217	18	the	the	DET
easat-3242	217	19	model	model	NOUN
easat-3242	217	20	to	to	PART
easat-3242	217	21	carry	carry	VERB
easat-3242	217	22	high	high	ADJ
easat-3242	217	23	-	-	PUNCT
easat-3242	217	24	resolution	resolution	NOUN
easat-3242	217	25	,	,	PUNCT
easat-3242	217	26	higher	high	ADJ
easat-3242	217	27	level	level	NOUN
easat-3242	217	28	semantics	semantic	NOUN
easat-3242	217	29	through	through	ADP
easat-3242	217	30	all	all	DET
easat-3242	217	31	stages	stage	NOUN
easat-3242	217	32	:	:	PUNCT
easat-3242	217	33	this	this	PRON
easat-3242	217	34	helps	help	VERB
easat-3242	217	35	getting	get	VERB
easat-3242	217	36	more	more	ADV
easat-3242	217	37	accurate	accurate	ADJ
easat-3242	217	38	segmentation	segmentation	NOUN
easat-3242	217	39	results	result	NOUN
easat-3242	217	40	.	.	PUNCT
easat-3242	218	1	therefore	therefore	ADV
easat-3242	218	2	,	,	PUNCT
easat-3242	218	3	the	the	DET
easat-3242	218	4	high	high	ADJ
easat-3242	218	5	performance	performance	NOUN
easat-3242	218	6	of	of	ADP
easat-3242	218	7	unet++vgg19	unet++vgg19	PROPN
easat-3242	218	8	in	in	ADP
easat-3242	218	9	capturing	capture	VERB
easat-3242	218	10	both	both	CCONJ
easat-3242	218	11	local	local	ADJ
easat-3242	218	12	and	and	CCONJ
easat-3242	218	13	global	global	ADJ
easat-3242	218	14	features	feature	NOUN
easat-3242	218	15	can	can	AUX
easat-3242	218	16	make	make	VERB
easat-3242	218	17	it	it	PRON
easat-3242	218	18	an	an	DET
easat-3242	218	19	ideal	ideal	ADJ
easat-3242	218	20	choice	choice	NOUN
easat-3242	218	21	for	for	ADP
easat-3242	218	22	applications	application	NOUN
easat-3242	218	23	such	such	ADJ
easat-3242	218	24	as	as	ADP
easat-3242	218	25	land	land	NOUN
easat-3242	218	26	cover	cover	NOUN
easat-3242	218	27	classification	classification	NOUN
easat-3242	218	28	,	,	PUNCT
easat-3242	218	29	urban	urban	ADJ
easat-3242	218	30	planning	planning	NOUN
easat-3242	218	31	or	or	CCONJ
easat-3242	218	32	disaster	disaster	NOUN
easat-3242	218	33	response	response	NOUN
easat-3242	218	34	where	where	SCONJ
easat-3242	218	35	reliable	reliable	ADJ
easat-3242	218	36	delimitation	delimitation	NOUN
easat-3242	218	37	of	of	ADP
easat-3242	218	38	different	different	ADJ
easat-3242	218	39	types	type	NOUN
easat-3242	218	40	of	of	ADP
easat-3242	218	41	terrains	terrain	NOUN
easat-3242	218	42	is	be	AUX
easat-3242	218	43	needed	need	VERB
easat-3242	218	44	.	.	PUNCT
easat-3242	219	1	in	in	ADP
easat-3242	219	2	general	general	ADJ
easat-3242	219	3	,	,	PUNCT
easat-3242	219	4	the	the	DET
easat-3242	219	5	application	application	NOUN
easat-3242	219	6	of	of	ADP
easat-3242	219	7	unet++	unet++	PRON
easat-3242	219	8	to	to	PART
easat-3242	219	9	vgg19	vgg19	VERB
easat-3242	219	10	contributes	contribute	VERB
easat-3242	219	11	important	important	ADJ
easat-3242	219	12	findings	finding	NOUN
easat-3242	219	13	in	in	ADP
easat-3242	219	14	satellite	satellite	NOUN
easat-3242	219	15	image	image	NOUN
easat-3242	219	16	segmentation	segmentation	NOUN
easat-3242	219	17	.	.	PUNCT
easat-3242	220	1	thus	thus	ADV
easat-3242	220	2	,	,	PUNCT
easat-3242	220	3	the	the	DET
easat-3242	220	4	model	model	NOUN
easat-3242	220	5	is	be	AUX
easat-3242	220	6	great	great	ADJ
easat-3242	220	7	for	for	ADP
easat-3242	220	8	high	high	ADJ
easat-3242	220	9	-	-	PUNCT
easat-3242	220	10	accuracy	accuracy	NOUN
easat-3242	220	11	and	and	CCONJ
easat-3242	220	12	precision	precision	NOUN
easat-3242	220	13	-	-	PUNCT
easat-3242	220	14	demanding	demand	VERB
easat-3242	220	15	tasks	task	NOUN
easat-3242	220	16	which	which	PRON
easat-3242	220	17	benefits	benefit	VERB
easat-3242	220	18	from	from	ADP
easat-3242	220	19	optimal	optimal	ADJ
easat-3242	220	20	feature	feature	NOUN
easat-3242	220	21	propagation	propagation	NOUN
easat-3242	220	22	as	as	ADP
easat-3242	220	23	in	in	ADP
easat-3242	220	24	unet++	unet++	NOUN
easat-3242	220	25	along	along	ADP
easat-3242	220	26	with	with	ADP
easat-3242	220	27	robust	robust	ADJ
easat-3242	220	28	feature	feature	NOUN
easat-3242	220	29	extraction	extraction	NOUN
easat-3242	220	30	through	through	ADP
easat-3242	220	31	vgg19	vgg19	PROPN
easat-3242	220	32	.	.	PUNCT
easat-3242	221	1	5715	5715	NUM
easat-3242	221	2	edelweiss	edelweiss	PROPN
easat-3242	221	3	applied	apply	VERB
easat-3242	221	4	science	science	NOUN
easat-3242	221	5	and	and	CCONJ
easat-3242	221	6	technology	technology	NOUN
easat-3242	221	7	issn	issn	PROPN
easat-3242	221	8	:	:	PUNCT
easat-3242	221	9	2576	2576	NUM
easat-3242	221	10	-	-	SYM
easat-3242	221	11	8484	8484	NUM
easat-3242	221	12	vol	vol	NOUN
easat-3242	221	13	.	.	PROPN
easat-3242	221	14	8	8	NUM
easat-3242	221	15	,	,	PUNCT
easat-3242	221	16	no	no	INTJ
easat-3242	221	17	.	.	NOUN
easat-3242	221	18	6	6	NUM
easat-3242	221	19	:	:	SYM
easat-3242	221	20	5707	5707	NUM
easat-3242	221	21	-	-	SYM
easat-3242	221	22	5722	5722	NUM
easat-3242	221	23	,	,	PUNCT
easat-3242	221	24	2024	2024	NUM
easat-3242	221	25	doi	doi	NOUN
easat-3242	221	26	:	:	PUNCT
easat-3242	221	27	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	221	28	©	©	PROPN
easat-3242	221	29	2024	2024	NUM
easat-3242	221	30	by	by	ADP
easat-3242	221	31	the	the	DET
easat-3242	221	32	authors	author	NOUN
easat-3242	221	33	;	;	PUNCT
easat-3242	221	34	licensee	licensee	PROPN
easat-3242	221	35	learning	learning	NOUN
easat-3242	221	36	gate	gate	NOUN
easat-3242	221	37	due	due	ADP
easat-3242	221	38	to	to	ADP
easat-3242	221	39	its	its	PRON
easat-3242	221	40	efficiency	efficiency	NOUN
easat-3242	221	41	in	in	ADP
easat-3242	221	42	carrying	carry	VERB
easat-3242	221	43	out	out	ADP
easat-3242	221	44	computational	computational	ADJ
easat-3242	221	45	and	and	CCONJ
easat-3242	221	46	analytical	analytical	ADJ
easat-3242	221	47	operations	operation	NOUN
easat-3242	221	48	,	,	PUNCT
easat-3242	221	49	as	as	ADV
easat-3242	221	50	well	well	ADV
easat-3242	221	51	capturing	capture	VERB
easat-3242	221	52	contextual	contextual	ADJ
easat-3242	221	53	information	information	NOUN
easat-3242	221	54	at	at	ADP
easat-3242	221	55	multiple	multiple	ADJ
easat-3242	221	56	scales	scale	NOUN
easat-3242	221	57	for	for	ADP
easat-3242	221	58	multi	multi	ADJ
easat-3242	221	59	-	-	ADJ
easat-3242	221	60	sensor	sensor	ADJ
easat-3242	221	61	image	image	NOUN
easat-3242	221	62	fusion	fusion	NOUN
easat-3242	221	63	of	of	ADP
easat-3242	221	64	complex	complex	ADJ
easat-3242	221	65	visual	visual	ADJ
easat-3242	221	66	data	datum	NOUN
easat-3242	221	67	;	;	PUNCT
easat-3242	221	68	geographic	geographic	ADJ
easat-3242	221	69	community	community	NOUN
easat-3242	221	70	find	find	VERB
easat-3242	221	71	it	it	PRON
easat-3242	221	72	the	the	DET
easat-3242	221	73	best	well	ADV
easat-3242	221	74	suited	suited	ADJ
easat-3242	221	75	technology	technology	NOUN
easat-3242	221	76	especially	especially	ADV
easat-3242	221	77	for	for	ADP
easat-3242	221	78	environmental	environmental	ADJ
easat-3242	221	79	monitoring	monitoring	NOUN
easat-3242	221	80	,	,	PUNCT
easat-3242	221	81	urban	urban	ADJ
easat-3242	221	82	mapping	mapping	NOUN
easat-3242	221	83	applications	application	NOUN
easat-3242	221	84	disaster	disaster	NOUN
easat-3242	221	85	management	management	NOUN
easat-3242	221	86	etc	etc	X
easat-3242	221	87	.	.	X
easat-3242	221	88	combined	combine	VERB
easat-3242	221	89	with	with	ADP
easat-3242	221	90	superior	superior	ADJ
easat-3242	221	91	architectural	architectural	ADJ
easat-3242	221	92	foundations	foundation	NOUN
easat-3242	221	93	,	,	PUNCT
easat-3242	221	94	the	the	DET
easat-3242	221	95	unet++vgg19	unet++vgg19	PROPN
easat-3242	221	96	model	model	NOUN
easat-3242	221	97	makes	make	VERB
easat-3242	221	98	a	a	DET
easat-3242	221	99	formidable	formidable	ADJ
easat-3242	221	100	tool	tool	NOUN
easat-3242	221	101	to	to	PART
easat-3242	221	102	help	help	VERB
easat-3242	221	103	address	address	VERB
easat-3242	221	104	limitations	limitation	NOUN
easat-3242	221	105	around	around	ADP
easat-3242	221	106	satellite	satellite	NOUN
easat-3242	221	107	image	image	NOUN
easat-3242	221	108	segmentation	segmentation	NOUN
easat-3242	221	109	in	in	ADP
easat-3242	221	110	diverse	diverse	ADJ
easat-3242	221	111	and	and	CCONJ
easat-3242	221	112	dynamic	dynamic	ADJ
easat-3242	221	113	surroundings	surrounding	NOUN
easat-3242	221	114	.	.	PUNCT
easat-3242	222	1	4	4	X
easat-3242	222	2	.	.	X
easat-3242	222	3	experimental	experimental	ADJ
easat-3242	222	4	results	result	NOUN
easat-3242	222	5	in	in	ADP
easat-3242	222	6	this	this	DET
easat-3242	222	7	section	section	NOUN
easat-3242	222	8	,	,	PUNCT
easat-3242	222	9	the	the	DET
easat-3242	222	10	complete	complete	ADJ
easat-3242	222	11	results	result	NOUN
easat-3242	222	12	of	of	ADP
easat-3242	222	13	performing	perform	VERB
easat-3242	222	14	simulation	simulation	NOUN
easat-3242	222	15	over	over	ADP
easat-3242	222	16	suggested	suggest	VERB
easat-3242	222	17	methodology	methodology	NOUN
easat-3242	222	18	is	be	AUX
easat-3242	222	19	presented	present	VERB
easat-3242	222	20	.	.	PUNCT
easat-3242	223	1	the	the	DET
easat-3242	223	2	kaggle	kaggle	NOUN
easat-3242	223	3	is	be	AUX
easat-3242	223	4	the	the	DET
easat-3242	223	5	source	source	NOUN
easat-3242	223	6	of	of	ADP
easat-3242	223	7	this	this	DET
easat-3242	223	8	investigation	investigation	NOUN
easat-3242	223	9	dataset	dataset	VERB
easat-3242	223	10	details	detail	NOUN
easat-3242	223	11	.	.	PUNCT
easat-3242	224	1	(	(	PUNCT
easat-3242	224	2	a	a	X
easat-3242	224	3	)	)	PUNCT
easat-3242	224	4	(	(	PUNCT
easat-3242	224	5	i	i	NOUN
easat-3242	224	6	)	)	PUNCT
easat-3242	224	7	(	(	PUNCT
easat-3242	224	8	b	b	X
easat-3242	224	9	)	)	PUNCT
easat-3242	224	10	(	(	PUNCT
easat-3242	224	11	ii	ii	NOUN
easat-3242	224	12	)	)	PUNCT
easat-3242	224	13	5716	5716	NUM
easat-3242	224	14	edelweiss	edelweiss	PROPN
easat-3242	224	15	applied	apply	VERB
easat-3242	224	16	science	science	NOUN
easat-3242	224	17	and	and	CCONJ
easat-3242	224	18	technology	technology	NOUN
easat-3242	224	19	issn	issn	PROPN
easat-3242	224	20	:	:	PUNCT
easat-3242	224	21	2576	2576	NUM
easat-3242	224	22	-	-	SYM
easat-3242	224	23	8484	8484	NUM
easat-3242	224	24	vol	vol	NOUN
easat-3242	224	25	.	.	PROPN
easat-3242	225	1	8	8	NUM
easat-3242	225	2	,	,	PUNCT
easat-3242	225	3	no	no	INTJ
easat-3242	225	4	.	.	NOUN
easat-3242	226	1	6	6	NUM
easat-3242	226	2	:	:	SYM
easat-3242	226	3	5707	5707	NUM
easat-3242	226	4	-	-	SYM
easat-3242	226	5	5722	5722	NUM
easat-3242	226	6	,	,	PUNCT
easat-3242	226	7	2024	2024	NUM
easat-3242	226	8	doi	doi	NOUN
easat-3242	226	9	:	:	PUNCT
easat-3242	226	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	226	11	©	©	PROPN
easat-3242	226	12	2024	2024	NUM
easat-3242	226	13	by	by	ADP
easat-3242	226	14	the	the	DET
easat-3242	226	15	authors	author	NOUN
easat-3242	226	16	;	;	PUNCT
easat-3242	226	17	licensee	licensee	PROPN
easat-3242	226	18	learning	learning	NOUN
easat-3242	226	19	gate	gate	NOUN
easat-3242	226	20	(	(	PUNCT
easat-3242	226	21	c	c	NOUN
easat-3242	226	22	)	)	PUNCT
easat-3242	226	23	(	(	PUNCT
easat-3242	226	24	iii	iii	NOUN
easat-3242	226	25	)	)	PUNCT
easat-3242	226	26	(	(	PUNCT
easat-3242	226	27	d	d	X
easat-3242	226	28	)	)	PUNCT
easat-3242	226	29	(	(	PUNCT
easat-3242	226	30	iv	iv	X
easat-3242	226	31	)	)	PUNCT
easat-3242	226	32	(	(	PUNCT
easat-3242	226	33	e	e	NOUN
easat-3242	226	34	)	)	PUNCT
easat-3242	226	35	(	(	PUNCT
easat-3242	226	36	v	v	NOUN
easat-3242	226	37	)	)	PUNCT
easat-3242	226	38	5717	5717	NUM
easat-3242	226	39	edelweiss	edelweiss	PROPN
