id	sid	tid	token	lemma	pos
fcis-10209	1	1	frontiers	frontier	NOUN
fcis-10209	1	2	in	in	ADP
fcis-10209	1	3	computing	computing	NOUN
fcis-10209	1	4	and	and	CCONJ
fcis-10209	1	5	intelligent	intelligent	ADJ
fcis-10209	1	6	systems	system	NOUN
fcis-10209	1	7	issn	issn	VERB
fcis-10209	1	8	:	:	PUNCT
fcis-10209	1	9	2832	2832	NUM
fcis-10209	1	10	-	-	SYM
fcis-10209	1	11	6024	6024	NUM
fcis-10209	1	12	|	|	NOUN
fcis-10209	1	13	vol	vol	NOUN
fcis-10209	1	14	.	.	PROPN
fcis-10209	2	1	4	4	NUM
fcis-10209	2	2	,	,	PUNCT
fcis-10209	2	3	no	no	INTJ
fcis-10209	2	4	.	.	NOUN
fcis-10209	2	5	2	2	NUM
fcis-10209	2	6	,	,	PUNCT
fcis-10209	2	7	2023	2023	NUM
fcis-10209	2	8	72	72	NUM
fcis-10209	2	9	weed	weed	NOUN
fcis-10209	2	10	recognition	recognition	NOUN
fcis-10209	2	11	method	method	NOUN
fcis-10209	2	12	based	base	VERB
fcis-10209	2	13	on	on	ADP
fcis-10209	2	14	hybrid	hybrid	ADJ
fcis-10209	2	15	cnn‐	cnn‐	NOUN
fcis-10209	2	16	transformer	transformer	NOUN
fcis-10209	2	17	model	model	NOUN
fcis-10209	2	18	jun	jun	PROPN
fcis-10209	2	19	zhang	zhang	PROPN
fcis-10209	2	20	*	*	PUNCT
fcis-10209	2	21	college	college	PROPN
fcis-10209	2	22	of	of	ADP
fcis-10209	2	23	software	software	NOUN
fcis-10209	2	24	engineering	engineering	NOUN
fcis-10209	2	25	,	,	PUNCT
fcis-10209	2	26	south	south	PROPN
fcis-10209	2	27	china	china	PROPN
fcis-10209	2	28	agriculture	agriculture	PROPN
fcis-10209	2	29	university	university	PROPN
fcis-10209	2	30	,	,	PUNCT
fcis-10209	2	31	guangzhou	guangzhou	PROPN
fcis-10209	2	32	,	,	PUNCT
fcis-10209	2	33	guangdong	guangdong	PROPN
fcis-10209	2	34	,	,	PUNCT
fcis-10209	2	35	510642	510642	NUM
fcis-10209	2	36	,	,	PUNCT
fcis-10209	2	37	china	china	PROPN
fcis-10209	2	38	*	*	PUNCT
fcis-10209	2	39	corresponding	correspond	VERB
fcis-10209	2	40	author	author	NOUN
fcis-10209	2	41	email	email	NOUN
fcis-10209	2	42	:	:	PUNCT
fcis-10209	3	1	opip47@163.com	opip47@163.com	X
fcis-10209	3	2	abstract	abstract	NOUN
fcis-10209	3	3	:	:	PUNCT
fcis-10209	3	4	as	as	ADP
fcis-10209	3	5	an	an	DET
fcis-10209	3	6	important	important	ADJ
fcis-10209	3	7	task	task	NOUN
fcis-10209	3	8	in	in	ADP
fcis-10209	3	9	precision	precision	NOUN
fcis-10209	3	10	agriculture	agriculture	NOUN
fcis-10209	3	11	,	,	PUNCT
fcis-10209	3	12	weed	weed	VERB
fcis-10209	3	13	recognition	recognition	NOUN
fcis-10209	3	14	plays	play	VERB
fcis-10209	3	15	a	a	DET
fcis-10209	3	16	crucial	crucial	ADJ
fcis-10209	3	17	role	role	NOUN
fcis-10209	3	18	in	in	ADP
fcis-10209	3	19	crop	crop	NOUN
fcis-10209	3	20	management	management	NOUN
fcis-10209	3	21	and	and	CCONJ
fcis-10209	3	22	yield	yield	NOUN
fcis-10209	3	23	increase	increase	NOUN
fcis-10209	3	24	.	.	PUNCT
fcis-10209	4	1	however	however	ADV
fcis-10209	4	2	,	,	PUNCT
fcis-10209	4	3	achieving	achieve	VERB
fcis-10209	4	4	high	high	ADJ
fcis-10209	4	5	accuracy	accuracy	NOUN
fcis-10209	4	6	and	and	CCONJ
fcis-10209	4	7	efficiency	efficiency	NOUN
fcis-10209	4	8	at	at	ADP
fcis-10209	4	9	the	the	DET
fcis-10209	4	10	same	same	ADJ
fcis-10209	4	11	time	time	NOUN
fcis-10209	4	12	remains	remain	VERB
fcis-10209	4	13	a	a	DET
fcis-10209	4	14	challenge	challenge	NOUN
fcis-10209	4	15	.	.	PUNCT
fcis-10209	5	1	to	to	PART
fcis-10209	5	2	address	address	VERB
fcis-10209	5	3	the	the	DET
fcis-10209	5	4	balance	balance	NOUN
fcis-10209	5	5	between	between	ADP
fcis-10209	5	6	accuracy	accuracy	NOUN
fcis-10209	5	7	and	and	CCONJ
fcis-10209	5	8	timeliness	timeliness	NOUN
fcis-10209	5	9	in	in	ADP
fcis-10209	5	10	weed	weed	NOUN
fcis-10209	5	11	recognition	recognition	NOUN
fcis-10209	5	12	,	,	PUNCT
fcis-10209	5	13	this	this	DET
fcis-10209	5	14	paper	paper	NOUN
fcis-10209	5	15	proposes	propose	VERB
fcis-10209	5	16	a	a	DET
fcis-10209	5	17	hybrid	hybrid	ADJ
fcis-10209	5	18	cnn	cnn	PROPN
fcis-10209	5	19	-	-	PUNCT
fcis-10209	5	20	transformer	transformer	NOUN
fcis-10209	5	21	model	model	NOUN
fcis-10209	5	22	for	for	ADP
fcis-10209	5	23	weed	weed	NOUN
fcis-10209	5	24	recognition	recognition	NOUN
fcis-10209	5	25	.	.	PUNCT
fcis-10209	6	1	the	the	DET
fcis-10209	6	2	model	model	NOUN
fcis-10209	6	3	uses	use	VERB
fcis-10209	6	4	a	a	DET
fcis-10209	6	5	combination	combination	NOUN
fcis-10209	6	6	of	of	ADP
fcis-10209	6	7	convolutional	convolutional	ADJ
fcis-10209	6	8	neural	neural	ADJ
fcis-10209	6	9	network	network	NOUN
fcis-10209	6	10	(	(	PUNCT
fcis-10209	6	11	cnn	cnn	PROPN
fcis-10209	6	12	)	)	PUNCT
fcis-10209	6	13	and	and	CCONJ
fcis-10209	6	14	transformer	transformer	NOUN
fcis-10209	6	15	structures	structure	NOUN
fcis-10209	6	16	for	for	ADP
fcis-10209	6	17	feature	feature	NOUN
fcis-10209	6	18	extraction	extraction	NOUN
fcis-10209	6	19	and	and	CCONJ
fcis-10209	6	20	classification	classification	NOUN
fcis-10209	6	21	,	,	PUNCT
fcis-10209	6	22	taking	take	VERB
fcis-10209	6	23	into	into	ADP
fcis-10209	6	24	account	account	NOUN
fcis-10209	6	25	both	both	CCONJ
fcis-10209	6	26	global	global	ADJ
fcis-10209	6	27	and	and	CCONJ
fcis-10209	6	28	local	local	ADJ
fcis-10209	6	29	information	information	NOUN
fcis-10209	6	30	.	.	PUNCT
fcis-10209	7	1	in	in	ADP
fcis-10209	7	2	addition	addition	NOUN
fcis-10209	7	3	,	,	PUNCT
fcis-10209	7	4	the	the	DET
fcis-10209	7	5	proposed	propose	VERB
fcis-10209	7	6	transformer	transformer	NOUN
fcis-10209	7	7	block	block	NOUN
fcis-10209	7	8	incorporates	incorporate	VERB
fcis-10209	7	9	the	the	DET
fcis-10209	7	10	sdta	sdta	PROPN
fcis-10209	7	11	(	(	PUNCT
fcis-10209	7	12	segmentation	segmentation	NOUN
fcis-10209	7	13	depth	depth	NOUN
fcis-10209	7	14	transpose	transpose	NOUN
fcis-10209	7	15	attention	attention	NOUN
fcis-10209	7	16	)	)	PUNCT
fcis-10209	7	17	mechanism	mechanism	NOUN
fcis-10209	7	18	to	to	PART
fcis-10209	7	19	improve	improve	VERB
fcis-10209	7	20	timeliness	timeliness	NOUN
fcis-10209	7	21	.	.	PUNCT
fcis-10209	8	1	furthermore	furthermore	ADV
fcis-10209	8	2	,	,	PUNCT
fcis-10209	8	3	this	this	DET
fcis-10209	8	4	paper	paper	NOUN
fcis-10209	8	5	improves	improve	VERB
fcis-10209	8	6	the	the	DET
fcis-10209	8	7	original	original	ADJ
fcis-10209	8	8	vit	vit	ADJ
fcis-10209	8	9	model	model	NOUN
fcis-10209	8	10	to	to	PART
fcis-10209	8	11	enhance	enhance	VERB
fcis-10209	8	12	its	its	PRON
fcis-10209	8	13	accuracy	accuracy	NOUN
fcis-10209	8	14	.	.	PUNCT
fcis-10209	9	1	experimental	experimental	ADJ
fcis-10209	9	2	results	result	NOUN
fcis-10209	9	3	on	on	ADP
fcis-10209	9	4	the	the	DET
fcis-10209	9	5	deep	deep	ADJ
fcis-10209	9	6	weeds	weed	NOUN
fcis-10209	9	7	dataset	dataset	VERB
fcis-10209	9	8	by	by	ADP
fcis-10209	9	9	olsen	olsen	PROPN
fcis-10209	9	10	et	et	PROPN
fcis-10209	9	11	al	al	PROPN
fcis-10209	9	12	.	.	PROPN
fcis-10209	9	13	show	show	VERB
fcis-10209	9	14	that	that	SCONJ
fcis-10209	9	15	the	the	DET
fcis-10209	9	16	proposed	propose	VERB
fcis-10209	9	17	hybrid	hybrid	NOUN
fcis-10209	9	18	model	model	NOUN
fcis-10209	9	19	outperforms	outperform	VERB
fcis-10209	9	20	the	the	DET
fcis-10209	9	21	original	original	ADJ
fcis-10209	9	22	vision	vision	NOUN
fcis-10209	9	23	transformer	transformer	NOUN
fcis-10209	9	24	model	model	NOUN
fcis-10209	9	25	in	in	ADP
fcis-10209	9	26	weed	weed	NOUN
fcis-10209	9	27	recognition	recognition	NOUN
fcis-10209	9	28	accuracy	accuracy	NOUN
fcis-10209	9	29	(	(	PUNCT
fcis-10209	9	30	89.43	89.43	NUM
fcis-10209	9	31	%	%	NOUN
fcis-10209	9	32	vs.	vs.	ADP
fcis-10209	9	33	96.08	96.08	NUM
fcis-10209	9	34	%	%	NOUN
fcis-10209	9	35	)	)	PUNCT
fcis-10209	9	36	.	.	PUNCT
fcis-10209	10	1	this	this	DET
fcis-10209	10	2	research	research	NOUN
fcis-10209	10	3	provides	provide	VERB
fcis-10209	10	4	an	an	DET
fcis-10209	10	5	effective	effective	ADJ
fcis-10209	10	6	solution	solution	NOUN
fcis-10209	10	7	for	for	ADP
fcis-10209	10	8	weed	weed	NOUN
fcis-10209	10	9	recognition	recognition	NOUN
fcis-10209	10	10	using	use	VERB
fcis-10209	10	11	a	a	DET
fcis-10209	10	12	hybrid	hybrid	ADJ
fcis-10209	10	13	model	model	NOUN
fcis-10209	10	14	,	,	PUNCT
fcis-10209	10	15	with	with	ADP
fcis-10209	10	16	high	high	ADJ
fcis-10209	10	17	practical	practical	ADJ
fcis-10209	10	18	value	value	NOUN
fcis-10209	10	19	and	and	CCONJ
fcis-10209	10	20	application	application	NOUN
fcis-10209	10	21	prospects	prospect	NOUN
fcis-10209	10	22	.	.	PUNCT
fcis-10209	11	1	keywords	keyword	NOUN
fcis-10209	11	2	:	:	PUNCT
fcis-10209	11	3	weed	weed	NOUN
fcis-10209	11	4	recognition	recognition	NOUN
fcis-10209	11	5	;	;	PUNCT
fcis-10209	11	6	vision	vision	NOUN
fcis-10209	11	7	transformer	transformer	NOUN
fcis-10209	11	8	;	;	PUNCT
fcis-10209	11	9	hybrid	hybrid	ADJ
fcis-10209	11	10	cnn	cnn	PROPN
fcis-10209	11	11	-	-	PUNCT
fcis-10209	11	12	transformer	transformer	NOUN
fcis-10209	11	13	model	model	NOUN
fcis-10209	11	14	;	;	PUNCT
fcis-10209	11	15	sdta	sdta	PROPN
fcis-10209	11	16	.	.	PROPN
fcis-10209	11	17	1	1	X
fcis-10209	11	18	.	.	X
fcis-10209	11	19	introduction	introduction	NOUN
fcis-10209	11	20	with	with	ADP
fcis-10209	11	21	the	the	DET
fcis-10209	11	22	rapid	rapid	ADJ
fcis-10209	11	23	development	development	NOUN
fcis-10209	11	24	of	of	ADP
fcis-10209	11	25	global	global	ADJ
fcis-10209	11	26	agriculture	agriculture	NOUN
fcis-10209	11	27	,	,	PUNCT
fcis-10209	11	28	weed	weed	NOUN
fcis-10209	11	29	problem	problem	NOUN
fcis-10209	11	30	has	have	AUX
fcis-10209	11	31	become	become	VERB
fcis-10209	11	32	one	one	NUM
fcis-10209	11	33	of	of	ADP
fcis-10209	11	34	the	the	DET
fcis-10209	11	35	urgent	urgent	ADJ
fcis-10209	11	36	issues	issue	NOUN
fcis-10209	11	37	to	to	PART
fcis-10209	11	38	be	be	AUX
fcis-10209	11	39	solved	solve	VERB
fcis-10209	11	40	in	in	ADP
fcis-10209	11	41	agricultural	agricultural	ADJ
fcis-10209	11	42	production[1	production[1	PROPN
fcis-10209	11	43	]	]	PUNCT
fcis-10209	11	44	.	.	PUNCT
fcis-10209	12	1	weeds	weed	NOUN
fcis-10209	12	2	grow	grow	VERB
fcis-10209	12	3	profusely	profusely	ADV
fcis-10209	12	4	and	and	CCONJ
fcis-10209	12	5	compete	compete	VERB
fcis-10209	12	6	with	with	ADP
fcis-10209	12	7	crops	crop	NOUN
fcis-10209	12	8	for	for	ADP
fcis-10209	12	9	nutrients	nutrient	NOUN
fcis-10209	12	10	,	,	PUNCT
fcis-10209	12	11	water	water	NOUN
fcis-10209	12	12	and	and	CCONJ
fcis-10209	12	13	sunlight	sunlight	NOUN
fcis-10209	12	14	,	,	PUNCT
fcis-10209	12	15	which	which	PRON
fcis-10209	12	16	not	not	PART
fcis-10209	12	17	only	only	ADV
fcis-10209	12	18	severely	severely	ADV
fcis-10209	12	19	affects	affect	VERB
fcis-10209	12	20	the	the	DET
fcis-10209	12	21	growth	growth	NOUN
fcis-10209	12	22	of	of	ADP
fcis-10209	12	23	crops	crop	NOUN
fcis-10209	12	24	,	,	PUNCT
fcis-10209	12	25	but	but	CCONJ
fcis-10209	12	26	also	also	ADV
fcis-10209	12	27	significantly	significantly	ADV
fcis-10209	12	28	reduces	reduce	VERB
fcis-10209	12	29	crop	crop	NOUN
fcis-10209	12	30	yields[2	yields[2	PROPN
fcis-10209	12	31	]	]	X
fcis-10209	12	32	.	.	PUNCT
fcis-10209	13	1	in	in	ADP
fcis-10209	13	2	order	order	NOUN
fcis-10209	13	3	to	to	PART
fcis-10209	13	4	achieve	achieve	VERB
fcis-10209	13	5	the	the	DET
fcis-10209	13	6	goal	goal	NOUN
fcis-10209	13	7	of	of	ADP
fcis-10209	13	8	weeding	weed	VERB
fcis-10209	13	9	,	,	PUNCT
fcis-10209	13	10	people	people	NOUN