easat-3242	226	40	applied	apply	VERB
easat-3242	226	41	science	science	NOUN
easat-3242	226	42	and	and	CCONJ
easat-3242	226	43	technology	technology	NOUN
easat-3242	226	44	issn	issn	PROPN
easat-3242	226	45	:	:	PUNCT
easat-3242	226	46	2576	2576	NUM
easat-3242	226	47	-	-	SYM
easat-3242	226	48	8484	8484	NUM
easat-3242	226	49	vol	vol	NOUN
easat-3242	226	50	.	.	PROPN
easat-3242	226	51	8	8	NUM
easat-3242	226	52	,	,	PUNCT
easat-3242	226	53	no	no	INTJ
easat-3242	226	54	.	.	NOUN
easat-3242	226	55	6	6	NUM
easat-3242	226	56	:	:	SYM
easat-3242	226	57	5707	5707	NUM
easat-3242	226	58	-	-	SYM
easat-3242	226	59	5722	5722	NUM
easat-3242	226	60	,	,	PUNCT
easat-3242	226	61	2024	2024	NUM
easat-3242	226	62	doi	doi	NOUN
easat-3242	226	63	:	:	PUNCT
easat-3242	226	64	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	226	65	©	©	PROPN
easat-3242	226	66	2024	2024	NUM
easat-3242	226	67	by	by	ADP
easat-3242	226	68	the	the	DET
easat-3242	226	69	authors	author	NOUN
easat-3242	226	70	;	;	PUNCT
easat-3242	226	71	licensee	licensee	PROPN
easat-3242	226	72	learning	learning	NOUN
easat-3242	226	73	gate	gate	NOUN
easat-3242	226	74	(	(	PUNCT
easat-3242	226	75	f	f	X
easat-3242	226	76	)	)	PUNCT
easat-3242	226	77	(	(	PUNCT
easat-3242	226	78	vi	vi	NOUN
easat-3242	226	79	)	)	PUNCT
easat-3242	226	80	figure	figure	NOUN
easat-3242	226	81	2	2	NUM
easat-3242	226	82	.	.	PUNCT
easat-3242	226	83	sample	sample	NOUN
easat-3242	226	84	images	image	NOUN
easat-3242	226	85	from	from	ADP
easat-3242	226	86	dataset	dataset	NOUN
easat-3242	226	87	.	.	PUNCT
easat-3242	227	1	figure	figure	NOUN
easat-3242	227	2	3	3	NUM
easat-3242	227	3	.	.	PUNCT
easat-3242	227	4	train	train	NOUN
easat-3242	227	5	loss	loss	NOUN
easat-3242	227	6	vs	vs	ADP
easat-3242	227	7	validation	validation	NOUN
easat-3242	227	8	loss	loss	NOUN
easat-3242	227	9	.	.	PUNCT
easat-3242	228	1	figure	figure	NOUN
easat-3242	228	2	3	3	NUM
easat-3242	228	3	also	also	ADV
easat-3242	228	4	shows	show	VERB
easat-3242	228	5	the	the	DET
easat-3242	228	6	training	training	NOUN
easat-3242	228	7	and	and	CCONJ
easat-3242	228	8	validation	validation	NOUN
easat-3242	228	9	loss	loss	NOUN
easat-3242	228	10	over	over	ADP
easat-3242	228	11	100	100	NUM
easat-3242	228	12	epochs	epoch	NOUN
easat-3242	228	13	,	,	PUNCT
easat-3242	228	14	indicating	indicate	VERB
easat-3242	228	15	how	how	SCONJ
easat-3242	228	16	much	much	ADV
easat-3242	228	17	better	well	ADJ
easat-3242	228	18	or	or	CCONJ
easat-3242	228	19	worse	bad	ADJ
easat-3242	228	20	our	our	PRON
easat-3242	228	21	model	model	NOUN
easat-3242	228	22	is	be	AUX
easat-3242	228	23	performing	perform	VERB
easat-3242	228	24	across	across	ADP
easat-3242	228	25	different	different	ADJ
easat-3242	228	26	epochs	epoch	NOUN
easat-3242	228	27	.	.	PUNCT
easat-3242	229	1	•	•	NUM
easat-3242	229	2	training	training	NOUN
easat-3242	229	3	loss	loss	NOUN
easat-3242	229	4	:	:	PUNCT
easat-3242	229	5	when	when	SCONJ
easat-3242	229	6	the	the	DET
easat-3242	229	7	model	model	NOUN
easat-3242	229	8	is	be	AUX
easat-3242	229	9	being	be	AUX
easat-3242	229	10	trained	train	VERB
easat-3242	229	11	,	,	PUNCT
easat-3242	229	12	it	it	PRON
easat-3242	229	13	tries	try	VERB
easat-3242	229	14	to	to	PART
easat-3242	229	15	reduce	reduce	VERB
easat-3242	229	16	the	the	DET
easat-3242	229	17	error	error	NOUN
easat-3242	229	18	in	in	ADP
easat-3242	229	19	on	on	ADP
easat-3242	229	20	training	training	NOUN
easat-3242	229	21	data	datum	NOUN
easat-3242	229	22	set	set	VERB
easat-3242	229	23	.	.	PUNCT
easat-3242	230	1	in	in	ADP
easat-3242	230	2	the	the	DET
easat-3242	230	3	fig	fig	NOUN
easat-3242	230	4	above	above	ADV
easat-3242	230	5	,	,	PUNCT
easat-3242	230	6	you	you	PRON
easat-3242	230	7	can	can	AUX
easat-3242	230	8	see	see	VERB
easat-3242	230	9	that	that	SCONJ
easat-3242	230	10	training	training	NOUN
easat-3242	230	11	loss	loss	NOUN
easat-3242	230	12	is	be	AUX
easat-3242	230	13	decreasing	decrease	VERB
easat-3242	230	14	rapidly	rapidly	ADV
easat-3242	230	15	in	in	ADP
easat-3242	230	16	first	first	ADJ
easat-3242	230	17	20	20	NUM
easat-3242	230	18	epochs	epoch	NOUN
easat-3242	230	19	and	and	CCONJ
easat-3242	230	20	then	then	ADV
easat-3242	230	21	slowly	slowly	ADV
easat-3242	230	22	this	this	PRON
easat-3242	230	23	indicates	indicate	VERB
easat-3242	230	24	algorithm	algorithm	NOUN
easat-3242	230	25	is	be	AUX
easat-3242	230	26	learning	learn	VERB
easat-3242	230	27	patters	patter	NOUN
easat-3242	230	28	from	from	ADP
easat-3242	230	29	data	datum	NOUN
easat-3242	230	30	and	and	CCONJ
easat-3242	230	31	improving	improve	VERB
easat-3242	230	32	itself	itself	PRON
easat-3242	230	33	on	on	ADP
easat-3242	230	34	its	its	PRON
easat-3242	230	35	5718	5718	NUM
easat-3242	230	36	edelweiss	edelweiss	PROPN
easat-3242	230	37	applied	apply	VERB
easat-3242	230	38	science	science	NOUN
easat-3242	230	39	and	and	CCONJ
easat-3242	230	40	technology	technology	NOUN
easat-3242	230	41	issn	issn	PROPN
easat-3242	230	42	:	:	PUNCT
easat-3242	230	43	2576	2576	NUM
easat-3242	230	44	-	-	SYM
easat-3242	230	45	8484	8484	NUM
easat-3242	230	46	vol	vol	NOUN
easat-3242	230	47	.	.	PROPN
easat-3242	231	1	8	8	NUM
easat-3242	231	2	,	,	PUNCT
easat-3242	231	3	no	no	INTJ
easat-3242	231	4	.	.	NOUN
easat-3242	232	1	6	6	NUM
easat-3242	232	2	:	:	SYM
easat-3242	232	3	5707	5707	NUM
easat-3242	232	4	-	-	SYM
easat-3242	232	5	5722	5722	NUM
easat-3242	232	6	,	,	PUNCT
easat-3242	232	7	2024	2024	NUM
easat-3242	232	8	doi	doi	NOUN
easat-3242	232	9	:	:	PUNCT
easat-3242	232	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	232	11	©	©	PROPN
easat-3242	232	12	2024	2024	NUM
easat-3242	232	13	by	by	ADP
easat-3242	232	14	the	the	DET
easat-3242	232	15	authors	author	NOUN
easat-3242	232	16	;	;	PUNCT
easat-3242	233	1	licensee	licensee	PROPN
easat-3242	233	2	learning	learn	VERB
easat-3242	233	3	gate	gate	PROPN
easat-3242	233	4	prediction	prediction	NOUN
easat-3242	233	5	.	.	PUNCT
easat-3242	234	1	training	training	NOUN
easat-3242	234	2	loss	loss	NOUN
easat-3242	234	3	going	go	VERB
easat-3242	234	4	down	down	ADP
easat-3242	234	5	means	mean	VERB
easat-3242	234	6	the	the	DET
easat-3242	234	7	model	model	NOUN
easat-3242	234	8	is	be	AUX
easat-3242	234	9	fitting	fit	VERB
easat-3242	234	10	the	the	DET
easat-3242	234	11	training	training	NOUN
easat-3242	234	12	data	datum	NOUN
easat-3242	234	13	,	,	PUNCT
easat-3242	234	14	but	but	CCONJ
easat-3242	234	15	that	that	PRON
easat-3242	234	16	does	do	AUX
easat-3242	234	17	not	not	PART
easat-3242	234	18	imply	imply	VERB
easat-3242	234	19	it	it	PRON
easat-3242	234	20	's	be	AUX
easat-3242	234	21	doing	do	VERB
easat-3242	234	22	good	good	NOUN
easat-3242	234	23	.	.	PUNCT
easat-3242	235	1	but	but	CCONJ
easat-3242	235	2	if	if	SCONJ
easat-3242	235	3	training	training	NOUN
easat-3242	235	4	loss	loss	NOUN
easat-3242	235	5	is	be	AUX
easat-3242	235	6	decreasing	decrease	VERB
easat-3242	235	7	but	but	CCONJ
easat-3242	235	8	validation	validation	NOUN
easat-3242	235	9	set	set	NOUN
easat-3242	235	10	,	,	PUNCT
easat-3242	235	11	it	it	PRON
easat-3242	235	12	will	will	AUX
easat-3242	235	13	be	be	AUX
easat-3242	235	14	overfitting	overfitte	VERB
easat-3242	235	15	in	in	ADP
easat-3242	235	16	which	which	DET
easat-3242	235	17	case	case	NOUN
easat-3242	235	18	.	.	PUNCT
easat-3242	236	1	in	in	ADP
easat-3242	236	2	this	this	DET
easat-3242	236	3	case	case	NOUN
easat-3242	236	4	,	,	PUNCT
easat-3242	236	5	make	make	VERB
easat-3242	236	6	sure	sure	ADJ
easat-3242	236	7	to	to	PART
easat-3242	236	8	monitor	monitor	VERB
easat-3242	236	9	both	both	DET
easat-3242	236	10	the	the	DET
easat-3242	236	11	losses	loss	NOUN
easat-3242	236	12	so	so	SCONJ
easat-3242	236	13	that	that	SCONJ
easat-3242	236	14	the	the	DET
easat-3242	236	15	model	model	NOUN
easat-3242	236	16	learns	learn	VERB
easat-3242	236	17	in	in	ADP
easat-3242	236	18	an	an	DET
easat-3242	236	19	appropriate	appropriate	ADJ
easat-3242	236	20	balanced	balanced	ADJ
easat-3242	236	21	fashion	fashion	NOUN
easat-3242	236	22	.	.	PUNCT
easat-3242	237	1	•	•	NUM
easat-3242	237	2	validation	validation	NOUN
easat-3242	237	3	loss	loss	NOUN
easat-3242	237	4	:	:	PUNCT
easat-3242	237	5	the	the	DET
easat-3242	237	6	validation	validation	NOUN
easat-3242	237	7	loss	loss	NOUN
easat-3242	237	8	which	which	PRON
easat-3242	237	9	represents	represent	VERB
easat-3242	237	10	how	how	SCONJ
easat-3242	237	11	good	good	ADJ
easat-3242	237	12	our	our	PRON
easat-3242	237	13	model	model	NOUN
easat-3242	237	14	can	can	AUX
easat-3242	237	15	generalize	generalize	VERB
easat-3242	237	16	over	over	ADP
easat-3242	237	17	unseen	unseen	ADJ
easat-3242	237	18	new	new	ADJ
easat-3242	237	19	data	datum	NOUN
easat-3242	237	20	.	.	PUNCT
easat-3242	238	1	it	it	PRON
easat-3242	238	2	is	be	AUX
easat-3242	238	3	initially	initially	ADV
easat-3242	238	4	high	high	ADJ
easat-3242	238	5	,	,	PUNCT
easat-3242	238	6	just	just	ADV
easat-3242	238	7	like	like	ADP
easat-3242	238	8	the	the	DET
easat-3242	238	9	training	training	NOUN
easat-3242	238	10	loss	loss	NOUN
easat-3242	238	11	and	and	CCONJ
easat-3242	238	12	drops	drop	NOUN
easat-3242	238	13	in	in	ADP
easat-3242	238	14	parallel	parallel	NOUN
easat-3242	238	15	to	to	ADP
easat-3242	238	16	lower	low	ADJ
easat-3242	238	17	values	value	NOUN
easat-3242	238	18	of	of	ADP
easat-3242	238	19	training	training	NOUN
easat-3242	238	20	loss	loss	NOUN
easat-3242	238	21	demonstrating	demonstrate	VERB
easat-3242	238	22	you	you	PRON
easat-3242	238	23	do	do	VERB
easat-3242	238	24	better	well	ADV
easat-3242	238	25	.	.	PUNCT
easat-3242	239	1	this	this	PRON
easat-3242	239	2	is	be	AUX
easat-3242	239	3	quite	quite	ADV
easat-3242	239	4	common	common	ADJ
easat-3242	239	5	as	as	SCONJ
easat-3242	239	6	the	the	DET
easat-3242	239	7	validation	validation	NOUN
easat-3242	239	8	set	set	NOUN
easat-3242	239	9	will	will	AUX
easat-3242	239	10	be	be	AUX
easat-3242	239	11	different	different	ADJ
easat-3242	239	12	from	from	ADP
easat-3242	239	13	training	training	NOUN
easat-3242	239	14	and	and	CCONJ
easat-3242	239	15	this	this	PRON
easat-3242	239	16	would	would	AUX
easat-3242	239	17	help	help	VERB
easat-3242	239	18	in	in	ADP
easat-3242	239	19	testing	test	VERB
easat-3242	239	20	how	how	SCONJ
easat-3242	239	21	well	well	ADV
easat-3242	239	22	or	or	CCONJ
easat-3242	239	23	bad	bad	ADJ
easat-3242	239	24	does	do	AUX
easat-3242	239	25	our	our	PRON
easat-3242	239	26	model	model	NOUN
easat-3242	239	27	generalize	generalize	VERB
easat-3242	239	28	on	on	ADP
easat-3242	239	29	unseen	unseen	ADJ
easat-3242	239	30	data	datum	NOUN
easat-3242	239	31	.	.	PUNCT
easat-3242	240	1	if	if	SCONJ
easat-3242	240	2	the	the	DET
easat-3242	240	3	validation	validation	NOUN
easat-3242	240	4	loss	loss	NOUN
easat-3242	240	5	starts	start	VERB
easat-3242	240	6	diverging	diverge	VERB
easat-3242	240	7	from	from	ADP
easat-3242	240	8	training	training	NOUN
easat-3242	240	9	loss	loss	NOUN
easat-3242	240	10	,	,	PUNCT
easat-3242	240	11	you	you	PRON
easat-3242	240	12	may	may	AUX
easat-3242	240	13	be	be	AUX
easat-3242	240	14	overfitting	overfitte	VERB
easat-3242	240	15	instead	instead	ADV
easat-3242	240	16	of	of	ADP
easat-3242	240	17	generalizing	generalize	VERB
easat-3242	240	18	.	.	PUNCT
easat-3242	241	1	if	if	SCONJ
easat-3242	241	2	,	,	PUNCT
easat-3242	241	3	however	however	ADV
easat-3242	241	4	,	,	PUNCT
easat-3242	241	5	the	the	DET
easat-3242	241	6	validation	validation	NOUN
easat-3242	241	7	loss	loss	NOUN
easat-3242	241	8	starts	start	VERB
easat-3242	241	9	rising	rise	VERB
easat-3242	241	10	but	but	CCONJ
easat-3242	241	11	training	training	NOUN
easat-3242	241	12	loss	loss	NOUN
easat-3242	241	13	still	still	ADV
easat-3242	241	14	decreases	decrease	VERB
easat-3242	241	15	your	your	PRON
easat-3242	241	16	model	model	NOUN
easat-3242	241	17	likely	likely	ADV
easat-3242	241	18	needs	need	VERB
easat-3242	241	19	regularization	regularization	NOUN
easat-3242	241	20	to	to	PART
easat-3242	241	21	avoid	avoid	VERB
easat-3242	241	22	overfitting	overfitte	VERB
easat-3242	241	23	and	and	CCONJ
easat-3242	241	24	become	become	VERB
easat-3242	241	25	more	more	ADV
easat-3242	241	26	generalizable	generalizable	ADJ
easat-3242	241	27	so	so	SCONJ
easat-3242	241	28	that	that	SCONJ
easat-3242	241	29	it	it	PRON
easat-3242	241	30	performs	perform	VERB
easat-3242	241	31	well	well	ADV
easat-3242	241	32	on	on	ADP
easat-3242	241	33	unseen	unseen	ADJ
easat-3242	241	34	data	datum	NOUN
easat-3242	241	35	.	.	PUNCT
easat-3242	242	1	figure	figure	NOUN
easat-3242	242	2	4	4	NUM
easat-3242	242	3	.	.	PUNCT
easat-3242	243	1	train	train	NOUN
easat-3242	243	2	miou	miou	NOUN
easat-3242	243	3	vs	vs	ADP
easat-3242	243	4	validation	validation	NOUN
easat-3242	243	5	miou	miou	NOUN
easat-3242	243	6	.	.	PUNCT
easat-3242	244	1	the	the	DET
easat-3242	244	2	mean	mean	ADJ
easat-3242	244	3	intersection	intersection	NOUN
easat-3242	244	4	over	over	ADP
easat-3242	244	5	union	union	NOUN
easat-3242	244	6	(	(	PUNCT
easat-3242	244	7	miou	miou	NOUN
easat-3242	244	8	)	)	PUNCT
easat-3242	244	9	metric	metric	NOUN
easat-3242	244	10	for	for	ADP
easat-3242	244	11	the	the	DET
easat-3242	244	12	training	training	NOUN
easat-3242	244	13	and	and	CCONJ
easat-3242	244	14	validation	validation	NOUN
easat-3242	244	15	datasets	dataset	NOUN
easat-3242	244	16	is	be	AUX
easat-3242	244	17	shown	show	VERB
easat-3242	244	18	as	as	ADP
easat-3242	244	19	a	a	DET
easat-3242	244	20	function	function	NOUN
easat-3242	244	21	of	of	ADP
easat-3242	244	22	100	100	NUM
easat-3242	244	23	epochs	epoch	NOUN
easat-3242	244	24	in	in	ADP
easat-3242	244	25	figure	figure	NOUN
easat-3242	244	26	4	4	NUM
easat-3242	244	27	.	.	PUNCT
easat-3242	245	1	miou	miou	NOUN
easat-3242	245	2	,	,	PUNCT
easat-3242	245	3	the	the	DET
easat-3242	245	4	intersection	intersection	NOUN
easat-3242	245	5	-	-	PUNCT
easat-3242	245	6	over	over	ADP
easat-3242	245	7	-	-	PUNCT
easat-3242	245	8	union	union	NOUN
easat-3242	245	9	(	(	PUNCT
easat-3242	245	10	iou	iou	NOUN
easat-3242	245	11	)	)	PUNCT
easat-3242	245	12	is	be	AUX
easat-3242	245	13	a	a	DET
easat-3242	245	14	common	common	ADJ
easat-3242	245	15	metric	metric	NOUN
easat-3242	245	16	to	to	PART
easat-3242	245	17	evaluate	evaluate	VERB
easat-3242	245	18	segmentation	segmentation	NOUN
easat-3242	245	19	models	model	NOUN
easat-3242	245	20	by	by	ADP
easat-3242	245	21	computing	compute	VERB
easat-3242	245	22	overlapping	overlap	VERB
easat-3242	245	23	region	region	NOUN
easat-3242	245	24	of	of	ADP
easat-3242	245	25	predicted	predict	VERB
easat-3242	245	26	mask	mask	NOUN
easat-3242	245	27	and	and	CCONJ
easat-3242	245	28	ground	ground	NOUN
easat-3242	245	29	truth	truth	NOUN
easat-3242	245	30	.	.	PUNCT
easat-3242	246	1	both	both	CCONJ
easat-3242	246	2	the	the	DET
easat-3242	246	3	train	train	NOUN
easat-3242	246	4	and	and	CCONJ
easat-3242	246	5	validation	validation	NOUN
easat-3242	246	6	miou	miou	NOUN
easat-3242	246	7	values	value	NOUN
easat-3242	246	8	increases	increase	VERB
easat-3242	246	9	slowly	slowly	ADV
easat-3242	246	10	to	to	PART
easat-3242	246	11	begin	begin	VERB
easat-3242	246	12	with	with	ADP
easat-3242	246	13	,	,	PUNCT
easat-3242	246	14	however	however	ADV
easat-3242	246	15	the	the	DET
easat-3242	246	16	training	training	NOUN
easat-3242	246	17	one	one	NOUN
easat-3242	246	18	gradually	gradually	ADV
easat-3242	246	19	improves	improve	VERB
easat-3242	246	20	faster	fast	ADV
easat-3242	246	21	.	.	PUNCT
easat-3242	247	1	after	after	ADP
easat-3242	247	2	around	around	ADV
easat-3242	247	3	20	20	NUM
easat-3242	247	4	epochs	epoch	NOUN
easat-3242	247	5	,	,	PUNCT
easat-3242	247	6	they	they	PRON
easat-3242	247	7	both	both	PRON
easat-3242	247	8	hover	hover	VERB
easat-3242	247	9	to	to	ADP
easat-3242	247	10	training	train	VERB
easat-3242	247	11	miou	miou	NOUN
easat-3242	247	12	of	of	ADP
easat-3242	247	13	~0.55	~0.55	PUNCT
easat-3242	247	14	and	and	CCONJ
easat-3242	247	15	validation	validation	NOUN
easat-3242	247	16	miou	miou	NOUN
easat-3242	247	17	about	about	ADP
easat-3242	247	18	to	to	PART
easat-3242	247	19	reach	reach	VERB
easat-3242	247	20	plateau	plateau	NOUN
easat-3242	247	21	a	a	DET
easat-3242	247	22	bit	bit	NOUN
easat-3242	247	23	below	below	ADP
easat-3242	247	24	~0.5	~0.5	NUM
easat-3242	247	25	.	.	PUNCT
easat-3242	248	1	this	this	PRON
easat-3242	248	2	is	be	AUX
easat-3242	248	3	an	an	DET
easat-3242	248	4	indication	indication	NOUN
easat-3242	248	5	that	that	SCONJ
easat-3242	248	6	the	the	DET
easat-3242	248	7	model	model	NOUN
easat-3242	248	8	in	in	ADP
easat-3242	248	9	fact	fact	NOUN
easat-3242	248	10	learning	learn	VERB
easat-3242	248	11	reasonable	reasonable	ADJ
easat-3242	248	12	segmentation	segmentation	NOUN
easat-3242	248	13	of	of	ADP
easat-3242	248	14	images	image	NOUN
easat-3242	248	15	on	on	ADP
easat-3242	248	16	both	both	PRON
easat-3242	248	17	training	training	NOUN
easat-3242	248	18	and	and	CCONJ
easat-3242	248	19	validation	validation	NOUN
easat-3242	248	20	sets	set	NOUN
easat-3242	248	21	.	.	PUNCT
easat-3242	249	1	the	the	DET
easat-3242	249	2	near	near	ADJ
easat-3242	249	3	values	value	NOUN
easat-3242	249	4	of	of	ADP
easat-3242	249	5	train	train	NOUN
easat-3242	249	6	and	and	CCONJ
easat-3242	249	7	validation	validation	NOUN
easat-3242	249	8	miou	miou	NOUN
easat-3242	249	9	indicate	indicate	VERB
easat-3242	249	10	no	no	DET
easat-3242	249	11	major	major	ADJ
easat-3242	249	12	overfitting	overfitting	NOUN
easat-3242	249	13	—	—	PUNCT
easat-3242	249	14	the	the	DET
easat-3242	249	15	model	model	NOUN
easat-3242	249	16	is	be	AUX
easat-3242	249	17	generalize	generalize	VERB
easat-3242	249	18	well	well	ADV
easat-3242	249	19	.	.	PUNCT
easat-3242	250	1	although	although	SCONJ
easat-3242	250	2	,	,	PUNCT
easat-3242	250	3	small	small	ADJ
easat-3242	250	4	5719	5719	NUM
easat-3242	250	5	edelweiss	edelweiss	PROPN
easat-3242	250	6	applied	apply	VERB
easat-3242	250	7	science	science	NOUN
easat-3242	250	8	and	and	CCONJ
easat-3242	250	9	technology	technology	NOUN
easat-3242	250	10	issn	issn	PROPN
easat-3242	250	11	:	:	PUNCT
easat-3242	250	12	2576	2576	NUM
easat-3242	250	13	-	-	SYM
easat-3242	250	14	8484	8484	NUM
easat-3242	250	15	vol	vol	NOUN
easat-3242	250	16	.	.	PROPN
easat-3242	250	17	8	8	NUM
easat-3242	250	18	,	,	PUNCT
easat-3242	250	19	no	no	INTJ
easat-3242	250	20	.	.	NOUN
easat-3242	251	1	6	6	NUM
easat-3242	251	2	:	:	SYM
easat-3242	251	3	5707	5707	NUM
easat-3242	251	4	-	-	SYM
easat-3242	251	5	5722	5722	NUM
easat-3242	251	6	,	,	PUNCT
easat-3242	251	7	2024	2024	NUM
easat-3242	251	8	doi	doi	NOUN
easat-3242	251	9	:	:	PUNCT
easat-3242	251	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	251	11	©	©	PROPN
easat-3242	251	12	2024	2024	NUM
easat-3242	251	13	by	by	ADP
easat-3242	251	14	the	the	DET
easat-3242	251	15	authors	author	NOUN
easat-3242	251	16	;	;	PUNCT
easat-3242	251	17	licensee	licensee	PROPN
easat-3242	251	18	learning	learn	VERB
easat-3242	251	19	gate	gate	NOUN
easat-3242	251	20	variations	variation	NOUN
easat-3242	251	21	on	on	ADP
easat-3242	251	22	validation	validation	NOUN
easat-3242	251	23	miou	miou	NOUN
easat-3242	251	24	can	can	AUX
easat-3242	251	25	also	also	ADV
easat-3242	251	26	be	be	AUX
easat-3242	251	27	attributed	attribute	VERB
easat-3242	251	28	to	to	ADP
easat-3242	251	29	the	the	DET
easat-3242	251	30	randomness	randomness	NOUN
easat-3242	251	31	involved	involve	VERB
easat-3242	251	32	when	when	SCONJ
easat-3242	251	33	dealing	deal	VERB
easat-3242	251	34	with	with	ADP
easat-3242	251	35	unseen	unseen	ADJ
easat-3242	251	36	data	datum	NOUN
easat-3242	251	37	.	.	PUNCT
easat-3242	252	1	figure	figure	NOUN
easat-3242	252	2	5	5	NUM
easat-3242	252	3	.	.	PUNCT
easat-3242	252	4	train	train	NOUN
easat-3242	252	5	accuracy	accuracy	NOUN
easat-3242	252	6	vs	vs	ADP
easat-3242	252	7	validation	validation	NOUN
easat-3242	252	8	accuracy	accuracy	NOUN
easat-3242	252	9	.	.	PUNCT
easat-3242	253	1	figure	figure	NOUN
easat-3242	253	2	5	5	NUM
easat-3242	253	3	shows	show	VERB
easat-3242	253	4	the	the	DET
easat-3242	253	5	accuracy	accuracy	NOUN
easat-3242	253	6	of	of	ADP
easat-3242	253	7	the	the	DET
easat-3242	253	8	model	model	NOUN
easat-3242	253	9	on	on	ADP
easat-3242	253	10	training	training	NOUN
easat-3242	253	11	and	and	CCONJ
easat-3242	253	12	validation	validation	NOUN
easat-3242	253	13	datasets	dataset	NOUN
easat-3242	253	14	over	over	ADP
easat-3242	253	15	100	100	NUM
easat-3242	253	16	epochs	epoch	NOUN
easat-3242	253	17	.	.	PUNCT
easat-3242	254	1	•	•	NOUN
easat-3242	254	2	training	training	NOUN
easat-3242	254	3	accuracy	accuracy	NOUN
easat-3242	254	4	is	be	AUX
easat-3242	254	5	the	the	DET
easat-3242	254	6	model	model	NOUN
easat-3242	254	7	accuracy	accuracy	NOUN
easat-3242	254	8	on	on	ADP
easat-3242	254	9	training	training	NOUN
easat-3242	254	10	dataset	dataset	NOUN
easat-3242	254	11	while	while	SCONJ
easat-3242	254	12	each	each	DET
easat-3242	254	13	epoch	epoch	NOUN
easat-3242	254	14	.	.	PUNCT
easat-3242	255	1	it	it	PRON
easat-3242	255	2	specifies	specify	VERB
easat-3242	255	3	how	how	SCONJ
easat-3242	255	4	well	well	ADV
easat-3242	255	5	the	the	DET
easat-3242	255	6	model	model	NOUN
easat-3242	255	7	is	be	AUX
easat-3242	255	8	learning	learn	VERB
easat-3242	255	9	from	from	ADP
easat-3242	255	10	exposure	exposure	NOUN
easat-3242	255	11	to	to	ADP
easat-3242	255	12	data	datum	NOUN
easat-3242	255	13	.	.	PUNCT
easat-3242	256	1	notice	notice	VERB
easat-3242	256	2	the	the	DET
easat-3242	256	3	training	training	NOUN
easat-3242	256	4	accuracy	accuracy	NOUN
easat-3242	256	5	grows	grow	VERB
easat-3242	256	6	rapidly	rapidly	ADV
easat-3242	256	7	in	in	ADP
easat-3242	256	8	the	the	DET
easat-3242	256	9	first	first	ADJ
easat-3242	256	10	few	few	ADJ
easat-3242	256	11	epochs	epoch	NOUN
easat-3242	256	12	,	,	PUNCT
easat-3242	256	13	since	since	SCONJ
easat-3242	256	14	it	it	PRON
easat-3242	256	15	depicts	depict	VERB
easat-3242	256	16	how	how	SCONJ
easat-3242	256	17	well	well	ADV
easat-3242	256	18	a	a	DET
easat-3242	256	19	model	model	NOUN
easat-3242	256	20	learned	learn	VERB
easat-3242	256	21	its	its	PRON
easat-3242	256	22	parameters	parameter	NOUN
easat-3242	256	23	(	(	PUNCT
easat-3242	256	24	weights	weight	NOUN
easat-3242	256	25	)	)	PUNCT
easat-3242	256	26	to	to	PART
easat-3242	256	27	minimize	minimize	VERB
easat-3242	256	28	loss	loss	NOUN
easat-3242	256	29	.	.	PUNCT
easat-3242	257	1	accuracy	accuracy	NOUN
easat-3242	257	2	plateaus	plateaus	NOUN
easat-3242	257	3	at	at	ADP
easat-3242	257	4	around	around	ADV
easat-3242	257	5	20	20	NUM
easat-3242	257	6	epochs	epoch	NOUN
easat-3242	257	7	,	,	PUNCT
easat-3242	257	8	indicating	indicate	VERB
easat-3242	257	9	the	the	DET
easat-3242	257	10	model	model	NOUN
easat-3242	257	11	is	be	AUX
easat-3242	257	12	close	close	ADJ
easat-3242	257	13	to	to	ADP
easat-3242	257	14	optimal	optimal	ADJ
easat-3242	257	15	for	for	ADP