fcis-10209	13	11	began	begin	VERB
fcis-10209	13	12	to	to	PART
fcis-10209	13	13	spray	spray	VERB
fcis-10209	13	14	large	large	ADJ
fcis-10209	13	15	amounts	amount	NOUN
fcis-10209	13	16	of	of	ADP
fcis-10209	13	17	herbicides	herbicide	NOUN
fcis-10209	13	18	for	for	ADP
fcis-10209	13	19	weeding	weed	VERB
fcis-10209	13	20	,	,	PUNCT
fcis-10209	13	21	but	but	CCONJ
fcis-10209	13	22	this	this	DET
fcis-10209	13	23	method	method	NOUN
fcis-10209	13	24	is	be	AUX
fcis-10209	13	25	prone	prone	ADJ
fcis-10209	13	26	to	to	ADP
fcis-10209	13	27	environmental	environmental	ADJ
fcis-10209	13	28	pollution	pollution	NOUN
fcis-10209	13	29	and	and	CCONJ
fcis-10209	13	30	hazards	hazard	NOUN
fcis-10209	13	31	.	.	PUNCT
fcis-10209	14	1	therefore	therefore	ADV
fcis-10209	14	2	,	,	PUNCT
fcis-10209	14	3	in	in	ADP
fcis-10209	14	4	order	order	NOUN
fcis-10209	14	5	to	to	PART
fcis-10209	14	6	improve	improve	VERB
fcis-10209	14	7	agricultural	agricultural	ADJ
fcis-10209	14	8	production	production	NOUN
fcis-10209	14	9	efficiency	efficiency	NOUN
fcis-10209	14	10	,	,	PUNCT
fcis-10209	14	11	reduce	reduce	VERB
fcis-10209	14	12	environmental	environmental	ADJ
fcis-10209	14	13	pollution	pollution	NOUN
fcis-10209	14	14	and	and	CCONJ
fcis-10209	14	15	hazards	hazard	NOUN
fcis-10209	14	16	,	,	PUNCT
fcis-10209	14	17	an	an	DET
fcis-10209	14	18	accurate	accurate	ADJ
fcis-10209	14	19	and	and	CCONJ
fcis-10209	14	20	efficient	efficient	ADJ
fcis-10209	14	21	method	method	NOUN
fcis-10209	14	22	to	to	PART
fcis-10209	14	23	identify	identify	VERB
fcis-10209	14	24	and	and	CCONJ
fcis-10209	14	25	classify	classify	VERB
fcis-10209	14	26	weeds	weed	NOUN
fcis-10209	14	27	is	be	AUX
fcis-10209	14	28	needed[3	needed[3	NUM
fcis-10209	14	29	]	]	PUNCT
fcis-10209	14	30	.	.	PUNCT
fcis-10209	15	1	weed	weed	NOUN
fcis-10209	15	2	identification	identification	NOUN
fcis-10209	15	3	and	and	CCONJ
fcis-10209	15	4	management	management	NOUN
fcis-10209	15	5	has	have	AUX
fcis-10209	15	6	become	become	VERB
fcis-10209	15	7	a	a	DET
fcis-10209	15	8	frontier	frontier	NOUN
fcis-10209	15	9	and	and	CCONJ
fcis-10209	15	10	hotspot	hotspot	NOUN
fcis-10209	15	11	issue	issue	NOUN
fcis-10209	15	12	in	in	ADP
fcis-10209	15	13	the	the	DET
fcis-10209	15	14	current	current	ADJ
fcis-10209	15	15	research	research	NOUN
fcis-10209	15	16	field	field	NOUN
fcis-10209	15	17	of	of	ADP
fcis-10209	15	18	weed	weed	NOUN
fcis-10209	15	19	control	control	NOUN
fcis-10209	15	20	,	,	PUNCT
fcis-10209	15	21	and	and	CCONJ
fcis-10209	15	22	the	the	DET
fcis-10209	15	23	development	development	NOUN
fcis-10209	15	24	of	of	ADP
fcis-10209	15	25	image	image	NOUN
fcis-10209	15	26	processing	processing	NOUN
fcis-10209	15	27	technology	technology	NOUN
fcis-10209	15	28	and	and	CCONJ
fcis-10209	15	29	artificial	artificial	ADJ
fcis-10209	15	30	intelligence	intelligence	NOUN
fcis-10209	15	31	technology	technology	NOUN
fcis-10209	15	32	provides	provide	VERB
fcis-10209	15	33	new	new	ADJ
fcis-10209	15	34	ideas	idea	NOUN
fcis-10209	15	35	and	and	CCONJ
fcis-10209	15	36	possibilities	possibility	NOUN
fcis-10209	15	37	for	for	ADP
fcis-10209	15	38	solving	solve	VERB
fcis-10209	15	39	this	this	PRON
fcis-10209	15	40	problem[4	problem[4	PROPN
fcis-10209	15	41	]	]	PUNCT
fcis-10209	15	42	.	.	PUNCT
fcis-10209	16	1	the	the	DET
fcis-10209	16	2	current	current	ADJ
fcis-10209	16	3	weed	weed	NOUN
fcis-10209	16	4	recognition	recognition	NOUN
fcis-10209	16	5	technology	technology	NOUN
fcis-10209	16	6	can	can	AUX
fcis-10209	16	7	be	be	AUX
fcis-10209	16	8	primarily	primarily	ADV
fcis-10209	16	9	divided	divide	VERB
fcis-10209	16	10	into	into	ADP
fcis-10209	16	11	two	two	NUM
fcis-10209	16	12	categories	category	NOUN
fcis-10209	16	13	in	in	ADP
fcis-10209	16	14	academic	academic	ADJ
fcis-10209	16	15	research	research	NOUN
fcis-10209	16	16	:	:	PUNCT
fcis-10209	16	17	machine	machine	NOUN
fcis-10209	16	18	learning	learn	VERB
fcis-10209	16	19	methods	method	NOUN
fcis-10209	16	20	based	base	VERB
fcis-10209	16	21	on	on	ADP
fcis-10209	16	22	traditional	traditional	ADJ
fcis-10209	16	23	manual	manual	ADJ
fcis-10209	16	24	feature	feature	NOUN
fcis-10209	16	25	extraction	extraction	NOUN
fcis-10209	16	26	,	,	PUNCT
fcis-10209	16	27	and	and	CCONJ
fcis-10209	16	28	deep	deep	ADJ
fcis-10209	16	29	learning	learning	NOUN
fcis-10209	16	30	models	model	NOUN
fcis-10209	16	31	based	base	VERB
fcis-10209	16	32	on	on	ADP
fcis-10209	16	33	automatic	automatic	ADJ
fcis-10209	16	34	feature	feature	NOUN
fcis-10209	16	35	extraction	extraction	NOUN
fcis-10209	16	36	.	.	PUNCT
fcis-10209	17	1	there	there	PRON
fcis-10209	17	2	algorithms	algorithm	VERB
fcis-10209	17	3	not	not	PART
fcis-10209	17	4	only	only	ADV
fcis-10209	17	5	have	have	AUX
fcis-10209	17	6	the	the	DET
fcis-10209	17	7	ability	ability	NOUN
fcis-10209	17	8	to	to	PART
fcis-10209	17	9	automatically	automatically	ADV
fcis-10209	17	10	extract	extract	VERB
fcis-10209	17	11	features	feature	NOUN
fcis-10209	17	12	from	from	ADP
fcis-10209	17	13	weed	weed	NOUN
fcis-10209	17	14	images	image	NOUN
fcis-10209	17	15	,	,	PUNCT
fcis-10209	17	16	but	but	CCONJ
fcis-10209	17	17	also	also	ADV
fcis-10209	17	18	achieve	achieve	VERB
fcis-10209	17	19	higher	high	ADJ
fcis-10209	17	20	recognition	recognition	NOUN
fcis-10209	17	21	accuracy	accuracy	NOUN
fcis-10209	17	22	and	and	CCONJ
fcis-10209	17	23	real	real	ADJ
fcis-10209	17	24	-	-	PUNCT
fcis-10209	17	25	time	time	NOUN
fcis-10209	17	26	performance	performance	NOUN
fcis-10209	17	27	.	.	PUNCT
fcis-10209	18	1	in	in	ADP
fcis-10209	18	2	current	current	ADJ
fcis-10209	18	3	weed	weed	NOUN
fcis-10209	18	4	identification	identification	NOUN
fcis-10209	18	5	technology	technology	NOUN
fcis-10209	18	6	,	,	PUNCT
fcis-10209	18	7	the	the	DET
fcis-10209	18	8	application	application	NOUN
fcis-10209	18	9	of	of	ADP
fcis-10209	18	10	machine	machine	NOUN
fcis-10209	18	11	learning	learning	NOUN
fcis-10209	18	12	and	and	CCONJ
fcis-10209	18	13	deep	deep	ADJ
fcis-10209	18	14	learning	learning	NOUN
fcis-10209	18	15	models	model	NOUN
fcis-10209	18	16	based	base	VERB
fcis-10209	18	17	on	on	ADP
fcis-10209	18	18	traditional	traditional	ADJ
fcis-10209	18	19	manual	manual	ADJ
fcis-10209	18	20	feature	feature	NOUN
fcis-10209	18	21	extraction	extraction	NOUN
fcis-10209	18	22	is	be	AUX
fcis-10209	18	23	becoming	become	VERB
fcis-10209	18	24	more	more	ADV
fcis-10209	18	25	and	and	CCONJ
fcis-10209	18	26	more	more	ADV
fcis-10209	18	27	popular	popular	ADJ
fcis-10209	18	28	.	.	PUNCT
fcis-10209	19	1	these	these	DET
fcis-10209	19	2	algorithms	algorithm	NOUN
fcis-10209	19	3	can	can	AUX
fcis-10209	19	4	not	not	PART
fcis-10209	19	5	only	only	ADV
fcis-10209	19	6	automatically	automatically	ADV
fcis-10209	19	7	extract	extract	VERB
fcis-10209	19	8	features	feature	NOUN
fcis-10209	19	9	from	from	ADP
fcis-10209	19	10	weed	weed	NOUN
fcis-10209	19	11	images	image	NOUN
fcis-10209	19	12	,	,	PUNCT
fcis-10209	19	13	but	but	CCONJ
fcis-10209	19	14	also	also	ADV
fcis-10209	19	15	have	have	VERB
fcis-10209	19	16	higher	high	ADJ
fcis-10209	19	17	recognition	recognition	NOUN
fcis-10209	19	18	accuracy	accuracy	NOUN
fcis-10209	19	19	and	and	CCONJ
fcis-10209	19	20	real	real	ADJ
fcis-10209	19	21	-	-	PUNCT
fcis-10209	19	22	time	time	NOUN
fcis-10209	19	23	performance[5	performance[5	PROPN
fcis-10209	19	24	]	]	PUNCT
fcis-10209	19	25	.	.	PUNCT
fcis-10209	20	1	method	method	PROPN
fcis-10209	20	2	based	base	VERB
fcis-10209	20	3	on	on	ADP
fcis-10209	20	4	manual	manual	ADJ
fcis-10209	20	5	feature	feature	NOUN
fcis-10209	20	6	extractor	extractor	NOUN
fcis-10209	20	7	and	and	CCONJ
fcis-10209	20	8	classifier	classifier	NOUN
fcis-10209	20	9	:	:	PUNCT
fcis-10209	20	10	this	this	DET
fcis-10209	20	11	method	method	NOUN
fcis-10209	20	12	usually	usually	ADV
fcis-10209	20	13	requires	require	VERB
fcis-10209	20	14	manual	manual	ADJ
fcis-10209	20	15	design	design	NOUN
fcis-10209	20	16	and	and	CCONJ
fcis-10209	20	17	extraction	extraction	NOUN
fcis-10209	20	18	of	of	ADP
fcis-10209	20	19	features	feature	NOUN
fcis-10209	20	20	related	relate	VERB
fcis-10209	20	21	to	to	ADP
fcis-10209	20	22	weeds	weed	NOUN
fcis-10209	20	23	,	,	PUNCT
fcis-10209	20	24	and	and	CCONJ
fcis-10209	20	25	uses	use	VERB
fcis-10209	20	26	classifiers	classifier	NOUN
fcis-10209	20	27	for	for	ADP
fcis-10209	20	28	classification	classification	NOUN
fcis-10209	20	29	,	,	PUNCT
fcis-10209	20	30	such	such	ADJ
fcis-10209	20	31	as	as	ADP
fcis-10209	20	32	support	support	NOUN
fcis-10209	20	33	vector	vector	NOUN
fcis-10209	20	34	machine	machine	NOUN
fcis-10209	20	35	(	(	PUNCT
fcis-10209	20	36	svm	svm	PROPN
fcis-10209	20	37	)	)	PUNCT
fcis-10209	20	38	,	,	PUNCT
fcis-10209	20	39	random	random	ADJ
fcis-10209	20	40	forest	forest	NOUN
fcis-10209	20	41	,	,	PUNCT
fcis-10209	20	42	etc	etc	X
fcis-10209	20	43	.	.	X
fcis-10209	21	1	[	[	X
fcis-10209	21	2	4].cristóbal[6	4].cristóbal[6	X
fcis-10209	21	3	]	]	X
fcis-10209	21	4	et	et	PROPN
fcis-10209	21	5	al	al	PROPN
fcis-10209	21	6	.	.	PROPN
fcis-10209	21	7	used	use	VERB
fcis-10209	21	8	hough	hough	PROPN
fcis-10209	21	9	transform	transform	VERB
fcis-10209	21	10	and	and	CCONJ
fcis-10209	21	11	svm	svm	VERB
fcis-10209	21	12	classifier	classifier	NOUN
fcis-10209	21	13	to	to	PART
fcis-10209	21	14	separate	separate	VERB
fcis-10209	21	15	weeds	weed	NOUN
fcis-10209	21	16	in	in	ADP
fcis-10209	21	17	corn	corn	NOUN
fcis-10209	21	18	fields	field	NOUN
fcis-10209	21	19	,	,	PUNCT
fcis-10209	21	20	achieving	achieve	VERB
fcis-10209	21	21	good	good	ADJ
fcis-10209	21	22	results	result	NOUN
fcis-10209	21	23	.	.	PUNCT
fcis-10209	22	1	lottes[7	lottes[7	VERB
fcis-10209	22	2	]	]	X
fcis-10209	22	3	et	et	PROPN
fcis-10209	22	4	al	al	PROPN
fcis-10209	22	5	.	.	PROPN
fcis-10209	22	6	proposed	propose	VERB
fcis-10209	22	7	a	a	DET
fcis-10209	22	8	weed	weed	NOUN
fcis-10209	22	9	species	species	NOUN
fcis-10209	22	10	detection	detection	NOUN
fcis-10209	22	11	method	method	NOUN
fcis-10209	22	12	based	base	VERB
fcis-10209	22	13	on	on	ADP
fcis-10209	22	14	dense	dense	ADJ
fcis-10209	22	15	stereo	stereo	NOUN
fcis-10209	22	16	vision	vision	NOUN
fcis-10209	22	17	,	,	PUNCT
fcis-10209	22	18	comparing	compare	VERB
fcis-10209	22	19	the	the	DET
fcis-10209	22	20	depth	depth	NOUN
fcis-10209	22	21	value	value	NOUN
fcis-10209	22	22	with	with	ADP
fcis-10209	22	23	manually	manually	ADV
fcis-10209	22	24	selected	select	VERB
fcis-10209	22	25	weed	weed	NOUN
fcis-10209	22	26	height	height	NOUN
fcis-10209	22	27	lines	line	NOUN
fcis-10209	22	28	to	to	PART
fcis-10209	22	29	identify	identify	VERB
fcis-10209	22	30	the	the	DET
fcis-10209	22	31	presence	presence	NOUN
fcis-10209	22	32	and	and	CCONJ
fcis-10209	22	33	species	specie	NOUN
fcis-10209	22	34	of	of	ADP
fcis-10209	22	35	weeds	weed	NOUN
fcis-10209	22	36	.	.	PUNCT
fcis-10209	23	1	zhang[7	zhang[7	X
fcis-10209	23	2	]	]	X
fcis-10209	23	3	et	et	PROPN
fcis-10209	23	4	al	al	PROPN
fcis-10209	23	5	.	.	PROPN
fcis-10209	23	6	preprocessed	preprocesse	VERB
fcis-10209	23	7	weed	weed	NOUN