easat-3242	257	16	correctly	correctly	ADV
easat-3242	257	17	classifying	classify	VERB
easat-3242	257	18	training	training	NOUN
easat-3242	257	19	data	datum	NOUN
easat-3242	257	20	.	.	PUNCT
easat-3242	258	1	as	as	ADV
easat-3242	258	2	long	long	ADV
easat-3242	258	3	as	as	SCONJ
easat-3242	258	4	the	the	DET
easat-3242	258	5	training	training	NOUN
easat-3242	258	6	accuracy	accuracy	NOUN
easat-3242	258	7	keeps	keep	VERB
easat-3242	258	8	increasing	increase	VERB
easat-3242	258	9	,	,	PUNCT
easat-3242	258	10	it	it	PRON
easat-3242	258	11	means	mean	VERB
easat-3242	258	12	that	that	SCONJ
easat-3242	258	13	the	the	DET
easat-3242	258	14	model	model	NOUN
easat-3242	258	15	is	be	AUX
easat-3242	258	16	learning	learn	VERB
easat-3242	258	17	better	well	ADV
easat-3242	258	18	to	to	ADP
easat-3242	258	19	capturing	capture	VERB
easat-3242	258	20	more	more	ADJ
easat-3242	258	21	patterns	pattern	NOUN
easat-3242	258	22	in	in	ADP
easat-3242	258	23	a	a	DET
easat-3242	258	24	data	data	NOUN
easat-3242	258	25	.	.	PUNCT
easat-3242	259	1	•	•	NUM
easat-3242	259	2	validation	validation	NOUN
easat-3242	259	3	accuracy	accuracy	NOUN
easat-3242	259	4	used	use	VERB
easat-3242	259	5	to	to	PART
easat-3242	259	6	compare	compare	VERB
easat-3242	259	7	the	the	DET
easat-3242	259	8	generalization	generalization	NOUN
easat-3242	259	9	capability	capability	NOUN
easat-3242	259	10	of	of	ADP
easat-3242	259	11	a	a	DET
easat-3242	259	12	model	model	NOUN
easat-3242	259	13	when	when	SCONJ
easat-3242	259	14	it	it	PRON
easat-3242	259	15	encounters	encounter	VERB
easat-3242	259	16	some	some	DET
easat-3242	259	17	unseen	unseen	ADJ
easat-3242	259	18	data	datum	NOUN
easat-3242	259	19	,	,	PUNCT
easat-3242	259	20	which	which	PRON
easat-3242	259	21	does	do	AUX
easat-3242	259	22	n't	not	PART
easat-3242	259	23	contain	contain	VERB
easat-3242	259	24	in	in	ADP
easat-3242	259	25	training	training	NOUN
easat-3242	259	26	set	set	NOUN
easat-3242	259	27	.	.	PUNCT
easat-3242	260	1	training	training	NOUN
easat-3242	260	2	accuracy	accuracy	NOUN
easat-3242	260	3	and	and	CCONJ
easat-3242	260	4	loss	loss	NOUN
easat-3242	260	5	behaviour	behaviour	NOUN
easat-3242	260	6	can	can	AUX
easat-3242	260	7	be	be	AUX
easat-3242	260	8	seen	see	VERB
easat-3242	260	9	in	in	ADP
easat-3242	260	10	the	the	DET
easat-3242	260	11	above	above	ADJ
easat-3242	260	12	bar	bar	NOUN
easat-3242	260	13	plot	plot	NOUN
easat-3242	260	14	,	,	PUNCT
easat-3242	260	15	with	with	ADP
easat-3242	260	16	validation	validation	NOUN
easat-3242	260	17	accuracy	accuracy	NOUN
easat-3242	260	18	following	follow	VERB
easat-3242	260	19	a	a	DET
easat-3242	260	20	slightly	slightly	ADV
easat-3242	260	21	smaller	small	ADJ
easat-3242	260	22	upward	upward	ADJ
easat-3242	260	23	trend	trend	NOUN
easat-3242	260	24	to	to	PART
easat-3242	260	25	again	again	ADV
easat-3242	260	26	settle	settle	VERB
easat-3242	260	27	around	around	ADP
easat-3242	260	28	the	the	DET
easat-3242	260	29	same	same	ADJ
easat-3242	260	30	point	point	NOUN
easat-3242	260	31	as	as	ADP
easat-3242	260	32	training	training	NOUN
easat-3242	260	33	.	.	PUNCT
easat-3242	261	1	validation	validation	NOUN
easat-3242	261	2	accuracy	accuracy	NOUN
easat-3242	261	3	is	be	AUX
easat-3242	261	4	close	close	ADJ
easat-3242	261	5	to	to	ADP
easat-3242	261	6	training	training	NOUN
easat-3242	261	7	accuracy	accuracy	NOUN
easat-3242	261	8	confirms	confirm	VERB
easat-3242	261	9	that	that	SCONJ
easat-3242	261	10	the	the	DET
easat-3242	261	11	model	model	NOUN
easat-3242	261	12	has	have	AUX
easat-3242	261	13	not	not	PART
easat-3242	261	14	been	be	AUX
easat-3242	261	15	overfitted	overfitte	VERB
easat-3242	261	16	and	and	CCONJ
easat-3242	261	17	it	it	PRON
easat-3242	261	18	will	will	AUX
easat-3242	261	19	generalize	generalize	VERB
easat-3242	261	20	well	well	ADV
easat-3242	261	21	with	with	ADP
easat-3242	261	22	out	out	ADP
easat-3242	261	23	of	of	ADP
easat-3242	261	24	sample	sample	NOUN
easat-3242	261	25	data	datum	NOUN
easat-3242	261	26	from	from	ADP
easat-3242	261	27	unseen	unseen	ADJ
easat-3242	261	28	dataset	dataset	NOUN
easat-3242	261	29	.	.	PUNCT
easat-3242	262	1	nevertheless	nevertheless	ADV
easat-3242	262	2	,	,	PUNCT
easat-3242	262	3	oscillations	oscillation	NOUN
easat-3242	262	4	around	around	ADP
easat-3242	262	5	steady	steady	ADJ
easat-3242	262	6	validation	validation	NOUN
easat-3242	262	7	accuracy	accuracy	NOUN
easat-3242	262	8	are	be	AUX
easat-3242	262	9	expected	expect	VERB
easat-3242	262	10	(	(	PUNCT
easat-3242	262	11	especially	especially	ADV
easat-3242	262	12	in	in	ADP
easat-3242	262	13	early	early	ADJ
easat-3242	262	14	epochs	epoch	NOUN
easat-3242	262	15	)	)	PUNCT
easat-3242	262	16	because	because	SCONJ
easat-3242	262	17	the	the	DET
easat-3242	262	18	model	model	NOUN
easat-3242	262	19	is	be	AUX
easat-3242	262	20	trained	train	VERB
easat-3242	262	21	on	on	ADP
easat-3242	262	22	one	one	NUM
easat-3242	262	23	set	set	NOUN
easat-3242	262	24	of	of	ADP
easat-3242	262	25	data	datum	NOUN
easat-3242	262	26	and	and	CCONJ
easat-3242	262	27	periodically	periodically	ADV
easat-3242	262	28	validated	validate	VERB
easat-3242	262	29	against	against	ADP
easat-3242	262	30	another	another	PRON
easat-3242	262	31	.	.	PUNCT
easat-3242	263	1	5720	5720	NUM
easat-3242	263	2	edelweiss	edelweiss	PROPN
easat-3242	263	3	applied	apply	VERB
easat-3242	263	4	science	science	NOUN
easat-3242	263	5	and	and	CCONJ
easat-3242	263	6	technology	technology	NOUN
easat-3242	263	7	issn	issn	PROPN
easat-3242	263	8	:	:	PUNCT
easat-3242	263	9	2576	2576	NUM
easat-3242	263	10	-	-	SYM
easat-3242	263	11	8484	8484	NUM
easat-3242	263	12	vol	vol	NOUN
easat-3242	263	13	.	.	PROPN
easat-3242	263	14	8	8	NUM
easat-3242	263	15	,	,	PUNCT
easat-3242	263	16	no	no	INTJ
easat-3242	263	17	.	.	NOUN
easat-3242	264	1	6	6	NUM
easat-3242	264	2	:	:	SYM
easat-3242	264	3	5707	5707	NUM
easat-3242	264	4	-	-	SYM
easat-3242	264	5	5722	5722	NUM
easat-3242	264	6	,	,	PUNCT
easat-3242	264	7	2024	2024	NUM
easat-3242	264	8	doi	doi	NOUN
easat-3242	264	9	:	:	PUNCT
easat-3242	264	10	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	264	11	©	©	PROPN
easat-3242	264	12	2024	2024	NUM
easat-3242	264	13	by	by	ADP
easat-3242	264	14	the	the	DET
easat-3242	264	15	authors	author	NOUN
easat-3242	264	16	;	;	PUNCT
easat-3242	264	17	licensee	licensee	PROPN
easat-3242	264	18	learning	learning	NOUN
easat-3242	264	19	gate	gate	PROPN
easat-3242	264	20	input	input	NOUN
easat-3242	264	21	image	image	NOUN
easat-3242	264	22	ground	ground	NOUN
easat-3242	264	23	truth	truth	NOUN
easat-3242	264	24	image	image	NOUN
easat-3242	264	25	result	result	VERB
easat-3242	264	26	image	image	NOUN
easat-3242	264	27	(	(	PUNCT
easat-3242	264	28	a	a	NOUN
easat-3242	264	29	)	)	PUNCT
easat-3242	264	30	(	(	PUNCT
easat-3242	264	31	b	b	X
easat-3242	264	32	)	)	PUNCT
easat-3242	264	33	(	(	PUNCT
easat-3242	264	34	c	c	X
easat-3242	264	35	)	)	PUNCT
easat-3242	264	36	(	(	PUNCT
easat-3242	264	37	d	d	X
easat-3242	264	38	)	)	PUNCT
easat-3242	264	39	figure	figure	NOUN
easat-3242	264	40	6	6	NUM
easat-3242	264	41	.	.	PUNCT
easat-3242	265	1	output	output	NOUN
easat-3242	265	2	segmented	segment	VERB
easat-3242	265	3	images	image	NOUN
easat-3242	265	4	.	.	PUNCT
easat-3242	266	1	5721	5721	NUM
easat-3242	266	2	edelweiss	edelweiss	PROPN
easat-3242	266	3	applied	apply	VERB
easat-3242	266	4	science	science	NOUN
easat-3242	266	5	and	and	CCONJ
easat-3242	266	6	technology	technology	NOUN
easat-3242	266	7	issn	issn	PROPN
easat-3242	266	8	:	:	PUNCT
easat-3242	266	9	2576	2576	NUM
easat-3242	266	10	-	-	SYM
easat-3242	266	11	8484	8484	NUM
easat-3242	266	12	vol	vol	NOUN
easat-3242	266	13	.	.	PROPN
easat-3242	266	14	8	8	NUM
easat-3242	266	15	,	,	PUNCT
easat-3242	266	16	no	no	INTJ
easat-3242	266	17	.	.	NOUN
easat-3242	266	18	6	6	NUM
easat-3242	266	19	:	:	SYM
easat-3242	266	20	5707	5707	NUM
easat-3242	266	21	-	-	SYM
easat-3242	266	22	5722	5722	NUM
easat-3242	266	23	,	,	PUNCT
easat-3242	266	24	2024	2024	NUM
easat-3242	266	25	doi	doi	NOUN
easat-3242	266	26	:	:	PUNCT
easat-3242	266	27	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	266	28	©	©	PROPN
easat-3242	266	29	2024	2024	NUM
easat-3242	266	30	by	by	ADP
easat-3242	266	31	the	the	DET
easat-3242	266	32	authors	author	NOUN
easat-3242	266	33	;	;	PUNCT
easat-3242	266	34	licensee	licensee	PROPN
easat-3242	266	35	learning	learn	VERB
easat-3242	266	36	gate	gate	NOUN
easat-3242	266	37	table	table	NOUN
easat-3242	266	38	1	1	NUM
easat-3242	266	39	.	.	PUNCT
easat-3242	266	40	evaluation	evaluation	NOUN
easat-3242	266	41	metrics	metric	NOUN
easat-3242	266	42	.	.	PUNCT
easat-3242	267	1	precision	precision	NOUN
easat-3242	267	2	recall	recall	NOUN
easat-3242	267	3	f1	f1	PROPN
easat-3242	267	4	score	score	NOUN
easat-3242	267	5	image	image	NOUN
easat-3242	267	6	1	1	NUM
easat-3242	267	7	0.835	0.835	NUM
easat-3242	267	8	0.830	0.830	NUM
easat-3242	267	9	0.832	0.832	NUM
easat-3242	267	10	image	image	NOUN
easat-3242	267	11	2	2	NUM
easat-3242	267	12	0.902	0.902	NUM
easat-3242	267	13	0.902	0.902	NUM
easat-3242	267	14	0.902	0.902	NUM
easat-3242	267	15	image	image	NOUN
easat-3242	267	16	3	3	NUM
easat-3242	267	17	0.776	0.776	NUM
easat-3242	267	18	0.787	0.787	NUM
easat-3242	267	19	0.782	0.782	NUM
easat-3242	267	20	image	image	NOUN
easat-3242	267	21	4	4	NUM
easat-3242	267	22	0.747	0.747	NUM
easat-3242	267	23	0.752	0.752	NUM
easat-3242	267	24	0.749	0.749	NUM
easat-3242	267	25	table	table	NOUN
easat-3242	267	26	1	1	NUM
easat-3242	267	27	gives	give	VERB
easat-3242	267	28	the	the	DET
easat-3242	267	29	precision	precision	NOUN
easat-3242	267	30	,	,	PUNCT
easat-3242	267	31	recall	recall	NOUN
easat-3242	267	32	,	,	PUNCT
easat-3242	267	33	and	and	CCONJ
easat-3242	267	34	f1	f1	NOUN
easat-3242	267	35	score	score	NOUN
easat-3242	267	36	for	for	ADP
easat-3242	267	37	four	four	NUM
easat-3242	267	38	images	image	NOUN
easat-3242	267	39	as	as	ADV
easat-3242	267	40	well	well	ADV
easat-3242	267	41	to	to	PART
easat-3242	267	42	have	have	VERB
easat-3242	267	43	an	an	DET
easat-3242	267	44	all	all	NOUN
easat-3242	267	45	in	in	ADP
easat-3242	267	46	one	one	NUM
easat-3242	267	47	understanding	understanding	NOUN
easat-3242	267	48	of	of	ADP
easat-3242	267	49	how	how	SCONJ
easat-3242	267	50	good	good	ADJ
easat-3242	267	51	was	be	AUX
easat-3242	267	52	a	a	DET
easat-3242	267	53	model	model	NOUN
easat-3242	267	54	on	on	ADP
easat-3242	267	55	these	these	DET
easat-3242	267	56	metrics	metric	NOUN
easat-3242	267	57	.	.	PUNCT
easat-3242	268	1	precision	precision	NOUN
easat-3242	268	2	is	be	AUX
easat-3242	268	3	the	the	DET
easat-3242	268	4	number	number	NOUN
easat-3242	268	5	of	of	ADP
easat-3242	268	6	actual	actual	ADJ
easat-3242	268	7	positives	positive	NOUN
easat-3242	268	8	from	from	ADP
easat-3242	268	9	all	all	PRON
easat-3242	268	10	predict	predict	VERB
easat-3242	268	11	positive	positive	ADJ
easat-3242	268	12	instances	instance	NOUN
easat-3242	268	13	,	,	PUNCT
easat-3242	268	14	whereas	whereas	SCONJ
easat-3242	268	15	recall	recall	NOUN
easat-3242	268	16	is	be	AUX
easat-3242	268	17	how	how	SCONJ
easat-3242	268	18	many	many	ADJ
easat-3242	268	19	true	true	ADJ
easat-3242	268	20	positives	positive	NOUN
easat-3242	268	21	out	out	ADV
easat-3242	268	22	there	there	ADV
easat-3242	268	23	.	.	PUNCT
easat-3242	269	1	the	the	DET
easat-3242	269	2	f1	f1	PROPN
easat-3242	269	3	metric	metric	NOUN
easat-3242	269	4	is	be	AUX
easat-3242	269	5	the	the	DET
easat-3242	269	6	harmonic	harmonic	ADJ
easat-3242	269	7	mean	mean	NOUN
easat-3242	269	8	of	of	ADP
easat-3242	269	9	precision	precision	NOUN
easat-3242	269	10	and	and	CCONJ
easat-3242	269	11	recall	recall	NOUN
easat-3242	269	12	giving	give	VERB
easat-3242	269	13	a	a	DET
easat-3242	269	14	performance	performance	NOUN
easat-3242	269	15	balance	balance	NOUN
easat-3242	269	16	between	between	ADP
easat-3242	269	17	various	various	ADJ
easat-3242	269	18	evaluation	evaluation	NOUN
easat-3242	269	19	parameters	parameter	NOUN
easat-3242	269	20	.	.	PUNCT
easat-3242	270	1	precision	precision	NOUN
easat-3242	270	2	,	,	PUNCT
easat-3242	270	3	recall	recall	NOUN
easat-3242	270	4	and	and	CCONJ
easat-3242	270	5	f1	f1	ADJ
easat-3242	270	6	score	score	NOUN
easat-3242	270	7	for	for	ADP
easat-3242	270	8	image	image	NOUN
easat-3242	270	9	1	1	NUM
easat-3242	270	10	are	be	AUX
easat-3242	270	11	close	close	ADJ
easat-3242	270	12	to	to	ADP
easat-3242	270	13	each	each	DET
easat-3242	270	14	other	other	ADJ
easat-3242	270	15	which	which	PRON
easat-3242	270	16	indicate	indicate	VERB
easat-3242	270	17	the	the	DET
easat-3242	270	18	model	model	NOUN
easat-3242	270	19	has	have	AUX
easat-3242	270	20	detected	detect	VERB
easat-3242	270	21	relevant	relevant	ADJ
easat-3242	270	22	instances	instance	NOUN
easat-3242	270	23	consistently	consistently	ADV
easat-3242	270	24	.	.	PUNCT
easat-3242	271	1	image	image	NOUN
easat-3242	271	2	2	2	NUM
easat-3242	271	3	displays	display	VERB
easat-3242	271	4	best	good	ADJ
easat-3242	271	5	scores	score	NOUN
easat-3242	271	6	in	in	ADP
easat-3242	271	7	each	each	DET
easat-3242	271	8	metric	metric	NOUN
easat-3242	271	9	,	,	PUNCT
easat-3242	271	10	meaning	mean	VERB
easat-3242	271	11	that	that	SCONJ
easat-3242	271	12	the	the	DET
easat-3242	271	13	model	model	NOUN
easat-3242	271	14	predicts	predict	VERB
easat-3242	271	15	this	this	DET
easat-3242	271	16	image	image	NOUN
easat-3242	271	17	accurately	accurately	ADV
easat-3242	271	18	.	.	PUNCT
easat-3242	272	1	images	image	NOUN
easat-3242	272	2	3	3	NUM
easat-3242	272	3	and	and	CCONJ
easat-3242	272	4	4	4	NUM
easat-3242	272	5	have	have	VERB
easat-3242	272	6	slightly	slightly	ADV
easat-3242	272	7	lower	low	ADJ
easat-3242	272	8	values	value	NOUN
easat-3242	272	9	,	,	PUNCT
easat-3242	272	10	in	in	ADP
easat-3242	272	11	particular	particular	ADJ
easat-3242	272	12	image	image	NOUN
easat-3242	272	13	3	3	NUM
easat-3242	272	14	with	with	ADP
easat-3242	272	15	a	a	DET
easat-3242	272	16	recall	recall	NOUN
easat-3242	272	17	higher	high	ADJ
easat-3242	272	18	than	than	ADP
easat-3242	272	19	its	its	PRON
easat-3242	272	20	precision	precision	NOUN
easat-3242	272	21	indicating	indicate	VERB
easat-3242	272	22	the	the	DET
easat-3242	272	23	model	model	NOUN
easat-3242	272	24	may	may	AUX
easat-3242	272	25	be	be	AUX
easat-3242	272	26	better	well	ADJ
easat-3242	272	27	at	at	ADP
easat-3242	272	28	finding	find	VERB
easat-3242	272	29	true	true	ADJ
easat-3242	272	30	positives	positive	NOUN
easat-3242	272	31	but	but	CCONJ
easat-3242	272	32	is	be	AUX
easat-3242	272	33	less	less	ADV
easat-3242	272	34	successful	successful	ADJ
easat-3242	272	35	avoiding	avoid	VERB
easat-3242	272	36	false	false	ADJ
easat-3242	272	37	positives	positive	NOUN
easat-3242	272	38	.	.	PUNCT
easat-3242	273	1	table	table	NOUN
easat-3242	273	2	2	2	NUM
easat-3242	273	3	.	.	PUNCT
easat-3242	273	4	comparative	comparative	ADJ
easat-3242	273	5	analysis	analysis	NOUN
easat-3242	273	6	.	.	PUNCT
easat-3242	274	1	method	method	NOUN
easat-3242	274	2	accuracy	accuracy	NOUN
easat-3242	274	3	unet++	unet++	PROPN
easat-3242	274	4	with	with	ADP
easat-3242	274	5	efficientnet	efficientnet	NOUN
easat-3242	274	6	-	-	PUNCT
easat-3242	274	7	b0	b0	NOUN
easat-3242	274	8	0.83	0.83	NUM
easat-3242	274	9	unet++	unet++	NOUN
easat-3242	274	10	with	with	ADP
easat-3242	274	11	mobilenet_s0	mobilenet_s0	NOUN
easat-3242	274	12	0.79	0.79	NUM
easat-3242	274	13	unet++	unet++	NOUN
easat-3242	274	14	with	with	ADP
easat-3242	274	15	mobilenet_v2	mobilenet_v2	NOUN
easat-3242	274	16	0.80	0.80	NUM
easat-3242	274	17	unet++	unet++	NOUN
easat-3242	274	18	with	with	ADP
easat-3242	274	19	vgg19	vgg19	PROPN
easat-3242	274	20	0.84	0.84	NUM
easat-3242	274	21	accuracy	accuracy	NOUN
easat-3242	274	22	of	of	ADP
easat-3242	274	23	unet++	unet++	PROPN
easat-3242	274	24	architectures	architecture	NOUN
easat-3242	274	25	coupled	couple	VERB
easat-3242	274	26	with	with	ADP
easat-3242	274	27	different	different	ADJ
easat-3242	274	28	backbone	backbone	NOUN
easat-3242	274	29	models	model	NOUN
easat-3242	274	30	are	be	AUX
easat-3242	274	31	summarized	summarize	VERB
easat-3242	274	32	in	in	ADP
easat-3242	274	33	table	table	NOUN
easat-3242	274	34	2	2	NUM
easat-3242	274	35	.	.	PUNCT
easat-3242	274	36	result	result	NOUN
easat-3242	274	37	shows	show	VERB
easat-3242	274	38	that	that	SCONJ
easat-3242	274	39	the	the	DET
easat-3242	274	40	combination	combination	NOUN
easat-3242	274	41	unet++	unet++	PRON
easat-3242	274	42	and	and	CCONJ
easat-3242	274	43	vgg19	vgg19	PROPN
easat-3242	274	44	gives	give	VERB
easat-3242	274	45	an	an	DET
easat-3242	274	46	accuracy	accuracy	NOUN
easat-3242	274	47	of	of	ADP
easat-3242	274	48	0.84	0.84	NUM
easat-3242	274	49	here	here	ADV
easat-3242	274	50	which	which	PRON
easat-3242	274	51	is	be	AUX
easat-3242	274	52	better	well	ADJ
easat-3242	274	53	than	than	SCONJ
easat-3242	274	54	other	other	ADJ
easat-3242	274	55	comparison	comparison	NOUN
easat-3242	274	56	made	make	VERB
easat-3242	274	57	above	above	ADP
easat-3242	274	58	unet++	unet++	NOUN
easat-3242	274	59	with	with	ADP
easat-3242	274	60	efficientnet	efficientnet	NOUN
easat-3242	274	61	-	-	PUNCT
easat-3242	274	62	b0	b0	NOUN
easat-3242	274	63	came	come	VERB
easat-3242	274	64	in	in	ADP
easat-3242	274	65	next	next	ADV
easat-3242	274	66	at	at	ADP
easat-3242	274	67	an	an	DET
easat-3242	274	68	accuracy	accuracy	NOUN
easat-3242	274	69	of	of	ADP
easat-3242	274	70	83	83	NUM
easat-3242	274	71	%	%	NOUN
easat-3242	274	72	suggesting	suggest	VERB
easat-3242	274	73	similarly	similarly	ADV
easat-3242	274	74	effective	effective	ADJ
easat-3242	274	75	performance	performance	NOUN
easat-3242	274	76	across	across	ADP
easat-3242	274	77	the	the	DET
easat-3242	274	78	dataset	dataset	NOUN
easat-3242	274	79	.	.	PUNCT
easat-3242	275	1	on	on	ADP
easat-3242	275	2	the	the	DET
easat-3242	275	3	other	other	ADJ
easat-3242	275	4	hand	hand	NOUN
easat-3242	275	5	,	,	PUNCT
easat-3242	275	6	unet++	unet++	PROPN
easat-3242	275	7	combined	combine	VERB
easat-3242	275	8	with	with	ADP
easat-3242	275	9	mobilenet_v2	mobilenet_v2	NOUN
easat-3242	275	10	yielded	yield	VERB
easat-3242	275	11	a	a	DET
easat-3242	275	12	slightly	slightly	ADV
easat-3242	275	13	lower	low	ADJ
easat-3242	275	14	accuracy	accuracy	NOUN
easat-3242	275	15	of	of	ADP
easat-3242	275	16	0.80	0.80	NUM
easat-3242	275	17	and	and	CCONJ
easat-3242	275	18	finally	finally	ADV
easat-3242	275	19	is	be	AUX
easat-3242	275	20	mobilenett_s0	mobilenett_s0	NOUN
easat-3242	275	21	showing	show	VERB
easat-3242	275	22	even	even	ADV
easat-3242	275	23	less	less	ADJ
easat-3242	275	24	accuracy	accuracy	NOUN
easat-3242	275	25	i.e.	i.e.	X
easat-3242	275	26	,	,	PUNCT
easat-3242	275	27	0.79	0.79	NUM
easat-3242	275	28	.	.	PUNCT
easat-3242	276	1	this	this	DET
easat-3242	276	2	comparison	comparison	NOUN
easat-3242	276	3	examine	examine	VERB
easat-3242	276	4	that	that	SCONJ
easat-3242	276	5	,	,	PUNCT
easat-3242	276	6	the	the	DET
easat-3242	276	7	segmentation	segmentation	NOUN
easat-3242	276	8	performance	performance	NOUN
easat-3242	276	9	for	for	ADP
easat-3242	276	10	vgg19	vgg19	PROPN
easat-3242	276	11	is	be	AUX
easat-3242	276	12	best	good	ADJ
easat-3242	276	13	among	among	ADP
easat-3242	276	14	other	other	ADJ
easat-3242	276	15	tested	test	VERB
easat-3242	276	16	backbon	backbon	NOUN
easat-3242	276	17	network	network	NOUN
easat-3242	276	18	and	and	CCONJ
easat-3242	276	19	choice	choice	NOUN
easat-3242	276	20	of	of	ADP
easat-3242	276	21	backbone	backbone	NOUN
easat-3242	276	22	network	network	NOUN
easat-3242	276	23	impacts	impact	VERB
easat-3242	276	24	segmentation	segmentation	NOUN
easat-3242	276	25	performance	performance	NOUN
easat-3242	276	26	.	.	PUNCT
easat-3242	277	1	5	5	X
easat-3242	277	2	.	.	X
easat-3242	277	3	conclusion	conclusion	NOUN
easat-3242	277	4	the	the	DET
easat-3242	277	5	experimental	experimental	ADJ
easat-3242	277	6	results	result	NOUN
easat-3242	277	7	demonstrated	demonstrate	VERB
easat-3242	277	8	the	the	DET
easat-3242	277	9	effectively	effectively	ADJ
easat-3242	277	10	of	of	ADP
easat-3242	277	11	the	the	DET
easat-3242	277	12	proposed	propose	VERB
easat-3242	277	13	unet++	unet++	NOUN
easat-3242	277	14	with	with	ADP
easat-3242	277	15	vgg19	vgg19	PRON
easat-3242	277	16	for	for	ADP
easat-3242	277	17	segmenting	segment	VERB
easat-3242	277	18	satellite	satellite	NOUN
easat-3242	277	19	images	image	NOUN
easat-3242	277	20	.	.	PUNCT
easat-3242	278	1	when	when	SCONJ
easat-3242	278	2	applied	apply	VERB
easat-3242	278	3	to	to	ADP
easat-3242	278	4	a	a	DET
easat-3242	278	5	large	large	ADJ
easat-3242	278	6	kaggle	kaggle	NOUN
easat-3242	278	7	dataset	dataset	NOUN
easat-3242	278	8	with	with	ADP
easat-3242	278	9	varying	vary	VERB
easat-3242	278	10	images	image	NOUN
easat-3242	278	11	,	,	PUNCT
easat-3242	278	12	the	the	DET
easat-3242	278	13	provided	provide	VERB
easat-3242	278	14	model	model	NOUN
easat-3242	278	15	exhibited	exhibit	VERB
easat-3242	278	16	high	high	ADJ
easat-3242	278	17	precision	precision	NOUN
easat-3242	278	18	,	,	PUNCT
easat-3242	278	19	recall	recall	NOUN
easat-3242	278	20	,	,	PUNCT
easat-3242	278	21	and	and	CCONJ
easat-3242	278	22	f1	f1	NOUN
easat-3242	278	23	metrics	metric	NOUN
easat-3242	278	24	consistently	consistently	ADV
easat-3242	278	25	across	across	ADP
easat-3242	278	26	different	different	ADJ
easat-3242	278	27	images	image	NOUN
easat-3242	278	28	.	.	PUNCT
easat-3242	279	1	this	this	PRON
easat-3242	279	2	corroborates	corroborate	VERB
easat-3242	279	3	that	that	SCONJ
easat-3242	279	4	the	the	DET
easat-3242	279	5	designed	design	VERB
easat-3242	279	6	system	system	NOUN
easat-3242	279	7	is	be	AUX
easat-3242	279	8	indeed	indeed	ADV
easat-3242	279	9	durable	durable	ADJ
easat-3242	279	10	enough	enough	ADV
easat-3242	279	11	to	to	PART
easat-3242	279	12	differentiate	differentiate	VERB
easat-3242	279	13	and	and	CCONJ
easat-3242	279	14	partition	partition	NOUN
easat-3242	279	15	land	land	NOUN
easat-3242	279	16	cover	cover	NOUN
easat-3242	279	17	patterns	pattern	NOUN
easat-3242	279	18	.	.	PUNCT
easat-3242	280	1	the	the	DET
easat-3242	280	2	model	model	NOUN
easat-3242	280	3	has	have	AUX
easat-3242	280	4	specifically	specifically	ADV
easat-3242	280	5	given	give	VERB
easat-3242	280	6	very	very	ADV
easat-3242	280	7	good	good	ADJ
easat-3242	280	8	results	result	NOUN
easat-3242	280	9	when	when	SCONJ
easat-3242	280	10	compared	compare	VERB
easat-3242	280	11	to	to	ADP
easat-3242	280	12	other	other	ADJ
easat-3242	280	13	configurations	configuration	NOUN
easat-3242	280	14	such	such	ADJ
easat-3242	280	15	as	as	ADP
easat-3242	280	16	unet++	unet++	NOUN
easat-3242	280	17	with	with	ADP
easat-3242	280	18	efficientnet	efficientnet	NOUN
easat-3242	280	19	-	-	PUNCT
easat-3242	280	20	b0	b0	NOUN
easat-3242	280	21	,	,	PUNCT