fcis-10209	23	8	images	image	NOUN
fcis-10209	23	9	using	use	VERB
fcis-10209	23	10	morphological	morphological	ADJ
fcis-10209	23	11	filtering	filtering	NOUN
fcis-10209	23	12	and	and	CCONJ
fcis-10209	23	13	threshold	threshold	NOUN
fcis-10209	23	14	segmentation	segmentation	NOUN
fcis-10209	23	15	,	,	PUNCT
fcis-10209	23	16	extracted	extract	VERB
fcis-10209	23	17	a	a	DET
fcis-10209	23	18	series	series	NOUN
fcis-10209	23	19	of	of	ADP
fcis-10209	23	20	features	feature	NOUN
fcis-10209	23	21	,	,	PUNCT
fcis-10209	23	22	and	and	CCONJ
fcis-10209	23	23	trained	train	VERB
fcis-10209	23	24	a	a	DET
fcis-10209	23	25	random	random	ADJ
fcis-10209	23	26	forest	forest	NOUN
fcis-10209	23	27	classifier	classifier	NOUN
fcis-10209	23	28	to	to	PART
fcis-10209	23	29	classify	classify	VERB
fcis-10209	23	30	different	different	ADJ
fcis-10209	23	31	types	type	NOUN
fcis-10209	23	32	of	of	ADP
fcis-10209	23	33	weeds	weed	NOUN
fcis-10209	23	34	.	.	PUNCT
fcis-10209	24	1	however	however	ADV
fcis-10209	24	2	,	,	PUNCT
fcis-10209	24	3	this	this	DET
fcis-10209	24	4	method	method	NOUN
fcis-10209	24	5	has	have	VERB
fcis-10209	24	6	some	some	DET
fcis-10209	24	7	drawbacks	drawback	NOUN
fcis-10209	24	8	.	.	PUNCT
fcis-10209	25	1	firstly	firstly	ADV
fcis-10209	25	2	,	,	PUNCT
fcis-10209	25	3	this	this	DET
fcis-10209	25	4	method	method	NOUN
fcis-10209	25	5	requires	require	VERB
fcis-10209	25	6	manual	manual	ADJ
fcis-10209	25	7	design	design	NOUN
fcis-10209	25	8	and	and	CCONJ
fcis-10209	25	9	extraction	extraction	NOUN
fcis-10209	25	10	of	of	ADP
fcis-10209	25	11	features	feature	NOUN
fcis-10209	25	12	related	relate	VERB
fcis-10209	25	13	to	to	ADP
fcis-10209	25	14	weeds	weed	NOUN
fcis-10209	25	15	,	,	PUNCT
fcis-10209	25	16	which	which	PRON
fcis-10209	25	17	may	may	AUX
fcis-10209	25	18	require	require	VERB
fcis-10209	25	19	expert	expert	ADJ
fcis-10209	25	20	experience	experience	NOUN
fcis-10209	25	21	and	and	CCONJ
fcis-10209	25	22	sample	sample	NOUN
fcis-10209	25	23	data	datum	NOUN
fcis-10209	25	24	,	,	PUNCT
fcis-10209	25	25	and	and	CCONJ
fcis-10209	25	26	important	important	ADJ
fcis-10209	25	27	information	information	NOUN
fcis-10209	25	28	may	may	AUX
fcis-10209	25	29	be	be	AUX
fcis-10209	25	30	lost	lose	VERB
fcis-10209	25	31	in	in	ADP
fcis-10209	25	32	the	the	DET
fcis-10209	25	33	feature	feature	NOUN
fcis-10209	25	34	extraction	extraction	NOUN
fcis-10209	25	35	process	process	NOUN
fcis-10209	25	36	.	.	PUNCT
fcis-10209	26	1	secondly	secondly	ADV
fcis-10209	26	2	,	,	PUNCT
fcis-10209	26	3	the	the	DET
fcis-10209	26	4	performance	performance	NOUN
fcis-10209	26	5	and	and	CCONJ
fcis-10209	26	6	accuracy	accuracy	NOUN
fcis-10209	26	7	of	of	ADP
fcis-10209	26	8	the	the	DET
fcis-10209	26	9	classifier	classifier	NOUN
fcis-10209	26	10	mainly	mainly	ADV
fcis-10209	26	11	depend	depend	VERB
fcis-10209	26	12	on	on	ADP
fcis-10209	26	13	the	the	DET
fcis-10209	26	14	selected	select	VERB
fcis-10209	26	15	features	feature	NOUN
fcis-10209	26	16	.	.	PUNCT
fcis-10209	27	1	if	if	SCONJ
fcis-10209	27	2	the	the	DET
fcis-10209	27	3	features	feature	NOUN
fcis-10209	27	4	are	be	AUX
fcis-10209	27	5	not	not	PART
fcis-10209	27	6	representative	representative	ADJ
fcis-10209	27	7	,	,	PUNCT
fcis-10209	27	8	the	the	DET
fcis-10209	27	9	recognition	recognition	NOUN
fcis-10209	27	10	effect	effect	NOUN
fcis-10209	27	11	and	and	CCONJ
fcis-10209	27	12	accuracy	accuracy	NOUN
fcis-10209	27	13	may	may	AUX
fcis-10209	27	14	be	be	AUX
fcis-10209	27	15	affected	affect	VERB
fcis-10209	27	16	.	.	PUNCT
fcis-10209	28	1	finally	finally	ADV
fcis-10209	28	2	,	,	PUNCT
fcis-10209	28	3	compared	compare	VERB
fcis-10209	28	4	with	with	ADP
fcis-10209	28	5	using	use	VERB
fcis-10209	28	6	deep	deep	ADJ
fcis-10209	28	7	learning	learning	NOUN
fcis-10209	28	8	methods	method	NOUN
fcis-10209	28	9	for	for	ADP
fcis-10209	28	10	weed	weed	NOUN
fcis-10209	28	11	identification	identification	NOUN
fcis-10209	28	12	,	,	PUNCT
fcis-10209	28	13	this	this	DET
fcis-10209	28	14	method	method	NOUN
fcis-10209	28	15	requires	require	VERB
fcis-10209	28	16	more	more	ADJ
fcis-10209	28	17	manual	manual	ADJ
fcis-10209	28	18	operations	operation	NOUN
fcis-10209	28	19	and	and	CCONJ
fcis-10209	28	20	time	time	NOUN
fcis-10209	28	21	-	-	PUNCT
fcis-10209	28	22	consuming[3	consuming[3	NOUN
fcis-10209	28	23	]	]	PUNCT
fcis-10209	28	24	.	.	PUNCT
fcis-10209	29	1	method	method	PROPN
fcis-10209	29	2	based	base	VERB
fcis-10209	29	3	on	on	ADP
fcis-10209	29	4	deep	deep	ADJ
fcis-10209	29	5	learning	learning	NOUN
fcis-10209	29	6	:	:	PUNCT
fcis-10209	29	7	this	this	PRON
fcis-10209	29	8	is	be	AUX
fcis-10209	29	9	a	a	DET
fcis-10209	29	10	method	method	NOUN
fcis-10209	29	11	that	that	PRON
fcis-10209	29	12	uses	use	VERB
fcis-10209	29	13	deep	deep	ADJ
fcis-10209	29	14	learning	learning	NOUN
fcis-10209	29	15	technology	technology	NOUN
fcis-10209	29	16	for	for	ADP
fcis-10209	29	17	object	object	NOUN
fcis-10209	29	18	extraction	extraction	NOUN
fcis-10209	29	19	,	,	PUNCT
fcis-10209	29	20	feature	feature	NOUN
fcis-10209	29	21	extraction	extraction	NOUN
fcis-10209	29	22	,	,	PUNCT
fcis-10209	29	23	and	and	CCONJ
fcis-10209	29	24	classification[8	classification[8	PRON
fcis-10209	29	25	]	]	PUNCT
fcis-10209	29	26	.	.	PUNCT
fcis-10209	30	1	due	due	ADP
fcis-10209	30	2	to	to	ADP
fcis-10209	30	3	the	the	DET
fcis-10209	30	4	characteristics	characteristic	NOUN
fcis-10209	30	5	of	of	ADP
fcis-10209	30	6	weed	weed	NOUN
fcis-10209	30	7	data	datum	NOUN
fcis-10209	30	8	,	,	PUNCT
fcis-10209	30	9	the	the	DET
fcis-10209	30	10	dataset	dataset	NOUN
fcis-10209	30	11	needs	need	VERB
fcis-10209	30	12	to	to	PART
fcis-10209	30	13	be	be	AUX
fcis-10209	30	14	preprocessed	preprocesse	VERB
fcis-10209	30	15	to	to	PART
fcis-10209	30	16	remove	remove	VERB
fcis-10209	30	17	the	the	DET
fcis-10209	30	18	background	background	NOUN
fcis-10209	30	19	from	from	ADP
fcis-10209	30	20	the	the	DET
fcis-10209	30	21	image	image	NOUN
fcis-10209	30	22	and	and	CCONJ
fcis-10209	30	23	only	only	ADV
fcis-10209	30	24	retain	retain	VERB
fcis-10209	30	25	the	the	DET
fcis-10209	30	26	weed	weed	NOUN
fcis-10209	30	27	body	body	NOUN
fcis-10209	30	28	to	to	PART
fcis-10209	30	29	improve	improve	VERB
fcis-10209	30	30	the	the	DET
fcis-10209	30	31	performance	performance	NOUN
fcis-10209	30	32	of	of	ADP
fcis-10209	30	33	the	the	DET
fcis-10209	30	34	subsequent	subsequent	ADJ
fcis-10209	30	35	model[9	model[9	NOUN
fcis-10209	30	36	]	]	PUNCT
fcis-10209	30	37	.	.	PUNCT
fcis-10209	31	1	common	common	ADJ
fcis-10209	31	2	preprocessing	preprocessing	NOUN
fcis-10209	31	3	methods	method	NOUN
fcis-10209	31	4	include	include	VERB
fcis-10209	31	5	the	the	DET
fcis-10209	31	6	supergreen	supergreen	PROPN
fcis-10209	31	7	algorithm[10	algorithm[10	PROPN
fcis-10209	31	8	]	]	PUNCT
fcis-10209	31	9	and	and	CCONJ
fcis-10209	31	10	otsu	otsu	ADJ
fcis-10209	31	11	threshold	threshold	NOUN
fcis-10209	31	12	segmentation	segmentation	NOUN
fcis-10209	31	13	algorithm[10	algorithm[10	ADV
fcis-10209	31	14	]	]	PUNCT
fcis-10209	31	15	.	.	PUNCT
fcis-10209	32	1	in	in	ADP
fcis-10209	32	2	terms	term	NOUN
fcis-10209	32	3	of	of	ADP
fcis-10209	32	4	feature	feature	NOUN
fcis-10209	32	5	extraction	extraction	NOUN
fcis-10209	32	6	,	,	PUNCT
fcis-10209	32	7	convolutional	convolutional	ADJ
fcis-10209	32	8	neural	neural	ADJ
fcis-10209	32	9	network	network	NOUN
fcis-10209	32	10	(	(	PUNCT
fcis-10209	32	11	cnn	cnn	PROPN
fcis-10209	32	12	)	)	PUNCT
fcis-10209	32	13	is	be	AUX
fcis-10209	32	14	one	one	NUM
fcis-10209	32	15	of	of	ADP
fcis-10209	32	16	the	the	DET
fcis-10209	32	17	most	most	ADV
fcis-10209	32	18	widely	widely	ADV
fcis-10209	32	19	used	use	VERB
fcis-10209	32	20	models	model	NOUN
fcis-10209	32	21	.	.	PUNCT
fcis-10209	33	1	it	it	PRON
fcis-10209	33	2	can	can	AUX
fcis-10209	33	3	automatically	automatically	ADV
fcis-10209	33	4	learn	learn	VERB
fcis-10209	33	5	the	the	DET
fcis-10209	33	6	most	most	ADV
fcis-10209	33	7	representative	representative	ADJ
fcis-10209	33	8	features	feature	NOUN
fcis-10209	33	9	from	from	ADP
fcis-10209	33	10	images	image	NOUN
fcis-10209	33	11	and	and	CCONJ
fcis-10209	33	12	generate	generate	VERB
fcis-10209	33	13	a	a	DET
fcis-10209	33	14	set	set	NOUN
fcis-10209	33	15	of	of	ADP
fcis-10209	33	16	high	high	ADJ
fcis-10209	33	17	-	-	PUNCT
fcis-10209	33	18	level	level	NOUN
fcis-10209	33	19	abstract	abstract	ADJ
fcis-10209	33	20	feature	feature	NOUN
fcis-10209	33	21	representations[10	representations[10	PROPN
fcis-10209	33	22	]	]	PUNCT
fcis-10209	33	23	.	.	PUNCT
fcis-10209	34	1	based	base	VERB
fcis-10209	34	2	on	on	ADP
fcis-10209	34	3	these	these	DET
fcis-10209	34	4	features	feature	NOUN
fcis-10209	34	5	,	,	PUNCT
fcis-10209	34	6	various	various	ADJ
fcis-10209	34	7	classifiers	classifier	NOUN
fcis-10209	34	8	can	can	AUX
fcis-10209	34	9	be	be	AUX
fcis-10209	34	10	used	use	VERB
fcis-10209	34	11	for	for	ADP
fcis-10209	34	12	classification	classification	NOUN
fcis-10209	34	13	,	,	PUNCT
fcis-10209	34	14	such	such	ADJ
fcis-10209	34	15	as	as	ADP
fcis-10209	34	16	support	support	NOUN
fcis-10209	34	17	vector	vector	NOUN
fcis-10209	34	18	machine	machine	NOUN
fcis-10209	34	19	(	(	PUNCT
fcis-10209	34	20	svm	svm	PROPN
fcis-10209	34	21	)	)	PUNCT
fcis-10209	34	22	,	,	PUNCT
fcis-10209	34	23	random	random	ADJ
fcis-10209	34	24	forest	forest	NOUN
fcis-10209	34	25	,	,	PUNCT
fcis-10209	34	26	etc	etc	X
fcis-10209	34	27	.	.	X
fcis-10209	35	1	in	in	ADP
fcis-10209	35	2	recent	recent	ADJ
fcis-10209	35	3	years	year	NOUN
fcis-10209	35	4	,	,	PUNCT
fcis-10209	35	5	many	many	ADJ
fcis-10209	35	6	new	new	ADJ
fcis-10209	35	7	deep	deep	ADJ
fcis-10209	35	8	learning	learning	NOUN
fcis-10209	35	9	models	model	NOUN
fcis-10209	35	10	and	and	CCONJ
fcis-10209	35	11	methods	method	NOUN
fcis-10209	35	12	have	have	AUX
fcis-10209	35	13	emerged	emerge	VERB
fcis-10209	35	14	.	.	PUNCT
fcis-10209	36	1	shah[11	shah[11	VERB
fcis-10209	36	2	]	]	PUNCT
fcis-10209	36	3	et	et	PROPN
fcis-10209	36	4	al	al	PROPN
fcis-10209	36	5	.	.	PROPN
fcis-10209	36	6	proposed	propose	VERB
fcis-10209	36	7	an	an	DET
fcis-10209	36	8	image	image	NOUN
fcis-10209	36	9	classification	classification	NOUN
fcis-10209	36	10	algorithm	algorithm	NOUN
fcis-10209	36	11	mainly	mainly	ADV
fcis-10209	36	12	targeting	target	VERB
fcis-10209	36	13	pests	pest	NOUN
fcis-10209	36	14	in	in	ADP
fcis-10209	36	15	soybeans	soybean	NOUN
fcis-10209	36	16	,	,	PUNCT
fcis-10209	36	17	which	which	PRON
fcis-10209	36	18	was	be	AUX
fcis-10209	36	19	trained	train	VERB
fcis-10209	36	20	and	and	CCONJ
fcis-10209	36	21	classified	classify	VERB
fcis-10209	36	22	using	use	VERB
fcis-10209	36	23	resnet-50	resnet-50	PROPN
fcis-10209	36	24	.	.	PUNCT
fcis-10209	37	1	zhan[12	zhan[12	NOUN
fcis-10209	37	2	]	]	X
fcis-10209	37	3	et	et	PROPN
fcis-10209	37	4	al	al	PROPN
fcis-10209	37	5	.	.	PROPN
fcis-10209	37	6	used	use	VERB
fcis-10209	37	7	highresolution	highresolution	NOUN
fcis-10209	37	8	remote	remote	ADJ
fcis-10209	37	9	sensing	sensing	NOUN
fcis-10209	37	10	images	image	NOUN