easat-3242	280	22	mobilenet_s0	mobilenet_s0	NOUN
easat-3242	280	23	and	and	CCONJ
easat-3242	280	24	mobilenet_v2	mobilenet_v2	NOUN
easat-3242	280	25	having	have	VERB
easat-3242	280	26	highest	high	ADJ
easat-3242	280	27	accuracy	accuracy	NOUN
easat-3242	280	28	of	of	ADP
easat-3242	280	29	0.84	0.84	NUM
easat-3242	280	30	.	.	PUNCT
easat-3242	281	1	the	the	DET
easat-3242	281	2	increased	increase	VERB
easat-3242	281	3	accuracy	accuracy	NOUN
easat-3242	281	4	might	might	AUX
easat-3242	281	5	be	be	AUX
easat-3242	281	6	notch	notch	ADJ
easat-3242	281	7	due	due	ADP
easat-3242	281	8	to	to	ADP
easat-3242	281	9	the	the	DET
easat-3242	281	10	higher	high	ADJ
easat-3242	281	11	design	design	NOUN
easat-3242	281	12	of	of	ADP
easat-3242	281	13	vgg19	vgg19	PRON
easat-3242	281	14	which	which	PRON
easat-3242	281	15	had	have	VERB
easat-3242	281	16	a	a	DET
easat-3242	281	17	very	very	ADV
easat-3242	281	18	deep	deep	ADJ
easat-3242	281	19	and	and	CCONJ
easat-3242	281	20	complicated	complicated	ADJ
easat-3242	281	21	architecture	architecture	NOUN
easat-3242	281	22	hence	hence	ADV
easat-3242	281	23	an	an	DET
easat-3242	281	24	accurate	accurate	ADJ
easat-3242	281	25	model	model	NOUN
easat-3242	281	26	may	may	AUX
easat-3242	281	27	have	have	AUX
easat-3242	281	28	been	be	AUX
easat-3242	281	29	build	build	VERB
easat-3242	281	30	in	in	ADP
easat-3242	281	31	comparison	comparison	NOUN
easat-3242	281	32	with	with	ADP
easat-3242	281	33	other	other	ADJ
easat-3242	281	34	sequences	sequence	NOUN
easat-3242	281	35	investigated	investigate	VERB
easat-3242	281	36	.	.	PUNCT
easat-3242	282	1	copyright	copyright	NOUN
easat-3242	282	2	:	:	PUNCT
easat-3242	282	3	©	©	PROPN
easat-3242	282	4	2024	2024	NUM
easat-3242	282	5	by	by	ADP
easat-3242	282	6	the	the	DET
easat-3242	282	7	authors	author	NOUN
easat-3242	282	8	.	.	PUNCT
easat-3242	283	1	this	this	DET
easat-3242	283	2	article	article	NOUN
easat-3242	283	3	is	be	AUX
easat-3242	283	4	an	an	DET
easat-3242	283	5	open	open	ADJ
easat-3242	283	6	access	access	NOUN
easat-3242	283	7	article	article	NOUN
easat-3242	283	8	distributed	distribute	VERB
easat-3242	283	9	under	under	ADP
easat-3242	283	10	the	the	DET
easat-3242	283	11	terms	term	NOUN
easat-3242	283	12	and	and	CCONJ
easat-3242	283	13	conditions	condition	NOUN
easat-3242	283	14	of	of	ADP
easat-3242	283	15	the	the	DET
easat-3242	283	16	creative	creative	ADJ
easat-3242	283	17	commons	common	NOUN
easat-3242	283	18	attribution	attribution	NOUN
easat-3242	283	19	(	(	PUNCT
easat-3242	283	20	cc	cc	NOUN
easat-3242	283	21	by	by	ADP
easat-3242	283	22	)	)	PUNCT
easat-3242	283	23	license	license	NOUN
easat-3242	283	24	(	(	PUNCT
easat-3242	283	25	https://creativecommons.org/licenses/by/4.0/	https://creativecommons.org/licenses/by/4.0/	PROPN
easat-3242	283	26	)	)	PUNCT
easat-3242	283	27	.	.	PUNCT
easat-3242	284	1	https://creativecommons.org/licenses/by/4.0/	https://creativecommons.org/licenses/by/4.0/	PROPN
easat-3242	284	2	5722	5722	NUM
easat-3242	284	3	edelweiss	edelweiss	PROPN
easat-3242	284	4	applied	apply	VERB
easat-3242	284	5	science	science	NOUN
easat-3242	284	6	and	and	CCONJ
easat-3242	284	7	technology	technology	NOUN
easat-3242	284	8	issn	issn	PROPN
easat-3242	284	9	:	:	PUNCT
easat-3242	284	10	2576	2576	NUM
easat-3242	284	11	-	-	SYM
easat-3242	284	12	8484	8484	NUM
easat-3242	284	13	vol	vol	NOUN
easat-3242	284	14	.	.	PROPN
easat-3242	284	15	8	8	NUM
easat-3242	284	16	,	,	PUNCT
easat-3242	284	17	no	no	INTJ
easat-3242	284	18	.	.	NOUN
easat-3242	284	19	6	6	NUM
easat-3242	284	20	:	:	SYM
easat-3242	284	21	5707	5707	NUM
easat-3242	284	22	-	-	SYM
easat-3242	284	23	5722	5722	NUM
easat-3242	284	24	,	,	PUNCT
easat-3242	284	25	2024	2024	NUM
easat-3242	284	26	doi	doi	NOUN
easat-3242	284	27	:	:	PUNCT
easat-3242	284	28	10.55214/25768484.v8i6.3242	10.55214/25768484.v8i6.3242	NUM
easat-3242	284	29	©	©	PROPN
easat-3242	284	30	2024	2024	NUM
easat-3242	284	31	by	by	ADP
easat-3242	284	32	the	the	DET
easat-3242	284	33	authors	author	NOUN
easat-3242	284	34	;	;	PUNCT
easat-3242	284	35	licensee	licensee	PROPN
easat-3242	284	36	learning	learn	VERB
easat-3242	284	37	gate	gate	NOUN
easat-3242	284	38	references	reference	NOUN
easat-3242	284	39	[	[	X
easat-3242	284	40	1	1	NUM
easat-3242	284	41	]	]	PUNCT
easat-3242	284	42	sertel	sertel	NOUN
easat-3242	284	43	,	,	PUNCT
easat-3242	284	44	elif	elif	NOUN
easat-3242	284	45	,	,	PUNCT
easat-3242	284	46	burakekim	burakekim	PROPN
easat-3242	284	47	,	,	PUNCT
easat-3242	284	48	pariaettehadiosgouei	pariaettehadiosgouei	ADV
easat-3242	284	49	,	,	PUNCT
easat-3242	284	50	and	and	CCONJ
easat-3242	284	51	m.	m.	NOUN
easat-3242	284	52	erdemkabadayi	erdemkabadayi	NOUN
easat-3242	284	53	.	.	PUNCT
easat-3242	285	1	"	"	PUNCT
easat-3242	285	2	land	land	NOUN
easat-3242	285	3	use	use	NOUN
easat-3242	285	4	and	and	CCONJ
easat-3242	285	5	land	land	NOUN
easat-3242	285	6	cover	cover	NOUN
easat-3242	285	7	mapping	mapping	NOUN
easat-3242	285	8	using	use	VERB
easat-3242	285	9	deep	deep	ADJ
easat-3242	285	10	learning	learning	NOUN
easat-3242	285	11	based	base	VERB
easat-3242	285	12	segmentation	segmentation	NOUN
easat-3242	285	13	approaches	approach	NOUN
easat-3242	285	14	and	and	CCONJ
easat-3242	285	15	vhr	vhr	VERB
easat-3242	285	16	worldview-3	worldview-3	DET
easat-3242	285	17	images	image	NOUN
easat-3242	285	18	.	.	PUNCT
easat-3242	285	19	"	"	PUNCT
easat-3242	286	1	remote	remote	ADJ
easat-3242	286	2	sensing	sense	VERB
easat-3242	286	3	14	14	NUM
easat-3242	286	4	,	,	PUNCT
easat-3242	286	5	no	no	INTJ
easat-3242	286	6	.	.	NOUN
easat-3242	286	7	18	18	NUM
easat-3242	286	8	(	(	PUNCT
easat-3242	286	9	2022	2022	NUM
easat-3242	286	10	):	):	PUNCT
easat-3242	286	11	4558	4558	NUM
easat-3242	286	12	.	.	PUNCT
easat-3242	287	1	[	[	X
easat-3242	287	2	2	2	NUM
easat-3242	287	3	]	]	X
easat-3242	287	4	kotaridis	kotaridis	PROPN
easat-3242	287	5	,	,	PUNCT
easat-3242	287	6	ioannis	ioanni	NOUN
easat-3242	287	7	,	,	PUNCT
easat-3242	287	8	and	and	CCONJ
easat-3242	287	9	maria	maria	PROPN
easat-3242	287	10	lazaridou	lazaridou	NOUN
easat-3242	287	11	.	.	PUNCT
easat-3242	288	1	"	"	PUNCT
easat-3242	288	2	remote	remote	ADJ
easat-3242	288	3	sensing	sense	VERB
easat-3242	288	4	image	image	NOUN
easat-3242	288	5	segmentation	segmentation	NOUN
easat-3242	288	6	advances	advance	NOUN
easat-3242	288	7	:	:	PUNCT
easat-3242	288	8	a	a	DET
easat-3242	288	9	meta	meta	ADJ
easat-3242	288	10	-	-	PUNCT
easat-3242	288	11	analysis	analysis	NOUN
easat-3242	288	12	.	.	PUNCT
easat-3242	288	13	"	"	PUNCT
easat-3242	288	14	isprs	isprs	PROPN
easat-3242	288	15	journal	journal	NOUN
easat-3242	288	16	of	of	ADP
easat-3242	288	17	photogrammetry	photogrammetry	NOUN
easat-3242	288	18	and	and	CCONJ
easat-3242	288	19	remote	remote	ADJ
easat-3242	288	20	sensing	sense	VERB
easat-3242	288	21	173	173	NUM
easat-3242	288	22	(	(	PUNCT
easat-3242	288	23	2021	2021	NUM
easat-3242	288	24	):	):	PUNCT
easat-3242	288	25	309	309	NUM
easat-3242	288	26	-	-	SYM
easat-3242	288	27	322	322	NUM
easat-3242	288	28	.	.	PUNCT
easat-3242	289	1	[	[	X
easat-3242	289	2	3	3	NUM
easat-3242	289	3	]	]	X
easat-3242	289	4	ayala	ayala	PROPN
easat-3242	289	5	,	,	PUNCT
easat-3242	289	6	christian	christian	PROPN
easat-3242	289	7	,	,	PUNCT
easat-3242	289	8	rubén	rubén	ADJ
easat-3242	289	9	sesma	sesma	NOUN
easat-3242	289	10	,	,	PUNCT
easat-3242	289	11	carlos	carlos	PROPN
easat-3242	289	12	aranda	aranda	PROPN
easat-3242	289	13	,	,	PUNCT
easat-3242	289	14	and	and	CCONJ
easat-3242	289	15	mikel	mikel	PROPN
easat-3242	289	16	galar	galar	PROPN
easat-3242	289	17	.	.	PUNCT
easat-3242	290	1	"	"	PUNCT
easat-3242	290	2	a	a	DET
easat-3242	290	3	deep	deep	ADJ
easat-3242	290	4	learning	learning	NOUN
easat-3242	290	5	approach	approach	NOUN
easat-3242	290	6	to	to	ADP
easat-3242	290	7	an	an	DET
easat-3242	290	8	enhanced	enhance	VERB
easat-3242	290	9	building	building	NOUN
easat-3242	290	10	footprint	footprint	NOUN
easat-3242	290	11	and	and	CCONJ
easat-3242	290	12	road	road	NOUN
easat-3242	290	13	detection	detection	NOUN
easat-3242	290	14	in	in	ADP
easat-3242	290	15	high	high	ADJ
easat-3242	290	16	-	-	PUNCT
easat-3242	290	17	resolution	resolution	NOUN
easat-3242	290	18	satellite	satellite	NOUN
easat-3242	290	19	imagery	imagery	NOUN
easat-3242	290	20	.	.	PUNCT
easat-3242	290	21	"	"	PUNCT
easat-3242	291	1	remote	remote	ADJ
easat-3242	291	2	sensing	sense	VERB
easat-3242	291	3	13	13	NUM
easat-3242	291	4	,	,	PUNCT
easat-3242	291	5	no	no	INTJ
easat-3242	291	6	.	.	NOUN
easat-3242	291	7	16	16	NUM
easat-3242	291	8	(	(	PUNCT
easat-3242	291	9	2021	2021	NUM
easat-3242	291	10	):	):	PUNCT
easat-3242	291	11	3135	3135	NUM
easat-3242	291	12	.	.	PUNCT
easat-3242	292	1	[	[	X
easat-3242	292	2	4	4	X
easat-3242	292	3	]	]	PUNCT
easat-3242	292	4	sheykhmousa	sheykhmousa	NOUN
easat-3242	292	5	,	,	PUNCT
easat-3242	292	6	mohammadreza	mohammadreza	NOUN
easat-3242	292	7	,	,	PUNCT
easat-3242	292	8	norman	norman	PROPN
easat-3242	292	9	kerle	kerle	PROPN
easat-3242	292	10	,	,	PUNCT
easat-3242	292	11	monika	monika	PROPN
easat-3242	292	12	kuffer	kuffer	PROPN
easat-3242	292	13	,	,	PUNCT
easat-3242	292	14	and	and	CCONJ
easat-3242	292	15	samanghaffarian	samanghaffarian	NOUN
easat-3242	292	16	.	.	PUNCT
easat-3242	293	1	"	"	PUNCT
easat-3242	293	2	post	post	ADJ
easat-3242	293	3	-	-	ADJ
easat-3242	293	4	disaster	disaster	ADJ
easat-3242	293	5	recovery	recovery	NOUN
easat-3242	293	6	assessment	assessment	NOUN
easat-3242	293	7	with	with	ADP
easat-3242	293	8	machine	machine	NOUN
easat-3242	293	9	learning	learning	NOUN
easat-3242	293	10	-	-	PUNCT
easat-3242	293	11	derived	derive	VERB
easat-3242	293	12	land	land	NOUN
easat-3242	293	13	cover	cover	NOUN
easat-3242	293	14	and	and	CCONJ
easat-3242	293	15	land	land	NOUN
easat-3242	293	16	use	use	NOUN
easat-3242	293	17	information	information	NOUN
easat-3242	293	18	.	.	PUNCT
easat-3242	293	19	"	"	PUNCT
easat-3242	294	1	remote	remote	ADJ
easat-3242	294	2	sensing	sense	VERB
easat-3242	294	3	11	11	NUM
easat-3242	294	4	,	,	PUNCT
easat-3242	294	5	no	no	INTJ
easat-3242	294	6	.	.	NOUN
easat-3242	295	1	10	10	NUM
easat-3242	296	1	(	(	PUNCT
easat-3242	296	2	2019	2019	NUM
easat-3242	296	3	):	):	PUNCT
easat-3242	296	4	1174	1174	NUM
easat-3242	296	5	.	.	PUNCT
easat-3242	297	1	[	[	X
easat-3242	297	2	5	5	NUM
easat-3242	297	3	]	]	X
easat-3242	297	4	wagner	wagner	PROPN
easat-3242	297	5	,	,	PUNCT
easat-3242	297	6	fabien	fabien	PROPN
easat-3242	297	7	h.	h.	PROPN
easat-3242	297	8	,	,	PUNCT
easat-3242	297	9	ricardo	ricardo	PROPN
easat-3242	297	10	dalagnol	dalagnol	NOUN
easat-3242	297	11	,	,	PUNCT
easat-3242	297	12	yuliyatarabalka	yuliyatarabalka	ADJ
easat-3242	297	13	,	,	PUNCT
easat-3242	297	14	tassiana	tassiana	NOUN
easat-3242	297	15	yf	yf	NOUN
easat-3242	297	16	segantine	segantine	NOUN
easat-3242	297	17	,	,	PUNCT
easat-3242	297	18	rogériothomé	rogériothomé	NOUN
easat-3242	297	19	,	,	PUNCT
easat-3242	297	20	and	and	CCONJ
easat-3242	297	21	mayumi	mayumi	PROPN
easat-3242	297	22	cm	cm	PROPN
easat-3242	297	23	hirye	hirye	NOUN
easat-3242	297	24	.	.	PUNCT
easat-3242	298	1	"	"	PUNCT
easat-3242	298	2	u	u	NOUN
easat-3242	298	3	-	-	NOUN
easat-3242	298	4	net	net	ADJ
easat-3242	298	5	-	-	PUNCT
easat-3242	298	6	id	id	NOUN
easat-3242	298	7	,	,	PUNCT
easat-3242	298	8	an	an	DET
easat-3242	298	9	instance	instance	NOUN
easat-3242	298	10	segmentation	segmentation	NOUN
easat-3242	298	11	model	model	NOUN