fcis-10209	37	11	as	as	ADP
fcis-10209	37	12	input	input	NOUN
fcis-10209	37	13	and	and	CCONJ
fcis-10209	37	14	achieved	achieve	VERB
fcis-10209	37	15	73	73	NUM
fcis-10209	37	16	automatic	automatic	ADJ
fcis-10209	37	17	segmentation	segmentation	NOUN
fcis-10209	37	18	of	of	ADP
fcis-10209	37	19	different	different	ADJ
fcis-10209	37	20	crop	crop	NOUN
fcis-10209	37	21	types	type	NOUN
fcis-10209	37	22	by	by	ADP
fcis-10209	37	23	training	train	VERB
fcis-10209	37	24	an	an	DET
fcis-10209	37	25	fcn	fcn	NOUN
fcis-10209	37	26	model	model	NOUN
fcis-10209	37	27	with	with	ADP
fcis-10209	37	28	dilated	dilate	VERB
fcis-10209	37	29	convolution	convolution	NOUN
fcis-10209	37	30	kernels	kernel	NOUN
fcis-10209	37	31	.	.	PUNCT
fcis-10209	38	1	zhu[13	zhu[13	PROPN
fcis-10209	38	2	]	]	PUNCT
fcis-10209	38	3	et	et	PROPN
fcis-10209	38	4	al	al	PROPN
fcis-10209	38	5	.	.	PROPN
fcis-10209	38	6	introduced	introduce	VERB
fcis-10209	38	7	a	a	DET
fcis-10209	38	8	self	self	NOUN
fcis-10209	38	9	-	-	PUNCT
fcis-10209	38	10	attention	attention	NOUN
fcis-10209	38	11	mechanism	mechanism	NOUN
fcis-10209	38	12	based	base	VERB
fcis-10209	38	13	on	on	ADP
fcis-10209	38	14	the	the	DET
fcis-10209	38	15	improved	improved	ADJ
fcis-10209	38	16	u	u	ADJ
fcis-10209	38	17	-	-	ADJ
fcis-10209	38	18	net	net	ADJ
fcis-10209	38	19	model	model	NOUN
fcis-10209	38	20	and	and	CCONJ
fcis-10209	38	21	achieved	achieve	VERB
fcis-10209	38	22	automatic	automatic	ADJ
fcis-10209	38	23	segmentation	segmentation	NOUN
fcis-10209	38	24	of	of	ADP
fcis-10209	38	25	different	different	ADJ
fcis-10209	38	26	crop	crop	NOUN
fcis-10209	38	27	types	type	NOUN
fcis-10209	38	28	through	through	ADP
fcis-10209	38	29	training	training	NOUN
fcis-10209	38	30	.	.	PUNCT
fcis-10209	39	1	these	these	DET
fcis-10209	39	2	models	model	NOUN
fcis-10209	39	3	have	have	VERB
fcis-10209	39	4	outstanding	outstanding	ADJ
fcis-10209	39	5	performance	performance	NOUN
fcis-10209	39	6	in	in	ADP
fcis-10209	39	7	image	image	NOUN
fcis-10209	39	8	segmentation	segmentation	NOUN
fcis-10209	39	9	and	and	CCONJ
fcis-10209	39	10	detection	detection	NOUN
fcis-10209	39	11	tasks	task	NOUN
fcis-10209	39	12	and	and	CCONJ
fcis-10209	39	13	also	also	ADV
fcis-10209	39	14	have	have	VERB
fcis-10209	39	15	potential	potential	NOUN
fcis-10209	39	16	in	in	ADP
fcis-10209	39	17	weed	weed	NOUN
fcis-10209	39	18	identification	identification	NOUN
fcis-10209	39	19	in	in	ADP
fcis-10209	39	20	the	the	DET
fcis-10209	39	21	agricultural	agricultural	ADJ
fcis-10209	39	22	field	field	NOUN
fcis-10209	39	23	.	.	PUNCT
fcis-10209	40	1	in	in	ADP
fcis-10209	40	2	summary	summary	NOUN
fcis-10209	40	3	,	,	PUNCT
fcis-10209	40	4	deep	deep	ADJ
fcis-10209	40	5	learning	learning	NOUN
fcis-10209	40	6	models	model	NOUN
fcis-10209	40	7	have	have	VERB
fcis-10209	40	8	the	the	DET
fcis-10209	40	9	advantages	advantage	NOUN
fcis-10209	40	10	of	of	ADP
fcis-10209	40	11	strong	strong	ADJ
fcis-10209	40	12	adaptive	adaptive	ADJ
fcis-10209	40	13	ability	ability	NOUN
fcis-10209	40	14	and	and	CCONJ
fcis-10209	40	15	high	high	ADJ
fcis-10209	40	16	recognition	recognition	NOUN
fcis-10209	40	17	rate	rate	NOUN
fcis-10209	40	18	,	,	PUNCT
fcis-10209	40	19	but	but	CCONJ
fcis-10209	40	20	they	they	PRON
fcis-10209	40	21	require	require	VERB
fcis-10209	40	22	a	a	DET
fcis-10209	40	23	large	large	ADJ
fcis-10209	40	24	amount	amount	NOUN
fcis-10209	40	25	of	of	ADP
fcis-10209	40	26	training	training	NOUN
fcis-10209	40	27	data	datum	NOUN
fcis-10209	40	28	and	and	CCONJ
fcis-10209	40	29	high	high	ADJ
fcis-10209	40	30	computing	computing	NOUN
fcis-10209	40	31	resources	resource	NOUN
fcis-10209	40	32	.	.	PUNCT
fcis-10209	41	1	however	however	ADV
fcis-10209	41	2	,	,	PUNCT
fcis-10209	41	3	weed	weed	VERB
fcis-10209	41	4	identification	identification	NOUN
fcis-10209	41	5	technology	technology	NOUN
fcis-10209	41	6	still	still	ADV
fcis-10209	41	7	has	have	VERB
fcis-10209	41	8	some	some	DET
fcis-10209	41	9	limitations	limitation	NOUN
fcis-10209	41	10	and	and	CCONJ
fcis-10209	41	11	deficiencies	deficiency	NOUN
fcis-10209	41	12	.	.	PUNCT
fcis-10209	42	1	firstly	firstly	ADV
fcis-10209	42	2	,	,	PUNCT
fcis-10209	42	3	weeds	weed	NOUN
fcis-10209	42	4	of	of	ADP
fcis-10209	42	5	different	different	ADJ
fcis-10209	42	6	species	specie	NOUN
fcis-10209	42	7	or	or	CCONJ
fcis-10209	42	8	growth	growth	NOUN
fcis-10209	42	9	stages	stage	NOUN
fcis-10209	42	10	have	have	VERB
fcis-10209	42	11	different	different	ADJ
fcis-10209	42	12	shapes	shape	NOUN
fcis-10209	42	13	and	and	CCONJ
fcis-10209	42	14	features	feature	NOUN
fcis-10209	42	15	,	,	PUNCT
fcis-10209	42	16	and	and	CCONJ
fcis-10209	42	17	weeds	weed	NOUN
fcis-10209	42	18	are	be	AUX
fcis-10209	42	19	very	very	ADV
fcis-10209	42	20	similar	similar	ADJ
fcis-10209	42	21	to	to	ADP
fcis-10209	42	22	crops	crop	NOUN
fcis-10209	42	23	,	,	PUNCT
fcis-10209	42	24	making	make	VERB
fcis-10209	42	25	the	the	DET
fcis-10209	42	26	recognition	recognition	NOUN
fcis-10209	42	27	task	task	NOUN
fcis-10209	42	28	more	more	ADV
fcis-10209	42	29	complex	complex	ADJ
fcis-10209	42	30	,	,	PUNCT
fcis-10209	42	31	which	which	PRON
fcis-10209	42	32	puts	put	VERB
fcis-10209	42	33	higher	high	ADJ
fcis-10209	42	34	demands	demand	NOUN
fcis-10209	42	35	on	on	ADP
fcis-10209	42	36	the	the	DET
fcis-10209	42	37	recognition	recognition	NOUN
fcis-10209	42	38	algorithm[3	algorithm[3	NOUN
fcis-10209	42	39	]	]	PUNCT
fcis-10209	42	40	.	.	PUNCT
fcis-10209	43	1	secondly	secondly	ADV
fcis-10209	43	2	,	,	PUNCT
fcis-10209	43	3	given	give	VERB
fcis-10209	43	4	that	that	SCONJ
fcis-10209	43	5	most	most	ADJ
fcis-10209	43	6	of	of	ADP
fcis-10209	43	7	the	the	DET
fcis-10209	43	8	existing	exist	VERB
fcis-10209	43	9	popular	popular	ADJ
fcis-10209	43	10	methods	method	NOUN
fcis-10209	43	11	pursue	pursue	VERB
fcis-10209	43	12	higher	high	ADJ
fcis-10209	43	13	accuracy	accuracy	NOUN
fcis-10209	43	14	,	,	PUNCT
fcis-10209	43	15	they	they	PRON
fcis-10209	43	16	can	can	AUX
fcis-10209	43	17	not	not	PART
fcis-10209	43	18	maintain	maintain	VERB
fcis-10209	43	19	faster	fast	ADJ
fcis-10209	43	20	inference	inference	NOUN
fcis-10209	43	21	speed	speed	NOUN
fcis-10209	43	22	,	,	PUNCT
fcis-10209	43	23	which	which	PRON
fcis-10209	43	24	leads	lead	VERB
fcis-10209	43	25	to	to	ADP
fcis-10209	43	26	poor	poor	ADJ
fcis-10209	43	27	performance	performance	NOUN
fcis-10209	43	28	in	in	ADP
fcis-10209	43	29	deployment	deployment	NOUN
fcis-10209	43	30	on	on	ADP
fcis-10209	43	31	resource	resource	NOUN
fcis-10209	43	32	-	-	PUNCT
fcis-10209	43	33	constrained	constrain	VERB
fcis-10209	43	34	edge	edge	NOUN
fcis-10209	43	35	devices[14	devices[14	PROPN
fcis-10209	43	36	]	]	PUNCT
fcis-10209	43	37	.	.	PUNCT
fcis-10209	44	1	finally	finally	ADV
fcis-10209	44	2	,	,	PUNCT
fcis-10209	44	3	currently	currently	ADV
fcis-10209	44	4	,	,	PUNCT
fcis-10209	44	5	most	most	ADJ
fcis-10209	44	6	of	of	ADP
fcis-10209	44	7	the	the	DET
fcis-10209	44	8	existing	exist	VERB
fcis-10209	44	9	weed	weed	NOUN
fcis-10209	44	10	identification	identification	NOUN
fcis-10209	44	11	algorithms	algorithm	NOUN
fcis-10209	44	12	are	be	AUX
fcis-10209	44	13	based	base	VERB
fcis-10209	44	14	on	on	ADP
fcis-10209	44	15	cnn	cnn	PROPN
fcis-10209	44	16	for	for	ADP
fcis-10209	44	17	feature	feature	NOUN
fcis-10209	44	18	extraction	extraction	NOUN
fcis-10209	44	19	.	.	PUNCT
fcis-10209	45	1	the	the	DET
fcis-10209	45	2	dense	dense	ADJ
fcis-10209	45	3	computation	computation	NOUN
fcis-10209	45	4	and	and	CCONJ
fcis-10209	45	5	parameter	parameter	NOUN
fcis-10209	45	6	sharing	sharing	NOUN
fcis-10209	45	7	of	of	ADP
fcis-10209	45	8	cnn	cnn	PROPN
fcis-10209	45	9	make	make	VERB
fcis-10209	45	10	its	its	PRON
fcis-10209	45	11	ability	ability	NOUN
fcis-10209	45	12	to	to	PART
fcis-10209	45	13	extract	extract	VERB
fcis-10209	45	14	local	local	ADJ
fcis-10209	45	15	semantic	semantic	ADJ
fcis-10209	45	16	information	information	NOUN
fcis-10209	45	17	quite	quite	ADV
fcis-10209	45	18	strong	strong	ADJ
fcis-10209	45	19	,	,	PUNCT
fcis-10209	45	20	but	but	CCONJ
fcis-10209	45	21	in	in	ADP
fcis-10209	45	22	real	real	ADJ
fcis-10209	45	23	-	-	PUNCT
fcis-10209	45	24	world	world	NOUN
fcis-10209	45	25	scenarios	scenario	NOUN
fcis-10209	45	26	,	,	PUNCT
fcis-10209	45	27	not	not	PART
fcis-10209	45	28	only	only	ADV
fcis-10209	45	29	local	local	ADJ
fcis-10209	45	30	information	information	NOUN
fcis-10209	45	31	needs	need	VERB
fcis-10209	45	32	to	to	PART
fcis-10209	45	33	be	be	AUX
fcis-10209	45	34	focused	focus	VERB
fcis-10209	45	35	on	on	ADP
fcis-10209	45	36	,	,	PUNCT
fcis-10209	45	37	but	but	CCONJ
fcis-10209	45	38	also	also	ADV
fcis-10209	45	39	global	global	ADJ
fcis-10209	45	40	semantic	semantic	ADJ
fcis-10209	45	41	information	information	NOUN
fcis-10209	45	42	needs	need	VERB
fcis-10209	45	43	to	to	PART
fcis-10209	45	44	be	be	AUX
fcis-10209	45	45	paid	pay	VERB
fcis-10209	45	46	attention	attention	NOUN
fcis-10209	45	47	to[15	to[15	NOUN
fcis-10209	45	48	]	]	PUNCT
fcis-10209	45	49	.	.	PUNCT
fcis-10209	46	1	therefore	therefore	ADV
fcis-10209	46	2	,	,	PUNCT
fcis-10209	46	3	in	in	ADP
fcis-10209	46	4	this	this	DET
fcis-10209	46	5	paper	paper	NOUN
fcis-10209	46	6	,	,	PUNCT
fcis-10209	46	7	a	a	DET
fcis-10209	46	8	weed	weed	NOUN
fcis-10209	46	9	identification	identification	NOUN
fcis-10209	46	10	method	method	NOUN
fcis-10209	46	11	combining	combine	VERB
fcis-10209	46	12	cnn	cnn	PROPN
fcis-10209	46	13	and	and	CCONJ
fcis-10209	46	14	transformer	transformer	ADJ
fcis-10209	46	15	hybrid	hybrid	ADJ
fcis-10209	46	16	model	model	NOUN
fcis-10209	46	17	will	will	AUX
fcis-10209	46	18	be	be	AUX
fcis-10209	46	19	proposed	propose	VERB
fcis-10209	46	20	to	to	PART
fcis-10209	46	21	solve	solve	VERB
fcis-10209	46	22	the	the	DET
fcis-10209	46	23	above	above	ADJ
fcis-10209	46	24	problems	problem	NOUN
fcis-10209	46	25	.	.	PUNCT
fcis-10209	47	1	this	this	DET
fcis-10209	47	2	method	method	NOUN
fcis-10209	47	3	first	first	ADV
fcis-10209	47	4	uses	use	VERB
fcis-10209	47	5	continuous	continuous	ADJ
fcis-10209	47	6	convolutional	convolutional	ADJ
fcis-10209	47	7	layers	layer	NOUN
fcis-10209	47	8	for	for	ADP
fcis-10209	47	9	local	local	ADJ
fcis-10209	47	10	feature	feature	NOUN
fcis-10209	47	11	extraction	extraction	NOUN
fcis-10209	47	12	,	,	PUNCT
fcis-10209	47	13	then	then	ADV
fcis-10209	47	14	uses	use	VERB
fcis-10209	47	15	transformer	transformer	NOUN
fcis-10209	47	16	block	block	NOUN
fcis-10209	47	17	for	for	ADP
fcis-10209	47	18	global	global	ADJ
fcis-10209	47	19	feature	feature	NOUN
fcis-10209	47	20	extraction	extraction	NOUN
fcis-10209	47	21	,	,	PUNCT
fcis-10209	47	22	and	and	CCONJ
fcis-10209	47	23	finally	finally	ADV
fcis-10209	47	24	combines	combine	VERB
fcis-10209	47	25	the	the	DET
fcis-10209	47	26	features	feature	NOUN
fcis-10209	47	27	of	of	ADP
fcis-10209	47	28	both	both	PRON
fcis-10209	47	29	to	to	PART
fcis-10209	47	30	conduct	conduct	VERB
fcis-10209	47	31	image	image	NOUN
fcis-10209	47	32	classification	classification	NOUN
fcis-10209	47	33	.	.	PUNCT
fcis-10209	48	1	among	among	ADP
fcis-10209	48	2	them	they	PRON
fcis-10209	48	3	,	,	PUNCT
fcis-10209	48	4	the	the	DET
fcis-10209	48	5	transformer	transformer	NOUN
fcis-10209	48	6	block	block	NOUN
fcis-10209	48	7	is	be	AUX
fcis-10209	48	8	based	base	VERB
fcis-10209	48	9	on	on	ADP
fcis-10209	48	10	the	the	DET
fcis-10209	48	11	edgenext	edgenext	NOUN
fcis-10209	48	12	architecture	architecture	NOUN
fcis-10209	48	13	,	,	PUNCT
fcis-10209	48	14	introducing	introduce	VERB
fcis-10209	48	15	sdta	sdta	NOUN
fcis-10209	48	16	(	(	PUNCT
fcis-10209	48	17	segmentation	segmentation	NOUN
fcis-10209	48	18	depth	depth	NOUN
fcis-10209	48	19	transpose	transpose	NOUN
fcis-10209	48	20	attention	attention	NOUN
fcis-10209	48	21	)	)	PUNCT
fcis-10209	48	22	,	,	PUNCT
fcis-10209	48	23	and	and	CCONJ
fcis-10209	48	24	using	use	VERB
fcis-10209	48	25	the	the	DET
fcis-10209	48	26	"	"	PUNCT
fcis-10209	48	27	patchify	patchify	NOUN
fcis-10209	48	28	"	"	PUNCT
fcis-10209	48	29	strategy	strategy	NOUN
fcis-10209	48	30	in	in	ADP
fcis-10209	48	31	the	the	DET
fcis-10209	48	32	input	input	NOUN
fcis-10209	48	33	section	section	NOUN
fcis-10209	48	34	,	,	PUNCT
fcis-10209	48	35	followed	follow	VERB
fcis-10209	48	36	by	by	ADP
fcis-10209	48	37	continuous	continuous	ADJ
fcis-10209	48	38	stacking	stacking	NOUN
fcis-10209	48	39	of	of	ADP
fcis-10209	48	40	three	three	NUM
fcis-10209	48	41	convolutional	convolutional	ADJ
fcis-10209	48	42	encoders	encoder	NOUN
fcis-10209	48	43	to	to	PART
fcis-10209	48	44	extract	extract	VERB
fcis-10209	48	45	local	local	ADJ
fcis-10209	48	46	features	feature	NOUN
fcis-10209	48	47	.	.	PUNCT
fcis-10209	49	1	at	at	ADP
fcis-10209	49	2	the	the	DET
fcis-10209	49	3	same	same	ADJ
fcis-10209	49	4	time	time	NOUN
fcis-10209	49	5	,	,	PUNCT
fcis-10209	49	6	this	this	DET
fcis-10209	49	7	paper	paper	NOUN
fcis-10209	49	8	will	will	AUX
fcis-10209	49	9	optimize	optimize	VERB
fcis-10209	49	10	it	it	PRON
fcis-10209	49	11	according	accord	VERB
fcis-10209	49	12	to	to	ADP
fcis-10209	49	13	the	the	DET
fcis-10209	49	14	characteristics	characteristic	NOUN
fcis-10209	49	15	of	of	ADP
fcis-10209	49	16	practical	practical	ADJ
fcis-10209	49	17	application	application	NOUN
fcis-10209	49	18	scenarios	scenario	NOUN
fcis-10209	49	19	and	and	CCONJ
fcis-10209	49	20	datasets	dataset	NOUN
fcis-10209	49	21	,	,	PUNCT
fcis-10209	49	22	and	and	CCONJ
fcis-10209	49	23	use	use	VERB
fcis-10209	49	24	real	real	ADJ
fcis-10209	49	25	data	datum	NOUN
fcis-10209	49	26	for	for	ADP
fcis-10209	49	27	testing	testing	NOUN
fcis-10209	49	28	to	to	PART
fcis-10209	49	29	verify	verify	VERB
fcis-10209	49	30	the	the	DET
fcis-10209	49	31	accuracy	accuracy	NOUN
fcis-10209	49	32	and	and	CCONJ
fcis-10209	49	33	precision	precision	NOUN
fcis-10209	49	34	of	of	ADP
fcis-10209	49	35	this	this	DET
fcis-10209	49	36	method	method	NOUN
fcis-10209	49	37	in	in	ADP
fcis-10209	49	38	weed	weed	NOUN
fcis-10209	49	39	identification	identification	NOUN
fcis-10209	49	40	,	,	PUNCT
fcis-10209	49	41	providing	provide	VERB
fcis-10209	49	42	scientific	scientific	ADJ
fcis-10209	49	43	support	support	NOUN
fcis-10209	49	44	for	for	ADP
fcis-10209	49	45	achieving	achieve	VERB
fcis-10209	49	46	more	more	ADV
fcis-10209	49	47	accurate	accurate	ADJ
fcis-10209	49	48	and	and	CCONJ
fcis-10209	49	49	efficient	efficient	ADJ
fcis-10209	49	50	weed	weed	NOUN
fcis-10209	49	51	identification	identification	NOUN
fcis-10209	49	52	.	.	PUNCT
fcis-10209	50	1	2	2	X
fcis-10209	50	2	.	.	X
fcis-10209	50	3	experimental	experimental	ADJ
fcis-10209	50	4	materials	material	NOUN
fcis-10209	50	5	2.1	2.1	NUM
fcis-10209	50	6	.	.	PUNCT
fcis-10209	51	1	dataset	dataset	VERB
fcis-10209	51	2	fig	fig	NOUN
fcis-10209	51	3	1	1	NUM
fcis-10209	51	4	.	.	PUNCT
fcis-10209	51	5	images	image	NOUN
fcis-10209	51	6	of	of	ADP
fcis-10209	51	7	different	different	ADJ
fcis-10209	51	8	weed	weed	NOUN
fcis-10209	51	9	categories	category	NOUN
fcis-10209	51	10	.	.	PUNCT
fcis-10209	52	1	the	the	DET
fcis-10209	52	2	weed	weed	NOUN
fcis-10209	52	3	dataset	dataset	NOUN
fcis-10209	52	4	used	use	VERB
fcis-10209	52	5	in	in	ADP
fcis-10209	52	6	this	this	DET
fcis-10209	52	7	project	project	NOUN
fcis-10209	52	8	comes	come	VERB
fcis-10209	52	9	from	from	ADP
fcis-10209	52	10	the	the	DET
fcis-10209	52	11	deepweeds	deepweed	NOUN
fcis-10209	52	12	weed	weed	VERB
fcis-10209	52	13	dataset	dataset	VERB
fcis-10209	52	14	publicly	publicly	ADV
fcis-10209	52	15	available	available	ADJ
fcis-10209	52	16	by	by	ADP
fcis-10209	52	17	olsen[16	olsen[16	PROPN
fcis-10209	52	18	]	]	PUNCT
fcis-10209	52	19	et	et	PROPN
fcis-10209	52	20	al	al	PROPN
fcis-10209	52	21	.	.	PUNCT
fcis-10209	53	1	the	the	DET
fcis-10209	53	2	dataset	dataset	NOUN
fcis-10209	53	3	consists	consist	VERB
fcis-10209	53	4	of	of	ADP
fcis-10209	53	5	9	9	NUM
fcis-10209	53	6	categories	category	NOUN
fcis-10209	53	7	,	,	PUNCT
fcis-10209	53	8	including	include	VERB
fcis-10209	53	9	8	8	NUM
fcis-10209	53	10	weed	weed	NOUN
fcis-10209	53	11	categories	category	NOUN
fcis-10209	53	12	and	and	CCONJ
fcis-10209	53	13	1	1	NUM
fcis-10209	53	14	negative	negative	ADJ
fcis-10209	53	15	class	class	NOUN
fcis-10209	53	16	,	,	PUNCT
fcis-10209	53	17	covering	cover	VERB
fcis-10209	53	18	some	some	DET
fcis-10209	53	19	common	common	ADJ
fcis-10209	53	20	weed	weed	NOUN
fcis-10209	53	21	species	specie	NOUN
fcis-10209	53	22	such	such	ADJ
fcis-10209	53	23	as	as	ADP
fcis-10209	53	24	chineseapple	chineseapple	ADJ
fcis-10209	53	25	,	,	PUNCT
fcis-10209	53	26	siam	siam	ADJ
fcis-10209	53	27	weed	weed	NOUN
fcis-10209	53	28	,	,	PUNCT
fcis-10209	53	29	parkinsonia	parkinsonia	NOUN
fcis-10209	53	30	,	,	PUNCT
fcis-10209	53	31	etc	etc	X
fcis-10209	53	32	.	.	X
fcis-10209	54	1	the	the	DET
fcis-10209	54	2	dataset	dataset	NOUN
fcis-10209	54	3	contains	contain	VERB
fcis-10209	54	4	images	image	NOUN
fcis-10209	54	5	captured	capture	VERB
fcis-10209	54	6	in	in	ADP
fcis-10209	54	7	natural	natural	ADJ
fcis-10209	54	8	environments	environment	NOUN
fcis-10209	54	9	with	with	ADP
fcis-10209	54	10	various	various	ADJ
fcis-10209	54	11	lighting	lighting	NOUN
fcis-10209	54	12	and	and	CCONJ
fcis-10209	54	13	angles	angle	NOUN
fcis-10209	54	14	.	.	PUNCT
fcis-10209	55	1	the	the	DET
fcis-10209	55	2	dataset	dataset	NOUN
fcis-10209	55	3	contains	contain	VERB
fcis-10209	55	4	a	a	DET
fcis-10209	55	5	total	total	NOUN
fcis-10209	55	6	of	of	ADP
fcis-10209	55	7	17,509	17,509	NUM
fcis-10209	55	8	images	image	NOUN
fcis-10209	55	9	,	,	PUNCT
fcis-10209	55	10	which	which	PRON
fcis-10209	55	11	have	have	AUX
fcis-10209	55	12	been	be	AUX
fcis-10209	55	13	divided	divide	VERB
fcis-10209	55	14	into	into	ADP
fcis-10209	55	15	14,036	14,036	NUM
fcis-10209	55	16	training	training	NOUN
fcis-10209	55	17	set	set	NOUN
fcis-10209	55	18	images	image	NOUN
fcis-10209	55	19	and	and	CCONJ
fcis-10209	55	20	3,473	3,473	NUM
fcis-10209	55	21	testing	testing	NOUN
fcis-10209	55	22	set	set	VERB
fcis-10209	55	23	images	image	NOUN
fcis-10209	55	24	in	in	ADP
fcis-10209	55	25	this	this	DET
fcis-10209	55	26	study	study	NOUN
fcis-10209	55	27	.	.	PUNCT
fcis-10209	56	1	sample	sample	NOUN
fcis-10209	56	2	images	image	NOUN
fcis-10209	56	3	from	from	ADP
fcis-10209	56	4	the	the	DET
fcis-10209	56	5	dataset	dataset	NOUN
fcis-10209	56	6	are	be	AUX
fcis-10209	56	7	shown	show	VERB
fcis-10209	56	8	in	in	ADP
fcis-10209	56	9	fig.1	fig.1	PROPN
fcis-10209	56	10	.	.	PUNCT
fcis-10209	57	1	2.2	2.2	NUM
fcis-10209	57	2	.	.	PUNCT
fcis-10209	57	3	data	datum	NOUN
fcis-10209	57	4	preprocessing	preprocesse	VERB
fcis-10209	57	5	vit	vit	NOUN
fcis-10209	57	6	model	model	NOUN
fcis-10209	57	7	was	be	AUX
fcis-10209	57	8	trained	train	VERB
fcis-10209	57	9	on	on	ADP
fcis-10209	57	10	the	the	DET
fcis-10209	57	11	imagenet	imagenet	NOUN
fcis-10209	57	12	dataset	dataset	NOUN
fcis-10209	57	13	,	,	PUNCT
fcis-10209	57	14	so	so	CCONJ
fcis-10209	57	15	its	its	PRON
fcis-10209	57	16	input	input	NOUN
fcis-10209	57	17	image	image	NOUN
fcis-10209	57	18	size	size	NOUN
fcis-10209	57	19	is	be	AUX
fcis-10209	57	20	224×224	224×224	NUM
fcis-10209	57	21	.	.	PUNCT
fcis-10209	58	1	in	in	ADP
fcis-10209	58	2	order	order	NOUN
fcis-10209	58	3	to	to	PART
fcis-10209	58	4	adapt	adapt	VERB
fcis-10209	58	5	the	the	DET
fcis-10209	58	6	deepweeds	deepweed	NOUN
fcis-10209	58	7	dataset	dataset	VERB
fcis-10209	58	8	to	to	ADP
fcis-10209	58	9	the	the	DET
fcis-10209	58	10	vit	vit	ADJ
fcis-10209	58	11	network	network	NOUN
fcis-10209	58	12	,	,	PUNCT
fcis-10209	58	13	improve	improve	VERB
fcis-10209	58	14	computational	computational	ADJ
fcis-10209	58	15	efficiency	efficiency	NOUN
fcis-10209	58	16	,	,	PUNCT
fcis-10209	58	17	and	and	CCONJ
fcis-10209	58	18	make	make	VERB
fcis-10209	58	19	the	the	DET
fcis-10209	58	20	model	model	NOUN
fcis-10209	58	21	easier	easy	ADJ
fcis-10209	58	22	to	to	PART
fcis-10209	58	23	train	train	VERB
fcis-10209	58	24	and	and	CCONJ
fcis-10209	58	25	optimize	optimize	VERB
fcis-10209	58	26	,	,	PUNCT
fcis-10209	58	27	all	all	DET
fcis-10209	58	28	images	image	NOUN
fcis-10209	58	29	were	be	AUX
fcis-10209	58	30	resized	resize	VERB
fcis-10209	58	31	to	to	ADP
fcis-10209	58	32	224×224	224×224	NUM
fcis-10209	58	33	and	and	CCONJ
fcis-10209	58	34	the	the	DET
fcis-10209	58	35	data	datum	NOUN
fcis-10209	58	36	was	be	AUX
fcis-10209	58	37	normalized	normalize	VERB
fcis-10209	58	38	the	the	DET
fcis-10209	58	39	data	datum	NOUN
fcis-10209	58	40	to	to	ADP
fcis-10209	58	41	[	[	X
fcis-10209	58	42	0,1	0,1	NUM
fcis-10209	58	43	]	]	PUNCT
fcis-10209	58	44	in	in	ADP
fcis-10209	58	45	this	this	DET
fcis-10209	58	46	study	study	NOUN
fcis-10209	58	47	to	to	PART
fcis-10209	58	48	avoid	avoid	VERB
fcis-10209	58	49	numerical	numerical	ADJ
fcis-10209	58	50	overflow	overflow	NOUN
fcis-10209	58	51	while	while	SCONJ
fcis-10209	58	52	maintaining	maintain	VERB
fcis-10209	58	53	numerical	numerical	ADJ
fcis-10209	58	54	stability	stability	NOUN
fcis-10209	58	55	.	.	PUNCT
fcis-10209	59	1	3	3	X
fcis-10209	59	2	.	.	X
fcis-10209	59	3	experimental	experimental	ADJ
fcis-10209	59	4	procedure	procedure	NOUN
fcis-10209	59	5	the	the	DET
fcis-10209	59	6	weed	weed	NOUN
fcis-10209	59	7	recognition	recognition	NOUN
fcis-10209	59	8	network	network	NOUN
fcis-10209	59	9	proposed	propose	VERB
fcis-10209	59	10	in	in	ADP
fcis-10209	59	11	this	this	DET
fcis-10209	59	12	paper	paper	NOUN
fcis-10209	59	13	is	be	AUX
fcis-10209	59	14	mainly	mainly	ADV
fcis-10209	59	15	composed	compose	VERB
fcis-10209	59	16	of	of	ADP
fcis-10209	59	17	a	a	DET
fcis-10209	59	18	cnn	cnn	PROPN
fcis-10209	59	19	-	-	PUNCT
fcis-10209	59	20	transformer	transformer	NOUN
fcis-10209	59	21	hybrid	hybrid	NOUN
fcis-10209	59	22	model	model	NOUN
fcis-10209	59	23	,	,	PUNCT
fcis-10209	59	24	which	which	PRON
fcis-10209	59	25	includes	include	VERB
fcis-10209	59	26	stemconv	stemconv	ADJ
fcis-10209	59	27	module	module	NOUN
fcis-10209	59	28	,	,	PUNCT
fcis-10209	59	29	convblock(cnn	convblock(cnn	NOUN
fcis-10209	59	30	)	)	PUNCT
fcis-10209	59	31	module	module	NOUN
fcis-10209	59	32	,	,	PUNCT