easat-3242	298	12	for	for	ADP
easat-3242	298	13	building	build	VERB
easat-3242	298	14	extraction	extraction	NOUN
easat-3242	298	15	from	from	ADP
easat-3242	298	16	satellite	satellite	NOUN
easat-3242	298	17	images	image	NOUN
easat-3242	298	18	—	—	PUNCT
easat-3242	298	19	case	case	NOUN
easat-3242	298	20	study	study	NOUN
easat-3242	298	21	in	in	ADP
easat-3242	298	22	the	the	DET
easat-3242	298	23	joanópolis	joanópolis	PROPN
easat-3242	298	24	city	city	PROPN
easat-3242	298	25	,	,	PUNCT
easat-3242	298	26	brazil	brazil	PROPN
easat-3242	298	27	.	.	PUNCT
easat-3242	298	28	"	"	PUNCT
easat-3242	299	1	remote	remote	ADJ
easat-3242	299	2	sensing	sense	VERB
easat-3242	299	3	12	12	NUM
easat-3242	299	4	,	,	PUNCT
easat-3242	299	5	no	no	INTJ
easat-3242	299	6	.	.	NOUN
easat-3242	299	7	10	10	NUM
easat-3242	299	8	(	(	PUNCT
easat-3242	299	9	2020	2020	NUM
easat-3242	299	10	):	):	PUNCT
easat-3242	299	11	1544	1544	NUM
easat-3242	299	12	.	.	PUNCT
easat-3242	300	1	[	[	X
easat-3242	300	2	6	6	NUM
easat-3242	300	3	]	]	SYM
easat-3242	300	4	li	li	PROPN
easat-3242	300	5	,	,	PUNCT
easat-3242	300	6	zhiwei	zhiwei	PROPN
easat-3242	300	7	,	,	PUNCT
easat-3242	300	8	huanfeng	huanfeng	PROPN
easat-3242	300	9	shen	shen	PROPN
easat-3242	300	10	,	,	PUNCT
easat-3242	300	11	qihaoweng	qihaoweng	PROPN
easat-3242	300	12	,	,	PUNCT
easat-3242	300	13	yuzhuo	yuzhuo	PROPN
easat-3242	300	14	zhang	zhang	PROPN
easat-3242	300	15	,	,	PUNCT
easat-3242	300	16	peng	peng	PROPN
easat-3242	300	17	dou	dou	PROPN
easat-3242	300	18	,	,	PUNCT
easat-3242	300	19	and	and	CCONJ
easat-3242	300	20	liangpei	liangpei	PROPN
easat-3242	300	21	zhang	zhang	PROPN
easat-3242	300	22	.	.	PUNCT
easat-3242	301	1	"	"	PUNCT
easat-3242	301	2	cloud	cloud	ADJ
easat-3242	301	3	and	and	CCONJ
easat-3242	301	4	cloud	cloud	NOUN
easat-3242	301	5	shadow	shadow	NOUN
easat-3242	301	6	detection	detection	NOUN
easat-3242	301	7	for	for	ADP
easat-3242	301	8	optical	optical	ADJ
easat-3242	301	9	satellite	satellite	NOUN
easat-3242	301	10	imagery	imagery	NOUN
easat-3242	301	11	:	:	PUNCT
easat-3242	301	12	features	feature	NOUN
easat-3242	301	13	,	,	PUNCT
easat-3242	301	14	algorithms	algorithm	NOUN
easat-3242	301	15	,	,	PUNCT
easat-3242	301	16	validation	validation	NOUN
easat-3242	301	17	,	,	PUNCT
easat-3242	301	18	and	and	CCONJ
easat-3242	301	19	prospects	prospect	NOUN
easat-3242	301	20	.	.	PUNCT
easat-3242	301	21	"	"	PUNCT
easat-3242	301	22	isprs	isprs	PROPN
easat-3242	301	23	journal	journal	NOUN
easat-3242	301	24	of	of	ADP
easat-3242	301	25	photogrammetry	photogrammetry	NOUN
easat-3242	301	26	and	and	CCONJ
easat-3242	301	27	remote	remote	ADJ
easat-3242	301	28	sensing	sense	VERB
easat-3242	301	29	188	188	NUM
easat-3242	301	30	(	(	PUNCT
easat-3242	301	31	2022	2022	NUM
easat-3242	301	32	):	):	PUNCT
easat-3242	301	33	89	89	NUM
easat-3242	301	34	-	-	SYM
easat-3242	301	35	108	108	NUM
easat-3242	301	36	.	.	PUNCT
easat-3242	302	1	[	[	X
easat-3242	302	2	7	7	NUM
easat-3242	302	3	]	]	X
easat-3242	302	4	rahaman	rahaman	NOUN
easat-3242	302	5	,	,	PUNCT
easat-3242	302	6	jarjish	jarjish	ADJ
easat-3242	302	7	,	,	PUNCT
easat-3242	302	8	and	and	CCONJ
easat-3242	302	9	mihir	mihir	PROPN
easat-3242	302	10	sing	sing	VERB
easat-3242	302	11	.	.	PUNCT
easat-3242	303	1	"	"	PUNCT
easat-3242	303	2	an	an	DET
easat-3242	303	3	efficient	efficient	ADJ
easat-3242	303	4	multilevel	multilevel	NOUN
easat-3242	303	5	thresholding	thresholding	NOUN
easat-3242	303	6	based	base	VERB
easat-3242	303	7	satellite	satellite	NOUN
easat-3242	303	8	image	image	NOUN
easat-3242	303	9	segmentation	segmentation	NOUN
easat-3242	303	10	approach	approach	NOUN
easat-3242	303	11	using	use	VERB
easat-3242	303	12	a	a	DET
easat-3242	303	13	new	new	ADJ
easat-3242	303	14	adaptive	adaptive	ADJ
easat-3242	303	15	cuckoo	cuckoo	NOUN
easat-3242	303	16	search	search	NOUN
easat-3242	303	17	algorithm	algorithm	NOUN
easat-3242	303	18	.	.	PUNCT
easat-3242	303	19	"	"	PUNCT
easat-3242	304	1	expert	expert	ADJ
easat-3242	304	2	systems	system	NOUN
easat-3242	304	3	with	with	ADP
easat-3242	304	4	applications	application	NOUN
easat-3242	304	5	174	174	NUM
easat-3242	304	6	(	(	PUNCT
easat-3242	304	7	2021	2021	NUM
easat-3242	304	8	):	):	PUNCT
easat-3242	304	9	114633	114633	NUM
easat-3242	304	10	.	.	PUNCT
easat-3242	305	1	[	[	X
easat-3242	305	2	8	8	NUM
easat-3242	305	3	]	]	SYM
easat-3242	305	4	sisodiya	sisodiya	ADJ
easat-3242	305	5	,	,	PUNCT
easat-3242	305	6	neha	neha	NOUN
easat-3242	305	7	,	,	PUNCT
easat-3242	305	8	nitantdube	nitantdube	NOUN
easat-3242	305	9	,	,	PUNCT
easat-3242	305	10	and	and	CCONJ
easat-3242	305	11	priyank	priyank	PROPN
easat-3242	305	12	thakkar	thakkar	NOUN
easat-3242	305	13	.	.	PUNCT
easat-3242	306	1	"	"	PUNCT
easat-3242	306	2	next	next	ADJ
easat-3242	306	3	-	-	PUNCT
easat-3242	306	4	generation	generation	NOUN
easat-3242	306	5	artificial	artificial	ADJ
easat-3242	306	6	intelligence	intelligence	NOUN
easat-3242	306	7	techniques	technique	NOUN
easat-3242	306	8	for	for	ADP
easat-3242	306	9	satellite	satellite	NOUN
easat-3242	306	10	data	datum	NOUN
easat-3242	306	11	processing	processing	NOUN
easat-3242	306	12	.	.	PUNCT
easat-3242	306	13	"	"	PUNCT
easat-3242	307	1	artificial	artificial	ADJ
easat-3242	307	2	intelligence	intelligence	NOUN
easat-3242	307	3	techniques	technique	NOUN
easat-3242	307	4	for	for	ADP
easat-3242	307	5	satellite	satellite	NOUN
easat-3242	307	6	image	image	NOUN
easat-3242	307	7	analysis	analysis	NOUN
easat-3242	307	8	(	(	PUNCT
easat-3242	307	9	2020	2020	NUM
easat-3242	307	10	):	):	PUNCT
easat-3242	307	11	235	235	NUM
easat-3242	307	12	-	-	SYM
easat-3242	307	13	254	254	NUM
easat-3242	307	14	.	.	PUNCT
easat-3242	308	1	[	[	X
easat-3242	308	2	9	9	NUM
easat-3242	308	3	]	]	SYM
easat-3242	308	4	yuan	yuan	NOUN
easat-3242	308	5	,	,	PUNCT
easat-3242	308	6	kunhao	kunhao	PROPN
easat-3242	308	7	,	,	PUNCT
easat-3242	308	8	xu	xu	PROPN
easat-3242	308	9	zhuang	zhuang	PROPN
easat-3242	308	10	,	,	PUNCT
easat-3242	308	11	gerald	gerald	PROPN
easat-3242	308	12	schaefer	schaefer	PROPN
easat-3242	308	13	,	,	PUNCT
easat-3242	308	14	jianxin	jianxin	PROPN
easat-3242	308	15	feng	feng	PROPN
easat-3242	308	16	,	,	PUNCT
easat-3242	308	17	lin	lin	PROPN
easat-3242	308	18	guan	guan	PROPN
easat-3242	308	19	,	,	PUNCT
easat-3242	308	20	and	and	CCONJ
easat-3242	308	21	hui	hui	PROPN
easat-3242	308	22	fang	fang	PROPN
easat-3242	308	23	.	.	PUNCT
easat-3242	309	1	"	"	PUNCT
easat-3242	309	2	deep	deep	ADJ
easat-3242	309	3	-	-	PUNCT
easat-3242	309	4	learning	learning	NOUN
easat-3242	309	5	-	-	PUNCT
easat-3242	309	6	based	base	VERB
easat-3242	309	7	multispectral	multispectral	ADJ
easat-3242	309	8	satellite	satellite	NOUN
easat-3242	309	9	image	image	NOUN
easat-3242	309	10	segmentation	segmentation	NOUN
easat-3242	309	11	for	for	ADP
easat-3242	309	12	water	water	NOUN
easat-3242	309	13	body	body	NOUN
easat-3242	309	14	detection	detection	NOUN
easat-3242	309	15	.	.	PUNCT
easat-3242	309	16	"	"	PUNCT
easat-3242	310	1	ieee	ieee	PROPN
easat-3242	310	2	journal	journal	NOUN
easat-3242	310	3	of	of	ADP
easat-3242	310	4	selected	select	VERB
easat-3242	310	5	topics	topic	NOUN
easat-3242	310	6	in	in	ADP
easat-3242	310	7	applied	apply	VERB
easat-3242	310	8	earth	earth	NOUN
easat-3242	310	9	observations	observation	NOUN
easat-3242	310	10	and	and	CCONJ
easat-3242	310	11	remote	remote	ADJ
easat-3242	310	12	sensing	sense	VERB
easat-3242	310	13	14	14	NUM
easat-3242	310	14	(	(	PUNCT
easat-3242	310	15	2021	2021	NUM
easat-3242	310	16	):	):	PUNCT
easat-3242	310	17	7422	7422	NUM
easat-3242	310	18	-	-	SYM
easat-3242	310	19	7434	7434	NUM
easat-3242	310	20	.	.	PUNCT
easat-3242	311	1	[	[	X
easat-3242	311	2	10	10	NUM
easat-3242	311	3	]	]	X
easat-3242	311	4	pan	pan	PROPN
easat-3242	311	5	,	,	PUNCT
easat-3242	311	6	zhuokun	zhuokun	NOUN
easat-3242	311	7	,	,	PUNCT
easat-3242	311	8	jiashu	jiashu	PROPN
easat-3242	311	9	xu	xu	PROPN
easat-3242	311	10	,	,	PUNCT
easat-3242	311	11	yubinguo	yubinguo	PROPN
easat-3242	311	12	,	,	PUNCT
easat-3242	311	13	yueming	yueming	NOUN
easat-3242	311	14	hu	hu	PROPN
easat-3242	311	15	,	,	PUNCT
easat-3242	311	16	and	and	CCONJ
easat-3242	311	17	guangxing	guangxe	VERB
easat-3242	311	18	wang	wang	PROPN
easat-3242	311	19	.	.	PUNCT
easat-3242	312	1	"	"	PUNCT
easat-3242	312	2	deep	deep	ADJ
easat-3242	312	3	learning	learn	VERB
easat-3242	312	4	segmentation	segmentation	NOUN
easat-3242	312	5	and	and	CCONJ
easat-3242	312	6	classification	classification	NOUN
easat-3242	312	7	for	for	ADP
easat-3242	312	8	urban	urban	ADJ
easat-3242	312	9	village	village	NOUN
easat-3242	312	10	using	use	VERB
easat-3242	312	11	a	a	DET
easat-3242	312	12	worldview	worldview	ADJ
easat-3242	312	13	satellite	satellite	NOUN
easat-3242	312	14	image	image	NOUN
easat-3242	312	15	based	base	VERB
easat-3242	312	16	on	on	ADP
easat-3242	312	17	u	u	NOUN
easat-3242	312	18	-	-	NOUN
easat-3242	312	19	net	net	NOUN
easat-3242	312	20	.	.	PUNCT
easat-3242	312	21	"	"	PUNCT
easat-3242	313	1	remote	remote	ADJ
easat-3242	313	2	sensing	sense	VERB
easat-3242	313	3	12	12	NUM
easat-3242	313	4	,	,	PUNCT
easat-3242	313	5	no	no	INTJ
easat-3242	313	6	.	.	NOUN
easat-3242	313	7	10	10	NUM
easat-3242	313	8	(	(	PUNCT
easat-3242	313	9	2020	2020	NUM
easat-3242	313	10	):	):	PUNCT
easat-3242	313	11	1574	1574	NUM
easat-3242	313	12	.	.	PUNCT
easat-3242	314	1	[	[	X
easat-3242	314	2	11	11	NUM
easat-3242	314	3	]	]	X
easat-3242	314	4	raghavan	raghavan	PROPN
easat-3242	314	5	,	,	PUNCT
easat-3242	314	6	ramesh	ramesh	PROPN
easat-3242	314	7	,	,	PUNCT
easat-3242	314	8	dinesh	dinesh	PROPN
easat-3242	314	9	chanderverma	chanderverma	PROPN
easat-3242	314	10	,	,	PUNCT
easat-3242	314	11	digvijay	digvijay	PROPN
easat-3242	314	12	pandey	pandey	PROPN
easat-3242	314	13	,	,	PUNCT
easat-3242	314	14	rohitanand	rohitanand	PROPN
easat-3242	314	15	,	,	PUNCT
easat-3242	314	16	binay	binay	PROPN
easat-3242	314	17	kumar	kumar	PROPN
easat-3242	314	18	pandey	pandey	PROPN
easat-3242	314	19	,	,	PUNCT
easat-3242	314	20	and	and	CCONJ
easat-3242	314	21	harinder	harinder	PROPN
easat-3242	314	22	singh	singh	PROPN
easat-3242	314	23	.	.	PUNCT
easat-3242	315	1	"	"	PUNCT
easat-3242	315	2	optimized	optimize	VERB
easat-3242	315	3	building	building	NOUN
easat-3242	315	4	extraction	extraction	NOUN
easat-3242	315	5	from	from	ADP
easat-3242	315	6	high	high	ADJ
easat-3242	315	7	-	-	PUNCT
easat-3242	315	8	resolution	resolution	NOUN
easat-3242	315	9	satellite	satellite	NOUN
easat-3242	315	10	imagery	imagery	NOUN
easat-3242	315	11	using	use	VERB
easat-3242	315	12	deep	deep	ADJ
easat-3242	315	13	learning	learning	NOUN
easat-3242	315	14	.	.	PUNCT
easat-3242	315	15	"	"	PUNCT
easat-3242	315	16	multimedia	multimedia	NOUN
easat-3242	315	17	tools	tool	NOUN
easat-3242	315	18	and	and	CCONJ
easat-3242	315	19	applications	application	NOUN
easat-3242	315	20	81	81	NUM
easat-3242	315	21	,	,	PUNCT
easat-3242	315	22	no	no	INTJ
easat-3242	315	23	.	.	NOUN
easat-3242	315	24	29	29	NUM
easat-3242	315	25	(	(	PUNCT
easat-3242	315	26	2022	2022	NUM
easat-3242	315	27	):	):	PUNCT
easat-3242	315	28	42309	42309	NUM
easat-3242	315	29	-	-	SYM
easat-3242	315	30	42323	42323	NUM
easat-3242	315	31	.	.	PUNCT
easat-3242	316	1	[	[	X
easat-3242	316	2	12	12	NUM
easat-3242	316	3	]	]	X
easat-3242	316	4	nalepa	nalepa	PROPN
easat-3242	316	5	,	,	PUNCT
easat-3242	316	6	jakub	jakub	PROPN
easat-3242	316	7	,	,	PUNCT
easat-3242	316	8	michal	michal	PROPN
easat-3242	316	9	myller	myller	PROPN
easat-3242	316	10	,	,	PUNCT
easat-3242	316	11	marcin	marcin	PROPN
easat-3242	316	12	cwiek	cwiek	NOUN
easat-3242	316	13	,	,	PUNCT
easat-3242	316	14	lukasz	lukasz	PROPN
easat-3242	316	15	zak	zak	PROPN
easat-3242	316	16	,	,	PUNCT
easat-3242	316	17	tomasz	tomasz	PROPN
easat-3242	316	18	lakota	lakota	PROPN
easat-3242	316	19	,	,	PUNCT
easat-3242	316	20	lukasz	lukasz	PROPN
easat-3242	316	21	tulczyjew	tulczyjew	PROPN
easat-3242	316	22	,	,	PUNCT
easat-3242	316	23	and	and	CCONJ
easat-3242	316	24	michal	michal	PROPN
easat-3242	316	25	kawulok	kawulok	PROPN
easat-3242	316	26	.	.	PUNCT
easat-3242	317	1	"	"	PUNCT
easat-3242	317	2	towards	towards	ADP
easat-3242	317	3	on	on	ADP
easat-3242	317	4	-	-	PUNCT
easat-3242	317	5	board	board	NOUN
easat-3242	317	6	hyperspectral	hyperspectral	ADJ
easat-3242	317	7	satellite	satellite	NOUN
easat-3242	317	8	image	image	NOUN
easat-3242	317	9	segmentation	segmentation	NOUN
easat-3242	317	10	:	:	PUNCT
easat-3242	317	11	understanding	understand	VERB
easat-3242	317	12	robustness	robustness	NOUN
easat-3242	317	13	of	of	ADP
easat-3242	317	14	deep	deep	ADJ
easat-3242	317	15	learning	learning	NOUN
easat-3242	317	16	through	through	ADP
easat-3242	317	17	simulating	simulate	VERB
easat-3242	317	18	acquisition	acquisition	NOUN
easat-3242	317	19	conditions	condition	NOUN
easat-3242	317	20	.	.	PUNCT
easat-3242	317	21	"	"	PUNCT
easat-3242	318	1	remote	remote	ADJ
easat-3242	318	2	sensing	sense	VERB
easat-3242	318	3	13	13	NUM
easat-3242	318	4	,	,	PUNCT
easat-3242	318	5	no	no	INTJ
easat-3242	318	6	.	.	NOUN
easat-3242	318	7	8	8	NUM
easat-3242	318	8	(	(	PUNCT
easat-3242	318	9	2021	2021	NUM
easat-3242	318	10	):	):	PUNCT
easat-3242	318	11	1532	1532	NUM
easat-3242	318	12	.	.	PUNCT
easat-3242	319	1	[	[	X
easat-3242	319	2	13	13	NUM
easat-3242	319	3	]	]	X
easat-3242	319	4	nguyen	nguyen	NOUN
easat-3242	319	5	,	,	PUNCT
easat-3242	319	6	thanh	thanh	PROPN
easat-3242	319	7	tam	tam	PROPN
easat-3242	319	8	,	,	PUNCT
easat-3242	319	9	thanhdat	thanhdat	PROPN
easat-3242	319	10	hoang	hoang	PROPN
easat-3242	319	11	,	,	PUNCT
easat-3242	319	12	minh	minh	PROPN
easat-3242	319	13	tam	tam	PROPN
easat-3242	319	14	pham	pham	PROPN
easat-3242	319	15	,	,	PUNCT
easat-3242	319	16	tuyet	tuyet	PROPN
easat-3242	319	17	trinh	trinh	PROPN
easat-3242	319	18	vu	vu	PROPN
easat-3242	319	19	,	,	PUNCT
easat-3242	319	20	thanh	thanh	PROPN
easat-3242	319	21	hung	hung	PROPN
easat-3242	319	22	nguyen	nguyen	PROPN
easat-3242	319	23	,	,	PUNCT
easat-3242	319	24	quyet	quyet	PROPN
easat-3242	319	25	-	-	PUNCT
easat-3242	319	26	thang	thang	PROPN
easat-3242	319	27	huynh	huynh	PROPN
easat-3242	319	28	,	,	PUNCT
easat-3242	319	29	and	and	CCONJ
easat-3242	319	30	jun	jun	PROPN
easat-3242	319	31	jo	jo	PROPN
easat-3242	319	32	.	.	PUNCT
easat-3242	320	1	"	"	PUNCT
easat-3242	320	2	monitoring	monitor	VERB
easat-3242	320	3	agriculture	agriculture	NOUN
easat-3242	320	4	areas	area	NOUN
easat-3242	320	5	with	with	ADP
easat-3242	320	6	satellite	satellite	NOUN
easat-3242	320	7	images	image	NOUN
easat-3242	320	8	and	and	CCONJ
easat-3242	320	9	deep	deep	ADJ
easat-3242	320	10	learning	learning	NOUN
easat-3242	320	11	.	.	PUNCT
easat-3242	320	12	"	"	PUNCT
easat-3242	321	1	applied	apply	VERB
easat-3242	321	2	soft	soft	ADJ
easat-3242	321	3	computing	compute	VERB
easat-3242	321	4	95	95	NUM
easat-3242	321	5	(	(	PUNCT
easat-3242	321	6	2020	2020	NUM
easat-3242	321	7	):	):	PUNCT
easat-3242	321	8	106565	106565	NUM
easat-3242	321	9	.	.	PUNCT
easat-3242	322	1	[	[	X
easat-3242	322	2	14	14	NUM
easat-3242	322	3	]	]	X
easat-3242	322	4	khan	khan	PROPN
easat-3242	322	5	,	,	PUNCT
easat-3242	322	6	sultan	sultan	PROPN
easat-3242	322	7	daud	daud	PROPN
easat-3242	322	8	,	,	PUNCT
easat-3242	322	9	louaialarabi	louaialarabi	NOUN
easat-3242	322	10	,	,	PUNCT
easat-3242	322	11	and	and	CCONJ
easat-3242	322	12	saleh	saleh	PROPN
easat-3242	322	13	basalamah	basalamah	PROPN
easat-3242	322	14	.	.	PUNCT
easat-3242	323	1	"	"	PUNCT
easat-3242	323	2	deep	deep	ADJ
easat-3242	323	3	hybrid	hybrid	NOUN
easat-3242	323	4	network	network	NOUN
easat-3242	323	5	for	for	ADP
easat-3242	323	6	land	land	NOUN
easat-3242	323	7	cover	cover	VERB
easat-3242	323	8	semantic	semantic	ADJ
easat-3242	323	9	segmentation	segmentation	NOUN
easat-3242	323	10	in	in	ADP
easat-3242	323	11	high	high	ADJ
easat-3242	323	12	-	-	PUNCT
easat-3242	323	13	spatial	spatial	ADJ
easat-3242	323	14	resolution	resolution	NOUN
easat-3242	323	15	satellite	satellite	NOUN
easat-3242	323	16	images	image	NOUN
easat-3242	323	17	.	.	PUNCT
easat-3242	323	18	"	"	PUNCT
easat-3242	324	1	information	information	NOUN
easat-3242	324	2	12	12	NUM
easat-3242	324	3	,	,	PUNCT
easat-3242	324	4	no	no	INTJ
easat-3242	324	5	.	.	NOUN
easat-3242	324	6	6	6	NUM
easat-3242	324	7	(	(	PUNCT
easat-3242	324	8	2021	2021	NUM
easat-3242	324	9	):	):	PUNCT
easat-3242	324	10	230	230	NUM
easat-3242	324	11	.	.	PUNCT
easat-3242	325	1	[	[	X
easat-3242	325	2	15	15	NUM
easat-3242	325	3	]	]	X
easat-3242	325	4	neupane	neupane	ADJ
easat-3242	325	5	,	,	PUNCT
easat-3242	325	6	bipul	bipul	NOUN
easat-3242	325	7	,	,	PUNCT
easat-3242	325	8	teerayuthoranont	teerayuthoranont	NOUN
easat-3242	325	9	,	,	PUNCT
easat-3242	325	10	and	and	CCONJ
easat-3242	325	11	jagannatharyal	jagannatharyal	NOUN
easat-3242	325	12	.	.	PUNCT
easat-3242	326	1	"	"	PUNCT
easat-3242	326	2	deep	deep	ADJ
easat-3242	326	3	learning	learning	NOUN
easat-3242	326	4	-	-	PUNCT
easat-3242	326	5	based	base	VERB
easat-3242	326	6	semantic	semantic	ADJ
easat-3242	326	7	segmentation	segmentation	NOUN
easat-3242	326	8	of	of	ADP
easat-3242	326	9	urban	urban	ADJ
easat-3242	326	10	features	feature	NOUN
easat-3242	326	11	in	in	ADP
easat-3242	326	12	satellite	satellite	NOUN
easat-3242	326	13	images	image	NOUN
easat-3242	326	14	:	:	PUNCT
easat-3242	326	15	a	a	DET
easat-3242	326	16	review	review	NOUN
easat-3242	326	17	and	and	CCONJ
easat-3242	326	18	meta	meta	ADJ
easat-3242	326	19	-	-	PUNCT
easat-3242	326	20	analysis	analysis	NOUN
easat-3242	326	21	.	.	PUNCT
easat-3242	326	22	"	"	PUNCT
easat-3242	327	1	remote	remote	ADJ
easat-3242	327	2	sensing	sense	VERB
easat-3242	327	3	13	13	NUM
easat-3242	327	4	,	,	PUNCT
easat-3242	327	5	no	no	INTJ
easat-3242	327	6	.	.	NOUN
easat-3242	327	7	4	4	NUM
easat-3242	327	8	(	(	PUNCT
easat-3242	327	9	2021	2021	NUM
easat-3242	327	10	):	):	PUNCT
easat-3242	327	11	808	808	NUM
easat-3242	327	12	.	.	PUNCT
easat-3242	328	1	[	[	X
easat-3242	328	2	16	16	NUM
easat-3242	328	3	]	]	X
easat-3242	328	4	tahir	tahir	PROPN
easat-3242	328	5	,	,	PUNCT
easat-3242	328	6	arsalan	arsalan	PROPN
easat-3242	328	7	,	,	PUNCT
easat-3242	328	8	hafiz	hafiz	PROPN
easat-3242	328	9	sulimanmunawar	sulimanmunawar	PROPN
easat-3242	328	10	,	,	PUNCT
easat-3242	328	11	junaidakram	junaidakram	PROPN
easat-3242	328	12	,	,	PUNCT
easat-3242	328	13	muhammad	muhammad	PROPN
easat-3242	328	14	adil	adil	PROPN
easat-3242	328	15	,	,	PUNCT
easat-3242	328	16	shehryar	shehryar	PROPN
easat-3242	328	17	ali	ali	PROPN
easat-3242	328	18	,	,	PUNCT
easat-3242	328	19	abbas	abbas	PROPN
easat-3242	328	20	z.	z.	PROPN
easat-3242	328	21	kouzani	kouzani	PROPN
easat-3242	328	22	,	,	PUNCT
easat-3242	328	23	and	and	CCONJ
easat-3242	328	24	ma	ma	PROPN
easat-3242	328	25	parvez	parvez	PROPN
easat-3242	328	26	mahmud	mahmud	PROPN
easat-3242	328	27	.	.	PUNCT
easat-3242	329	1	"	"	PUNCT
easat-3242	329	2	automatic	automatic	ADJ
easat-3242	329	3	target	target	NOUN
easat-3242	329	4	detection	detection	NOUN
easat-3242	329	5	from	from	ADP
easat-3242	329	6	satellite	satellite	NOUN
easat-3242	329	7	imagery	imagery	NOUN
easat-3242	329	8	using	use	VERB
easat-3242	329	9	machine	machine	NOUN
easat-3242	329	10	learning	learning	NOUN
easat-3242	329	11	.	.	PUNCT
easat-3242	329	12	"	"	PUNCT
easat-3242	330	1	sensors	sensor	NOUN
easat-3242	330	2	22	22	NUM
easat-3242	330	3	,	,	PUNCT
easat-3242	330	4	no	no	INTJ
easat-3242	330	5	.	.	NOUN
easat-3242	330	6	3	3	NUM
easat-3242	330	7	(	(	PUNCT
easat-3242	330	8	2022	2022	NUM
easat-3242	330	9	):	):	PUNCT
easat-3242	330	10	1147	1147	NUM
easat-3242	330	11	.	.	PUNCT
easat-3242	331	1	[	[	X
easat-3242	331	2	17	17	NUM
easat-3242	331	3	]	]	X
easat-3242	331	4	rashkovetsky	rashkovetsky	NOUN
easat-3242	331	5	,	,	PUNCT
easat-3242	331	6	dmitry	dmitry	PROPN
easat-3242	331	7	,	,	PUNCT
easat-3242	331	8	florian	florian	ADJ
easat-3242	331	9	mauracher	mauracher	PROPN
easat-3242	331	10	,	,	PUNCT
easat-3242	331	11	martin	martin	PROPN
easat-3242	331	12	langer	langer	PROPN
easat-3242	331	13	,	,	PUNCT
easat-3242	331	14	and	and	CCONJ
easat-3242	331	15	michael	michael	PROPN
easat-3242	331	16	schmitt	schmitt	PROPN
easat-3242	331	17	.	.	PUNCT
easat-3242	332	1	"	"	PUNCT
easat-3242	332	2	wildfire	wildfire	NOUN
easat-3242	332	3	detection	detection	NOUN
easat-3242	332	4	from	from	ADP
easat-3242	332	5	multisensor	multisensor	NOUN
easat-3242	332	6	satellite	satellite	NOUN
easat-3242	332	7	imagery	imagery	NOUN
easat-3242	332	8	using	use	VERB
easat-3242	332	9	deep	deep	ADJ
easat-3242	332	10	semantic	semantic	ADJ
easat-3242	332	11	segmentation	segmentation	NOUN
easat-3242	332	12	.	.	PUNCT
easat-3242	332	13	"	"	PUNCT
easat-3242	333	1	ieee	ieee	PROPN
easat-3242	333	2	journal	journal	NOUN
easat-3242	333	3	of	of	ADP
easat-3242	333	4	selected	select	VERB
easat-3242	333	5	topics	topic	NOUN
easat-3242	333	6	in	in	ADP
easat-3242	333	7	applied	apply	VERB
easat-3242	333	8	earth	earth	NOUN
easat-3242	333	9	observations	observation	NOUN
easat-3242	333	10	and	and	CCONJ
easat-3242	333	11	remote	remote	ADJ
easat-3242	333	12	sensing	sense	VERB
easat-3242	333	13	14	14	NUM
easat-3242	333	14	(	(	PUNCT
easat-3242	333	15	2021	2021	NUM
easat-3242	333	16	):	):	PUNCT
easat-3242	333	17	7001	7001	NUM
easat-3242	333	18	-	-	SYM
easat-3242	333	19	7016	7016	NUM
easat-3242	333	20	.	.	PUNCT
easat-3242	334	1	[	[	X
easat-3242	334	2	18	18	NUM
easat-3242	334	3	]	]	X
easat-3242	334	4	saralioglu	saralioglu	NOUN
easat-3242	334	5	,	,	PUNCT
easat-3242	334	6	ekrem	ekrem	VERB
easat-3242	334	7	,	,	PUNCT
easat-3242	334	8	and	and	CCONJ
easat-3242	334	9	oguzgungor	oguzgungor	ADJ
easat-3242	334	10	.	.	PUNCT
easat-3242	335	1	"	"	PUNCT
easat-3242	335	2	semantic	semantic	ADJ
easat-3242	335	3	segmentation	segmentation	NOUN
easat-3242	335	4	of	of	ADP
easat-3242	335	5	land	land	NOUN
easat-3242	335	6	cover	cover	NOUN
easat-3242	335	7	from	from	ADP
easat-3242	335	8	high	high	ADJ
easat-3242	335	9	resolution	resolution	NOUN
easat-3242	335	10	multispectral	multispectral	ADJ
easat-3242	335	11	satellite	satellite	NOUN
easat-3242	335	12	images	image	NOUN
easat-3242	335	13	by	by	ADP
easat-3242	335	14	spectral	spectral	ADJ
easat-3242	335	15	-	-	PUNCT
easat-3242	335	16	spatial	spatial	ADJ
easat-3242	335	17	convolutional	convolutional	ADJ
easat-3242	335	18	neural	neural	ADJ
easat-3242	335	19	network	network	NOUN
easat-3242	335	20	.	.	PUNCT
easat-3242	335	21	"	"	PUNCT
easat-3242	336	1	geocarto	geocarto	PROPN
easat-3242	336	2	international	international	PROPN
easat-3242	336	3	37	37	NUM
easat-3242	336	4	,	,	PUNCT
easat-3242	336	5	no	no	INTJ
easat-3242	336	6	.	.	NOUN
easat-3242	336	7	2	2	NUM
easat-3242	336	8	(	(	PUNCT
easat-3242	336	9	2022	2022	NUM
easat-3242	336	10	):	):	PUNCT
easat-3242	336	11	657	657	NUM
easat-3242	336	12	-	-	SYM
easat-3242	336	13	677	677	NUM
easat-3242	336	14	.	.	PUNCT