fcis-10209	59	33	transformer	transformer	NOUN
fcis-10209	59	34	encoder	encoder	NOUN
fcis-10209	59	35	module	module	NOUN
fcis-10209	59	36	,	,	PUNCT
fcis-10209	59	37	and	and	CCONJ
fcis-10209	59	38	pooling	pool	VERB
fcis-10209	59	39	fully	fully	ADV
fcis-10209	59	40	connected	connected	ADJ
fcis-10209	59	41	classification	classification	NOUN
fcis-10209	59	42	module	module	NOUN
fcis-10209	59	43	.	.	PUNCT
fcis-10209	60	1	3.1	3.1	NUM
fcis-10209	60	2	.	.	PUNCT
fcis-10209	60	3	overall	overall	ADJ
fcis-10209	60	4	model	model	NOUN
fcis-10209	60	5	architecture	architecture	NOUN
fcis-10209	60	6	fig	fig	NOUN
fcis-10209	60	7	2	2	NUM
fcis-10209	60	8	.	.	PUNCT
fcis-10209	60	9	overall	overall	ADJ
fcis-10209	60	10	architecture	architecture	NOUN
fcis-10209	60	11	diagram	diagram	NOUN
fcis-10209	60	12	of	of	ADP
fcis-10209	60	13	the	the	DET
fcis-10209	60	14	hybrid	hybrid	ADJ
fcis-10209	60	15	model	model	NOUN
fcis-10209	60	16	fig	fig	PROPN
fcis-10209	60	17	3	3	X
fcis-10209	60	18	.	.	PUNCT
fcis-10209	61	1	the	the	DET
fcis-10209	61	2	specific	specific	ADJ
fcis-10209	61	3	structure	structure	NOUN
fcis-10209	61	4	of	of	ADP
fcis-10209	61	5	block	block	NOUN
fcis-10209	61	6	i	i	PROPN
fcis-10209	61	7	(	(	PUNCT
fcis-10209	61	8	1~4	1~4	NUM
fcis-10209	61	9	)	)	PUNCT
fcis-10209	61	10	the	the	DET
fcis-10209	61	11	overall	overall	ADJ
fcis-10209	61	12	structure	structure	NOUN
fcis-10209	61	13	of	of	ADP
fcis-10209	61	14	the	the	DET
fcis-10209	61	15	cnn	cnn	PROPN
fcis-10209	61	16	-	-	PUNCT
fcis-10209	61	17	transformer	transformer	NOUN
fcis-10209	61	18	hybrid	hybrid	NOUN
fcis-10209	61	19	model	model	NOUN
fcis-10209	61	20	proposed	propose	VERB
fcis-10209	61	21	in	in	ADP
fcis-10209	61	22	this	this	DET
fcis-10209	61	23	study	study	NOUN
fcis-10209	61	24	is	be	AUX
fcis-10209	61	25	shown	show	VERB
fcis-10209	61	26	in	in	ADP
fcis-10209	61	27	fig.2	fig.2	PROPN
fcis-10209	61	28	.	.	PUNCT
fcis-10209	62	1	firstly	firstly	ADV
fcis-10209	62	2	,	,	PUNCT
fcis-10209	62	3	the	the	DET
fcis-10209	62	4	input	input	NOUN
fcis-10209	62	5	image	image	NOUN
fcis-10209	62	6	is	be	AUX
fcis-10209	62	7	passed	pass	VERB
fcis-10209	62	8	into	into	ADP
fcis-10209	62	9	the	the	DET
fcis-10209	62	10	stemconv	stemconv	ADJ
fcis-10209	62	11	layer	layer	NOUN
fcis-10209	62	12	(	(	PUNCT
fcis-10209	62	13	which	which	PRON
fcis-10209	62	14	74	74	NUM
fcis-10209	62	15	includes	include	VERB
fcis-10209	62	16	four	four	NUM
fcis-10209	62	17	convolutional	convolutional	ADJ
fcis-10209	62	18	layers	layer	NOUN
fcis-10209	62	19	)	)	PUNCT
fcis-10209	62	20	to	to	PART
fcis-10209	62	21	perform	perform	VERB
fcis-10209	62	22	preliminary	preliminary	ADJ
fcis-10209	62	23	feature	feature	NOUN
fcis-10209	62	24	extraction	extraction	NOUN
fcis-10209	62	25	and	and	CCONJ
fcis-10209	62	26	downsampling	downsampling	NOUN
fcis-10209	62	27	.	.	PUNCT
fcis-10209	63	1	this	this	PRON
fcis-10209	63	2	mainly	mainly	ADV
fcis-10209	63	3	aims	aim	VERB
fcis-10209	63	4	to	to	PART
fcis-10209	63	5	reduce	reduce	VERB
fcis-10209	63	6	network	network	NOUN
fcis-10209	63	7	computation	computation	NOUN
fcis-10209	63	8	and	and	CCONJ
fcis-10209	63	9	improve	improve	VERB
fcis-10209	63	10	time	time	NOUN
fcis-10209	63	11	efficiency	efficiency	NOUN
fcis-10209	63	12	,	,	PUNCT
fcis-10209	63	13	as	as	SCONJ
fcis-10209	63	14	high	high	ADJ
fcis-10209	63	15	-	-	PUNCT
fcis-10209	63	16	resolution	resolution	NOUN
fcis-10209	63	17	images	image	NOUN
fcis-10209	63	18	require	require	VERB
fcis-10209	63	19	high	high	ADJ
fcis-10209	63	20	-	-	PUNCT
fcis-10209	63	21	performance	performance	NOUN
fcis-10209	63	22	devices	device	NOUN
fcis-10209	63	23	for	for	ADP
fcis-10209	63	24	training	train	VERB
fcis-10209	63	25	a	a	DET
fcis-10209	63	26	network	network	NOUN
fcis-10209	63	27	using	use	VERB
fcis-10209	63	28	transformer	transformer	NOUN
fcis-10209	63	29	to	to	PART
fcis-10209	63	30	extract	extract	VERB
fcis-10209	63	31	features	feature	NOUN
fcis-10209	63	32	.	.	PUNCT
fcis-10209	64	1	additionally	additionally	ADV
fcis-10209	64	2	,	,	PUNCT
fcis-10209	64	3	this	this	DET
fcis-10209	64	4	step	step	NOUN
fcis-10209	64	5	transforms	transform	VERB
fcis-10209	64	6	the	the	DET
fcis-10209	64	7	data	datum	NOUN
fcis-10209	64	8	to	to	ADP
fcis-10209	64	9	a	a	DET
fcis-10209	64	10	form	form	NOUN
fcis-10209	64	11	suitable	suitable	ADJ
fcis-10209	64	12	for	for	ADP
fcis-10209	64	13	subsequent	subsequent	ADJ
fcis-10209	64	14	neural	neural	ADJ
fcis-10209	64	15	network	network	NOUN
fcis-10209	64	16	processing	processing	NOUN
fcis-10209	64	17	.	.	PUNCT
fcis-10209	65	1	secondly	secondly	ADV
fcis-10209	65	2	,	,	PUNCT
fcis-10209	65	3	the	the	DET
fcis-10209	65	4	preprocessed	preprocesse	VERB
fcis-10209	65	5	data	data	NOUN
fcis-10209	65	6	is	be	AUX
fcis-10209	65	7	fed	feed	VERB
fcis-10209	65	8	into	into	ADP
fcis-10209	65	9	a	a	DET
fcis-10209	65	10	four	four	NUM
fcis-10209	65	11	-	-	PUNCT
fcis-10209	65	12	stage	stage	NOUN
fcis-10209	65	13	network	network	NOUN
fcis-10209	65	14	structure	structure	NOUN
fcis-10209	65	15	,	,	PUNCT
fcis-10209	65	16	where	where	SCONJ
fcis-10209	65	17	each	each	DET
fcis-10209	65	18	block	block	NOUN
fcis-10209	65	19	receives	receive	VERB
fcis-10209	65	20	the	the	DET
fcis-10209	65	21	output	output	NOUN
fcis-10209	65	22	of	of	ADP
fcis-10209	65	23	the	the	DET
fcis-10209	65	24	previous	previous	ADJ
fcis-10209	65	25	block	block	NOUN
fcis-10209	65	26	as	as	ADP
fcis-10209	65	27	its	its	PRON
fcis-10209	65	28	input	input	NOUN
fcis-10209	65	29	and	and	CCONJ
fcis-10209	65	30	performs	perform	VERB
fcis-10209	65	31	further	further	ADJ
fcis-10209	65	32	feature	feature	NOUN
fcis-10209	65	33	extraction	extraction	NOUN
fcis-10209	65	34	and	and	CCONJ
fcis-10209	65	35	transformation	transformation	NOUN
fcis-10209	65	36	.	.	PUNCT
fcis-10209	66	1	specifically	specifically	ADV
fcis-10209	66	2	,	,	PUNCT
fcis-10209	66	3	this	this	PRON
fcis-10209	66	4	includes	include	VERB
fcis-10209	66	5	n	n	NOUN
fcis-10209	66	6	convblocks	convblock	NOUN
fcis-10209	66	7	at	at	ADP
fcis-10209	66	8	the	the	DET
fcis-10209	66	9	beginning	beginning	NOUN
fcis-10209	66	10	and	and	CCONJ
fcis-10209	66	11	1	1	NUM
fcis-10209	66	12	transblock	transblock	NOUN
fcis-10209	66	13	at	at	ADP
fcis-10209	66	14	the	the	DET
fcis-10209	66	15	end	end	NOUN
fcis-10209	66	16	,	,	PUNCT
fcis-10209	66	17	as	as	SCONJ
fcis-10209	66	18	shown	show	VERB
fcis-10209	66	19	in	in	ADP
fcis-10209	66	20	fig.3	fig.3	PROPN
fcis-10209	66	21	,	,	PUNCT
fcis-10209	66	22	with	with	ADP
fcis-10209	66	23	the	the	DET
fcis-10209	66	24	number	number	NOUN
fcis-10209	66	25	of	of	ADP
fcis-10209	66	26	convblocks	convblock	NOUN
fcis-10209	66	27	adjusted	adjust	VERB
fcis-10209	66	28	based	base	VERB
fcis-10209	66	29	on	on	ADP
fcis-10209	66	30	pretraining	pretraine	VERB
fcis-10209	66	31	(	(	PUNCT
fcis-10209	66	32	in	in	ADP
fcis-10209	66	33	this	this	DET
fcis-10209	66	34	study	study	NOUN
fcis-10209	66	35	,	,	PUNCT
fcis-10209	66	36	the	the	DET
fcis-10209	66	37	number	number	NOUN
fcis-10209	66	38	of	of	ADP
fcis-10209	66	39	convblocks	convblock	NOUN
fcis-10209	66	40	for	for	ADP
fcis-10209	66	41	the	the	DET
fcis-10209	66	42	four	four	NUM
fcis-10209	66	43	blocks	block	NOUN
fcis-10209	66	44	were	be	AUX
fcis-10209	66	45	3	3	NUM
fcis-10209	66	46	,	,	PUNCT
fcis-10209	66	47	4	4	NUM
fcis-10209	66	48	,	,	PUNCT
fcis-10209	66	49	10	10	NUM
fcis-10209	66	50	,	,	PUNCT
fcis-10209	66	51	and	and	CCONJ
fcis-10209	66	52	3	3	NUM
fcis-10209	66	53	respectively	respectively	ADV
fcis-10209	66	54	)	)	PUNCT
fcis-10209	66	55	.	.	PUNCT
fcis-10209	67	1	finally	finally	ADV
fcis-10209	67	2	,	,	PUNCT
fcis-10209	67	3	the	the	DET
fcis-10209	67	4	model	model	NOUN
fcis-10209	67	5	performs	perform	VERB
fcis-10209	67	6	final	final	ADJ
fcis-10209	67	7	classification	classification	NOUN
fcis-10209	67	8	by	by	ADP
fcis-10209	67	9	reducing	reduce	VERB
fcis-10209	67	10	the	the	DET
fcis-10209	67	11	number	number	NOUN
fcis-10209	67	12	of	of	ADP
fcis-10209	67	13	features	feature	NOUN
fcis-10209	67	14	through	through	ADP
fcis-10209	67	15	a	a	DET
fcis-10209	67	16	pooling	pool	VERB
fcis-10209	67	17	layer	layer	NOUN
fcis-10209	67	18	,	,	PUNCT
fcis-10209	67	19	taking	take	VERB
fcis-10209	67	20	the	the	DET
fcis-10209	67	21	average	average	ADJ
fcis-10209	67	22	value	value	NOUN
fcis-10209	67	23	at	at	ADP
fcis-10209	67	24	each	each	DET
fcis-10209	67	25	feature	feature	NOUN
fcis-10209	67	26	position	position	NOUN
fcis-10209	67	27	in	in	ADP
fcis-10209	67	28	the	the	DET
fcis-10209	67	29	feature	feature	NOUN
fcis-10209	67	30	map	map	NOUN
fcis-10209	67	31	and	and	CCONJ
fcis-10209	67	32	mapping	map	VERB
fcis-10209	67	33	it	it	PRON
fcis-10209	67	34	to	to	ADP
fcis-10209	67	35	the	the	DET
fcis-10209	67	36	number	number	NOUN
fcis-10209	67	37	of	of	ADP
fcis-10209	67	38	output	output	NOUN
fcis-10209	67	39	categories	category	NOUN
fcis-10209	67	40	through	through	ADP
fcis-10209	67	41	a	a	DET
fcis-10209	67	42	fully	fully	ADV
fcis-10209	67	43	connected	connect	VERB
fcis-10209	67	44	layer	layer	NOUN
fcis-10209	67	45	to	to	PART
fcis-10209	67	46	display	display	VERB
fcis-10209	67	47	the	the	DET
fcis-10209	67	48	predicted	predict	VERB
fcis-10209	67	49	results	result	NOUN
fcis-10209	67	50	of	of	ADP
fcis-10209	67	51	the	the	DET
fcis-10209	67	52	input	input	NOUN
fcis-10209	67	53	image	image	NOUN
fcis-10209	67	54	.	.	PUNCT
fcis-10209	68	1	3.2	3.2	NUM
fcis-10209	68	2	.	.	PUNCT
fcis-10209	69	1	convblock	convblock	PROPN
fcis-10209	69	2	the	the	DET
fcis-10209	69	3	convblock	convblock	NOUN
fcis-10209	69	4	is	be	AUX
fcis-10209	69	5	designed	design	VERB
fcis-10209	69	6	to	to	PART
fcis-10209	69	7	extract	extract	VERB
fcis-10209	69	8	high	high	ADJ
fcis-10209	69	9	-	-	PUNCT
fcis-10209	69	10	level	level	NOUN
fcis-10209	69	11	features	feature	NOUN
fcis-10209	69	12	from	from	ADP
fcis-10209	69	13	input	input	NOUN
fcis-10209	69	14	data	datum	NOUN
fcis-10209	69	15	in	in	ADP
fcis-10209	69	16	a	a	DET
fcis-10209	69	17	deep	deep	ADJ
fcis-10209	69	18	learning	learning	NOUN
fcis-10209	69	19	model	model	NOUN
fcis-10209	69	20	,	,	PUNCT
fcis-10209	69	21	improving	improve	VERB
fcis-10209	69	22	its	its	PRON
fcis-10209	69	23	generalization	generalization	NOUN
fcis-10209	69	24	performance	performance	NOUN
fcis-10209	69	25	and	and	CCONJ
fcis-10209	69	26	enabling	enable	VERB
fcis-10209	69	27	it	it	PRON
fcis-10209	69	28	to	to	PART
fcis-10209	69	29	better	well	ADV
fcis-10209	69	30	complete	complete	VERB
fcis-10209	69	31	various	various	ADJ
fcis-10209	69	32	tasks	task	NOUN
fcis-10209	69	33	.	.	PUNCT
fcis-10209	70	1	through	through	ADP
fcis-10209	70	2	operations	operation	NOUN
fcis-10209	70	3	such	such	ADJ
fcis-10209	70	4	as	as	ADP
fcis-10209	70	5	convolutional	convolutional	ADJ
fcis-10209	70	6	layers	layer	NOUN
fcis-10209	70	7	,	,	PUNCT
fcis-10209	70	8	pooling	pool	VERB
fcis-10209	70	9	layers	layer	NOUN
fcis-10209	70	10	,	,	PUNCT
fcis-10209	70	11	batch	batch	VERB
fcis-10209	70	12	normalization	normalization	NOUN
fcis-10209	70	13	layers	layer	NOUN
fcis-10209	70	14	,	,	PUNCT
fcis-10209	70	15	and	and	CCONJ
fcis-10209	70	16	nonlinear	nonlinear	ADJ
fcis-10209	70	17	activation	activation	NOUN
fcis-10209	70	18	functions	function	NOUN
fcis-10209	70	19	,	,	PUNCT
fcis-10209	70	20	the	the	DET
fcis-10209	70	21	input	input	NOUN
fcis-10209	70	22	data	data	NOUN
fcis-10209	70	23	is	be	AUX
fcis-10209	70	24	gradually	gradually	ADV
fcis-10209	70	25	abstracted	abstract	VERB
fcis-10209	70	26	and	and	CCONJ
fcis-10209	70	27	compressed	compress	VERB
fcis-10209	70	28	,	,	PUNCT
fcis-10209	70	29	enhancing	enhance	VERB
fcis-10209	70	30	the	the	DET
fcis-10209	70	31	model	model	NOUN
fcis-10209	70	32	's	's	PART
fcis-10209	70	33	performance	performance	NOUN
fcis-10209	70	34	and	and	CCONJ
fcis-10209	70	35	generalization	generalization	NOUN
fcis-10209	70	36	abilities	ability	NOUN
fcis-10209	70	37	.	.	PUNCT
fcis-10209	71	1	in	in	ADP
fcis-10209	71	2	this	this	DET
fcis-10209	71	3	paper	paper	NOUN
fcis-10209	71	4	,	,	PUNCT
fcis-10209	71	5	we	we	PRON
fcis-10209	71	6	continuously	continuously	ADV
fcis-10209	71	7	trained	train	VERB
fcis-10209	71	8	the	the	DET
fcis-10209	71	9	model	model	NOUN
fcis-10209	71	10	and	and	CCONJ
fcis-10209	71	11	designed	design	VERB
fcis-10209	71	12	the	the	DET
fcis-10209	71	13	convolutional	convolutional	ADJ
fcis-10209	71	14	block	block	NOUN
fcis-10209	71	15	shown	show	VERB
fcis-10209	71	16	in	in	ADP
fcis-10209	71	17	fig.4	fig.4	PROPN
fcis-10209	71	18	.	.	PUNCT
fcis-10209	72	1	when	when	SCONJ
fcis-10209	72	2	the	the	DET
fcis-10209	72	3	input	input	NOUN
fcis-10209	72	4	is	be	AUX
fcis-10209	72	5	passed	pass	VERB
fcis-10209	72	6	into	into	ADP
fcis-10209	72	7	the	the	DET
fcis-10209	72	8	convblock	convblock	NOUN
fcis-10209	72	9	,	,	PUNCT
fcis-10209	72	10	the	the	DET
fcis-10209	72	11	data	data	NOUN
fcis-10209	72	12	is	be	AUX
fcis-10209	72	13	first	first	ADV
fcis-10209	72	14	grouped	group	VERB
fcis-10209	72	15	locally	locally	ADV
fcis-10209	72	16	through	through	ADP
fcis-10209	72	17	patch	patch	NOUN
fcis-10209	72	18	embedding	embed	VERB
fcis-10209	72	19	for	for	ADP
fcis-10209	72	20	subsequent	subsequent	ADJ
fcis-10209	72	21	convolutional	convolutional	ADJ
fcis-10209	72	22	operations	operation	NOUN
fcis-10209	72	23	.	.	PUNCT
fcis-10209	73	1	then	then	ADV
fcis-10209	73	2	,	,	PUNCT
fcis-10209	73	3	3x3	3x3	NUM
fcis-10209	73	4	and	and	CCONJ
fcis-10209	73	5	1x1	1x1	NUM
fcis-10209	73	6	sliding	slide	VERB
fcis-10209	73	7	windows	window	NOUN
fcis-10209	73	8	are	be	AUX
fcis-10209	73	9	used	use	VERB
fcis-10209	73	10	to	to	PART
fcis-10209	73	11	perform	perform	VERB
fcis-10209	73	12	convolutional	convolutional	ADJ
fcis-10209	73	13	operations	operation	NOUN
fcis-10209	73	14	on	on	ADP
fcis-10209	73	15	the	the	DET
fcis-10209	73	16	input	input	NOUN
fcis-10209	73	17	,	,	PUNCT
fcis-10209	73	18	which	which	PRON
fcis-10209	73	19	are	be	AUX
fcis-10209	73	20	activated	activate	VERB
fcis-10209	73	21	through	through	ADP
fcis-10209	73	22	bn	bn	NOUN
fcis-10209	73	23	batch	batch	NOUN
fcis-10209	73	24	standardization	standardization	NOUN
fcis-10209	73	25	and	and	CCONJ
fcis-10209	73	26	gelu	gelu	NOUN
fcis-10209	73	27	functions	function	NOUN
fcis-10209	73	28	.	.	PUNCT
fcis-10209	74	1	subsequently	subsequently	ADV
fcis-10209	74	2	,	,	PUNCT
fcis-10209	74	3	pooling	pool	VERB
fcis-10209	74	4	operations	operation	NOUN
fcis-10209	74	5	are	be	AUX
fcis-10209	74	6	used	use	VERB
fcis-10209	74	7	to	to	PART
fcis-10209	74	8	reduce	reduce	VERB
fcis-10209	74	9	the	the	DET
fcis-10209	74	10	size	size	NOUN
fcis-10209	74	11	of	of	ADP
fcis-10209	74	12	the	the	DET
fcis-10209	74	13	feature	feature	NOUN
fcis-10209	74	14	map	map	NOUN
fcis-10209	74	15	,	,	PUNCT
fcis-10209	74	16	followed	follow	VERB
fcis-10209	74	17	by	by	ADP
fcis-10209	74	18	batch	batch	NOUN
fcis-10209	74	19	normalization	normalization	NOUN
fcis-10209	74	20	and	and	CCONJ
fcis-10209	74	21	finally	finally	ADV
fcis-10209	74	22	a	a	DET
fcis-10209	74	23	fully	fully	ADV
fcis-10209	74	24	connected	connect	VERB
fcis-10209	74	25	layer	layer	NOUN
fcis-10209	74	26	that	that	PRON
fcis-10209	74	27	abstracts	abstract	VERB
fcis-10209	74	28	the	the	DET
fcis-10209	74	29	high	high	ADJ
fcis-10209	74	30	-	-	PUNCT
fcis-10209	74	31	level	level	NOUN
fcis-10209	74	32	features	feature	NOUN
fcis-10209	74	33	of	of	ADP
fcis-10209	74	34	the	the	DET
fcis-10209	74	35	data	datum	NOUN
fcis-10209	74	36	for	for	ADP
fcis-10209	74	37	downstream	downstream	ADJ
fcis-10209	74	38	tasks	task	NOUN
fcis-10209	74	39	.	.	PUNCT
fcis-10209	75	1	fig	fig	NOUN
fcis-10209	75	2	4	4	NUM
fcis-10209	75	3	.	.	PUNCT
fcis-10209	76	1	the	the	DET
fcis-10209	76	2	structure	structure	NOUN
fcis-10209	76	3	of	of	ADP
fcis-10209	76	4	convblock	convblock	PROPN
fcis-10209	76	5	3.3	3.3	NUM
fcis-10209	76	6	.	.	PUNCT
fcis-10209	77	1	transblock	transblock	NOUN
fcis-10209	77	2	in	in	ADP
fcis-10209	77	3	the	the	DET
fcis-10209	77	4	transblock	transblock	NOUN
fcis-10209	77	5	module	module	NOUN
fcis-10209	77	6	of	of	ADP
fcis-10209	77	7	the	the	DET
fcis-10209	77	8	neural	neural	ADJ
fcis-10209	77	9	network	network	NOUN
fcis-10209	77	10	structure	structure	NOUN
fcis-10209	77	11	proposed	propose	VERB
fcis-10209	77	12	in	in	ADP
fcis-10209	77	13	this	this	DET
fcis-10209	77	14	paper	paper	NOUN
fcis-10209	77	15	,	,	PUNCT
fcis-10209	77	16	the	the	DET
fcis-10209	77	17	edgenext	edgenext	NOUN
fcis-10209	77	18	[	[	X
fcis-10209	77	19	15	15	NUM
fcis-10209	77	20	]	]	X
fcis-10209	77	21	architecture	architecture	NOUN
fcis-10209	77	22	is	be	AUX
fcis-10209	77	23	used	use	VERB
fcis-10209	77	24	to	to	PART
fcis-10209	77	25	segment	segment	VERB
fcis-10209	77	26	and	and	CCONJ
fcis-10209	77	27	extract	extract	VERB
fcis-10209	77	28	features	feature	NOUN
fcis-10209	77	29	from	from	ADP
fcis-10209	77	30	input	input	NOUN
fcis-10209	77	31	data	datum	NOUN
fcis-10209	77	32	.	.	PUNCT
fcis-10209	78	1	the	the	DET
fcis-10209	78	2	edgenext	edgenext	NOUN
fcis-10209	78	3	model	model	NOUN
fcis-10209	78	4	is	be	AUX
fcis-10209	78	5	an	an	DET
fcis-10209	78	6	efficient	efficient	ADJ
fcis-10209	78	7	and	and	CCONJ
fcis-10209	78	8	deployable	deployable	ADJ
fcis-10209	78	9	deep	deep	ADJ
fcis-10209	78	10	learning	learning	NOUN
fcis-10209	78	11	architecture	architecture	NOUN
fcis-10209	78	12	that	that	PRON
fcis-10209	78	13	can	can	AUX
fcis-10209	78	14	be	be	AUX
fcis-10209	78	15	applied	apply	VERB
fcis-10209	78	16	to	to	ADP
fcis-10209	78	17	mobile	mobile	ADJ
fcis-10209	78	18	visual	visual	ADJ
fcis-10209	78	19	tasks	task	NOUN
fcis-10209	78	20	.	.	PUNCT
fcis-10209	79	1	the	the	DET
fcis-10209	79	2	model	model	NOUN
fcis-10209	79	3	consists	consist	VERB
fcis-10209	79	4	of	of	ADP
fcis-10209	79	5	two	two	NUM
fcis-10209	79	6	main	main	ADJ
fcis-10209	79	7	components	component	NOUN
fcis-10209	79	8	:	:	PUNCT
fcis-10209	79	9	the	the	DET
fcis-10209	79	10	convolution	convolution	NOUN
fcis-10209	79	11	encoder	encoder	NOUN
fcis-10209	79	12	for	for	ADP
fcis-10209	79	13	extracting	extract	VERB
fcis-10209	79	14	image	image	NOUN
fcis-10209	79	15	features	feature	NOUN
fcis-10209	79	16	,	,	PUNCT
fcis-10209	79	17	and	and	CCONJ
fcis-10209	79	18	the	the	DET
fcis-10209	79	19	stda	stda	NOUN
fcis-10209	79	20	encoder	encoder	NOUN
fcis-10209	79	21	for	for	ADP
fcis-10209	79	22	combining	combine	VERB
fcis-10209	79	23	time	time	NOUN
fcis-10209	79	24	and	and	CCONJ
fcis-10209	79	25	space	space	NOUN
fcis-10209	79	26	information	information	NOUN
fcis-10209	79	27	to	to	PART
fcis-10209	79	28	achieve	achieve	VERB
fcis-10209	79	29	more	more	ADV
fcis-10209	79	30	accurate	accurate	ADJ
fcis-10209	79	31	object	object	NOUN
fcis-10209	79	32	tracking	tracking	NOUN
fcis-10209	79	33	.	.	PUNCT
fcis-10209	80	1	the	the	DET
fcis-10209	80	2	specific	specific	ADJ
fcis-10209	80	3	structure	structure	NOUN
fcis-10209	80	4	of	of	ADP
fcis-10209	80	5	edgenext	edgenext	NOUN
fcis-10209	80	6	can	can	AUX
fcis-10209	80	7	be	be	AUX
fcis-10209	80	8	seen	see	VERB
fcis-10209	80	9	in	in	ADP
fcis-10209	80	10	fig.5	fig.5	PROPN
fcis-10209	80	11	,	,	PUNCT
fcis-10209	80	12	which	which	PRON
fcis-10209	80	13	follows	follow	VERB
fcis-10209	80	14	the	the	DET
fcis-10209	80	15	standard	standard	ADJ
fcis-10209	80	16	"	"	PUNCT
fcis-10209	80	17	four	four	NUM
fcis-10209	80	18	-	-	PUNCT
fcis-10209	80	19	stage	stage	NOUN
fcis-10209	80	20	"	"	PUNCT
fcis-10209	80	21	pyramid	pyramid	NOUN
fcis-10209	80	22	-	-	PUNCT
fcis-10209	80	23	style	style	NOUN
fcis-10209	80	24	design	design	NOUN
fcis-10209	80	25	specification	specification	NOUN
fcis-10209	80	26	and	and	CCONJ
fcis-10209	80	27	includes	include	VERB
fcis-10209	80	28	two	two	NUM
fcis-10209	80	29	core	core	NOUN
fcis-10209	80	30	modules	module	NOUN
fcis-10209	80	31	:	:	PUNCT
fcis-10209	80	32	the	the	DET
fcis-10209	80	33	convolutional	convolutional	ADJ
fcis-10209	80	34	encoder	encoder	NOUN
fcis-10209	80	35	and	and	CCONJ
fcis-10209	80	36	the	the	DET
fcis-10209	80	37	stda	stda	NOUN
fcis-10209	80	38	encoder	encoder	NOUN
fcis-10209	80	39	.	.	PUNCT
fcis-10209	81	1	at	at	ADP
fcis-10209	81	2	the	the	DET
fcis-10209	81	3	beginning	beginning	NOUN
fcis-10209	81	4	of	of	ADP
fcis-10209	81	5	the	the	DET
fcis-10209	81	6	overall	overall	ADJ
fcis-10209	81	7	structure	structure	NOUN
fcis-10209	81	8	,	,	PUNCT
fcis-10209	81	9	a	a	DET
fcis-10209	81	10	"	"	PUNCT
fcis-10209	81	11	patchify	patchify	NOUN
fcis-10209	81	12	"	"	PUNCT
fcis-10209	81	13	strategy	strategy	NOUN
fcis-10209	81	14	similar	similar	ADJ
fcis-10209	81	15	to	to	ADP
fcis-10209	81	16	vit	vit	NOUN
fcis-10209	81	17	and	and	CCONJ
fcis-10209	81	18	swintransformer	swintransformer	NOUN
fcis-10209	81	19	is	be	AUX
fcis-10209	81	20	used	use	VERB
fcis-10209	81	21	,	,	PUNCT
fcis-10209	81	22	employing	employ	VERB
fcis-10209	81	23	nonoverlapping	nonoverlapping	ADJ
fcis-10209	81	24	convolutions	convolution	NOUN
fcis-10209	81	25	of	of	ADP
fcis-10209	81	26	size	size	NOUN
fcis-10209	81	27	4x4	4x4	NUM
fcis-10209	81	28	to	to	PART
fcis-10209	81	29	achieve	achieve	VERB
fcis-10209	81	30	better	well	ADJ
fcis-10209	81	31	pooling	pool	VERB
fcis-10209	81	32	effects	effect	NOUN
fcis-10209	81	33	,	,	PUNCT
fcis-10209	81	34	resulting	result	VERB
fcis-10209	81	35	in	in	ADP
fcis-10209	81	36	an	an	DET
fcis-10209	81	37	output	output	NOUN
fcis-10209	81	38	size	size	NOUN
fcis-10209	81	39	of	of	ADP
fcis-10209	81	40	h/4xw/4xc1	h/4xw/4xc1	NOUN
fcis-10209	81	41	.	.	PUNCT
fcis-10209	82	1	next	next	ADJ
fcis-10209	82	2	,	,	PUNCT
fcis-10209	82	3	three	three	NUM
fcis-10209	82	4	stacked	stacked	ADJ
fcis-10209	82	5	convolutional	convolutional	ADJ
fcis-10209	82	6	encoders	encoder	NOUN
fcis-10209	82	7	of	of	ADP
fcis-10209	82	8	kernel	kernel	PROPN
fcis-10209	82	9	size	size	NOUN
fcis-10209	82	10	3x3	3x3	NUM
fcis-10209	82	11	are	be	AUX
fcis-10209	82	12	used	use	VERB
fcis-10209	82	13	to	to	PART
fcis-10209	82	14	extract	extract	VERB
fcis-10209	82	15	local	local	ADJ
fcis-10209	82	16	features	feature	NOUN
fcis-10209	82	17	without	without	ADP
fcis-10209	82	18	changing	change	VERB
fcis-10209	82	19	the	the	DET
fcis-10209	82	20	feature	feature	NOUN
fcis-10209	82	21	map	map	NOUN
fcis-10209	82	22	size	size	NOUN
fcis-10209	82	23	.	.	PUNCT
fcis-10209	83	1	in	in	ADP
fcis-10209	83	2	stage	stage	NOUN
fcis-10209	83	3	2	2	NUM
fcis-10209	83	4	,	,	PUNCT
fcis-10209	83	5	down	down	ADV
fcis-10209	83	6	-	-	PUNCT
fcis-10209	83	7	sampling	sampling	NOUN
fcis-10209	83	8	is	be	AUX
fcis-10209	83	9	achieved	achieve	VERB
fcis-10209	83	10	through	through	ADP
fcis-10209	83	11	convolution	convolution	NOUN
fcis-10209	83	12	with	with	ADP
fcis-10209	83	13	a	a	DET
fcis-10209	83	14	stride	stride	NOUN
fcis-10209	83	15	of	of	ADP
fcis-10209	83	16	(	(	PUNCT
fcis-10209	83	17	2,2	2,2	NUM
fcis-10209	83	18	)	)	PUNCT
fcis-10209	83	19	,	,	PUNCT
fcis-10209	83	20	after	after	ADP
fcis-10209	83	21	which	which	PRON
fcis-10209	83	22	two	two	NUM
fcis-10209	83	23	consecutive	consecutive	ADJ
fcis-10209	83	24	encoder	encoder	NOUN
fcis-10209	83	25	layers	layer	NOUN
fcis-10209	83	26	with	with	ADP
fcis-10209	83	27	kernel	kernel	PROPN
fcis-10209	83	28	size	size	NOUN
fcis-10209	83	29	5x5	5x5	NUM
fcis-10209	83	30	are	be	AUX
fcis-10209	83	31	stacked	stack	VERB
fcis-10209	83	32	and	and	CCONJ
fcis-10209	83	33	position	position	NOUN
fcis-10209	83	34	encoding	encoding	NOUN
fcis-10209	83	35	is	be	AUX
fcis-10209	83	36	added	add	VERB
fcis-10209	83	37	via	via	ADP
fcis-10209	83	38	element	element	ADJ
fcis-10209	83	39	-	-	ADJ
fcis-10209	83	40	wise	wise	ADJ
fcis-10209	83	41	addition	addition	NOUN
fcis-10209	83	42	before	before	ADP
fcis-10209	83	43	entering	enter	VERB
fcis-10209	83	44	the	the	DET
fcis-10209	83	45	stda	stda	NOUN
fcis-10209	83	46	module	module	NOUN
fcis-10209	83	47	.	.	PUNCT
fcis-10209	84	1	in	in	ADP
fcis-10209	84	2	stages	stage	NOUN
fcis-10209	84	3	3	3	NUM
fcis-10209	84	4	and	and	CCONJ
fcis-10209	84	5	4	4	NUM
fcis-10209	84	6	,	,	PUNCT
fcis-10209	84	7	summing	sum	VERB
fcis-10209	84	8	convolutional	convolutional	ADJ
fcis-10209	84	9	encoding	encoding	NOUN
fcis-10209	84	10	layers	layer	NOUN
fcis-10209	84	11	are	be	AUX
fcis-10209	84	12	used	use	VERB
fcis-10209	84	13	,	,	PUNCT
fcis-10209	84	14	and	and	CCONJ
fcis-10209	84	15	different	different	ADJ
fcis-10209	84	16	kernel	kernel	NOUN
fcis-10209	84	17	sizes	size	NOUN
fcis-10209	84	18	are	be	AUX
fcis-10209	84	19	applied	apply	VERB
fcis-10209	84	20	to	to	PART
fcis-10209	84	21	implement	implement	VERB
fcis-10209	84	22	an	an	DET
fcis-10209	84	23	adaptive	adaptive	ADJ
fcis-10209	84	24	kernel	kernel	NOUN
fcis-10209	84	25	size	size	NOUN
fcis-10209	84	26	mechanism	mechanism	NOUN
fcis-10209	84	27	.	.	PUNCT
fcis-10209	85	1	this	this	DET
fcis-10209	85	2	design	design	NOUN
fcis-10209	85	3	was	be	AUX
fcis-10209	85	4	adopted	adopt	VERB
fcis-10209	85	5	to	to	PART
fcis-10209	85	6	increase	increase	VERB
fcis-10209	85	7	the	the	DET
fcis-10209	85	8	cnn	cnn	PROPN
fcis-10209	85	9	's	's	PART
fcis-10209	85	10	local	local	ADJ
fcis-10209	85	11	receptive	receptive	ADJ
fcis-10209	85	12	field	field	NOUN
fcis-10209	85	13	and	and	CCONJ
fcis-10209	85	14	improve	improve	VERB
fcis-10209	85	15	model	model	NOUN
fcis-10209	85	16	performance	performance	NOUN
fcis-10209	85	17	while	while	SCONJ
fcis-10209	85	18	avoiding	avoid	VERB
fcis-10209	85	19	the	the	DET
fcis-10209	85	20	expensive	expensive	ADJ
fcis-10209	85	21	computational	computational	ADJ
fcis-10209	85	22	cost	cost	NOUN
fcis-10209	85	23	associated	associate	VERB
fcis-10209	85	24	with	with	ADP
fcis-10209	85	25	directly	directly	ADV
fcis-10209	85	26	using	use	VERB
fcis-10209	85	27	larger	large	ADJ
fcis-10209	85	28	kernels	kernel	NOUN
fcis-10209	85	29	.	.	PUNCT
fcis-10209	86	1	therefore	therefore	ADV
fcis-10209	86	2	,	,	PUNCT
fcis-10209	86	3	the	the	DET
fcis-10209	86	4	pyramid	pyramid	NOUN
fcis-10209	86	5	-	-	PUNCT
fcis-10209	86	6	style	style	NOUN
fcis-10209	86	7	design	design	NOUN
fcis-10209	86	8	is	be	AUX
fcis-10209	86	9	a	a	DET
fcis-10209	86	10	reasonable	reasonable	ADJ
fcis-10209	86	11	approach	approach	NOUN
fcis-10209	86	12	.	.	PUNCT
fcis-10209	87	1	furthermore	furthermore	ADV
fcis-10209	87	2	,	,	PUNCT
fcis-10209	87	3	adding	add	VERB
fcis-10209	87	4	position	position	NOUN
fcis-10209	87	5	encoding	encode	VERB
fcis-10209	87	6	only	only	ADV
fcis-10209	87	7	once	once	ADV
fcis-10209	87	8	in	in	ADP
fcis-10209	87	9	the	the	DET
fcis-10209	87	10	four	four	NUM
fcis-10209	87	11	stages	stage	NOUN
fcis-10209	87	12	reduces	reduce	VERB
fcis-10209	87	13	the	the	DET
fcis-10209	87	14	impact	impact	NOUN
fcis-10209	87	15	on	on	ADP
fcis-10209	87	16	detection	detection	NOUN
fcis-10209	87	17	,	,	PUNCT
fcis-10209	87	18	segmentation	segmentation	NOUN
fcis-10209	87	19	,	,	PUNCT
fcis-10209	87	20	and	and	CCONJ
fcis-10209	87	21	other	other	ADJ
fcis-10209	87	22	tasks	task	NOUN
fcis-10209	87	23	while	while	SCONJ
fcis-10209	87	24	improving	improve	VERB
fcis-10209	87	25	model	model	NOUN
fcis-10209	87	26	inference	inference	NOUN
fcis-10209	87	27	speed	speed	NOUN
fcis-10209	87	28	.	.	PUNCT
fcis-10209	88	1	fig	fig	NOUN
fcis-10209	88	2	5	5	NUM
fcis-10209	88	3	.	.	PUNCT
fcis-10209	89	1	the	the	DET
fcis-10209	89	2	schematic	schematic	ADJ
fcis-10209	89	3	diagram	diagram	NOUN
fcis-10209	89	4	of	of	ADP
fcis-10209	89	5	the	the	DET
fcis-10209	89	6	edgenext	edgenext	NOUN
fcis-10209	89	7	architecture	architecture	NOUN
fcis-10209	89	8	3.4	3.4	NUM
fcis-10209	89	9	.	.	PUNCT
fcis-10209	90	1	sdta	sdta	NOUN
fcis-10209	90	2	(	(	PUNCT
fcis-10209	90	3	segmentation	segmentation	NOUN
fcis-10209	90	4	depthwise	depthwise	NOUN
fcis-10209	90	5	transpose	transpose	NOUN
fcis-10209	90	6	attention	attention	NOUN
fcis-10209	90	7	)	)	PUNCT
fcis-10209	90	8	when	when	SCONJ
fcis-10209	90	9	trying	try	VERB
fcis-10209	90	10	to	to	PART
fcis-10209	90	11	combine	combine	VERB
fcis-10209	90	12	the	the	DET
fcis-10209	90	13	excellent	excellent	ADJ
fcis-10209	90	14	characteristics	characteristic	NOUN
fcis-10209	90	15	of	of	ADP
fcis-10209	90	16	vision	vision	NOUN
fcis-10209	90	17	transformer	transformer	NOUN
fcis-10209	90	18	(	(	PUNCT
fcis-10209	90	19	vit	vit	NOUN
fcis-10209	90	20	)	)	PUNCT
fcis-10209	90	21	and	and	CCONJ
fcis-10209	90	22	cnn	cnn	PROPN
fcis-10209	90	23	,	,	PUNCT
fcis-10209	90	24	any	any	DET
fcis-10209	90	25	combination	combination	NOUN
fcis-10209	90	26	method	method	NOUN
fcis-10209	90	27	inevitably	inevitably	ADV
fcis-10209	90	28	brings	bring	VERB
fcis-10209	90	29	a	a	DET
fcis-10209	90	30	new	new	ADJ
fcis-10209	90	31	problem	problem	NOUN
fcis-10209	90	32	:	:	PUNCT
fcis-10209	90	33	due	due	ADP
fcis-10209	90	34	to	to	ADP
fcis-10209	90	35	the	the	DET
fcis-10209	90	36	characteristics	characteristic	NOUN
fcis-10209	90	37	of	of	ADP
fcis-10209	90	38	self	self	NOUN
fcis-10209	90	39	-	-	PUNCT
fcis-10209	90	40	attention	attention	NOUN
fcis-10209	90	41	calculation	calculation	NOUN
fcis-10209	90	42	,	,	PUNCT
fcis-10209	90	43	the	the	DET
fcis-10209	90	44	inference	inference	NOUN
fcis-10209	90	45	speed	speed	NOUN
fcis-10209	90	46	of	of	ADP
fcis-10209	90	47	the	the	DET
fcis-10209	90	48	architecture	architecture	NOUN
fcis-10209	90	49	is	be	AUX
fcis-10209	90	50	severely	severely	ADV
fcis-10209	90	51	limited	limited	ADJ
fcis-10209	90	52	.	.	PUNCT
fcis-10209	91	1	therefore	therefore	ADV
fcis-10209	91	2	,	,	PUNCT
fcis-10209	91	3	in	in	ADP
fcis-10209	91	4	order	order	NOUN
fcis-10209	91	5	to	to	PART
fcis-10209	91	6	make	make	VERB
fcis-10209	91	7	the	the	DET
fcis-10209	91	8	model	model	NOUN
fcis-10209	91	9	balance	balance	NOUN
fcis-10209	91	10	performance	performance	NOUN
fcis-10209	91	11	and	and	CCONJ
fcis-10209	91	12	inference	inference	NOUN
fcis-10209	91	13	speed	speed	NOUN
fcis-10209	91	14	,	,	PUNCT
fcis-10209	91	15	the	the	DET
fcis-10209	91	16	authors	author	NOUN
fcis-10209	91	17	of	of	ADP
fcis-10209	91	18	edgenext	edgenext	NOUN
fcis-10209	91	19	introduce	introduce	VERB
fcis-10209	91	20	a	a	DET
fcis-10209	91	21	segmentation	segmentation	NOUN
fcis-10209	91	22	depthwise	depthwise	NOUN
fcis-10209	91	23	transpose	transpose	NOUN
fcis-10209	91	24	attention	attention	NOUN
fcis-10209	91	25	(	(	PUNCT
fcis-10209	91	26	sdta[15	sdta[15	X
fcis-10209	91	27	]	]	NOUN
fcis-10209	91	28	)	)	PUNCT
fcis-10209	91	29	encoder	encoder	NOUN
fcis-10209	91	30	as	as	SCONJ
fcis-10209	91	31	shown	show	VERB
fcis-10209	91	32	in	in	ADP
fcis-10209	91	33	fig.6	fig.6	PROPN
fcis-10209	91	34	,	,	PUNCT
fcis-10209	91	35	which	which	PRON
fcis-10209	91	36	achieves	achieve	VERB
fcis-10209	91	37	efficient	efficient	ADJ
fcis-10209	91	38	integration	integration	NOUN
fcis-10209	91	39	without	without	ADP
fcis-10209	91	40	increasing	increase	VERB
fcis-10209	91	41	additional	additional	ADJ
fcis-10209	91	42	parameter	parameter	NOUN
fcis-10209	91	43	volume	volume	NOUN
fcis-10209	91	44	and	and	CCONJ
fcis-10209	91	45	multiplication	multiplication	NOUN
fcis-10209	91	46	-	-	PUNCT
fcis-10209	91	47	addition	addition	NOUN
fcis-10209	91	48	operation	operation	NOUN
fcis-10209	91	49	amount	amount	NOUN
fcis-10209	91	50	(	(	PUNCT
fcis-10209	91	51	madds	madds	PROPN
fcis-10209	91	52	)	)	PUNCT
fcis-10209	91	53	.	.	PUNCT
fcis-10209	92	1	the	the	DET
fcis-10209	92	2	sdta	sdta	PROPN
fcis-10209	92	3	encoder	encoder	NOUN
fcis-10209	92	4	consists	consist	VERB
fcis-10209	92	5	of	of	ADP
fcis-10209	92	6	two	two	NUM
fcis-10209	92	7	main	main	ADJ
fcis-10209	92	8	components	component	NOUN
fcis-10209	92	9	:	:	PUNCT
fcis-10209	92	10	the	the	DET
fcis-10209	92	11	feature	feature	NOUN
fcis-10209	92	12	encoding	encoding	NOUN
fcis-10209	92	13	module	module	NOUN
fcis-10209	92	14	and	and	CCONJ
fcis-10209	92	15	the	the	DET
fcis-10209	92	16	self	self	NOUN
fcis-10209	92	17	-	-	PUNCT
fcis-10209	92	18	attention	attention	NOUN
fcis-10209	92	19	calculation	calculation	NOUN
fcis-10209	92	20	module	module	NOUN
fcis-10209	92	21	.	.	PUNCT
fcis-10209	93	1	in	in	ADP
fcis-10209	93	2	the	the	DET
fcis-10209	93	3	feature	feature	NOUN
fcis-10209	93	4	encoding	encoding	NOUN
fcis-10209	93	5	module	module	NOUN
fcis-10209	93	6	,	,	PUNCT
fcis-10209	93	7	the	the	DET
fcis-10209	93	8	input	input	NOUN
fcis-10209	93	9	features	feature	NOUN
fcis-10209	93	10	are	be	AUX
fcis-10209	93	11	first	first	ADV
fcis-10209	93	12	divided	divide	VERB
fcis-10209	93	13	into	into	ADP
fcis-10209	93	14	s	s	PROPN
fcis-10209	93	15	subsets	subset	NOUN
fcis-10209	93	16	,	,	PUNCT
fcis-10209	93	17	each	each	PRON
fcis-10209	93	18	with	with	ADP
fcis-10209	93	19	an	an	DET
fcis-10209	93	20	equal	equal	ADJ
fcis-10209	93	21	size	size	NOUN
fcis-10209	93	22	,	,	PUNCT
fcis-10209	93	23	and	and	CCONJ
fcis-10209	93	24	each	each	DET
fcis-10209	93	25	subset	subset	NOUN
fcis-10209	93	26	is	be	AUX
fcis-10209	93	27	encoded	encode	VERB
fcis-10209	93	28	by	by	ADP
fcis-10209	93	29	fusing	fuse	VERB
fcis-10209	93	30	the	the	DET
fcis-10209	93	31	output	output	NOUN
fcis-10209	93	32	features	feature	NOUN
fcis-10209	93	33	of	of	ADP
fcis-10209	93	34	the	the	DET
fcis-10209	93	35	previous	previous	ADJ
fcis-10209	93	36	subset	subset	NOUN
fcis-10209	93	37	and	and	CCONJ
fcis-10209	93	38	then	then	ADV
fcis-10209	93	39	passing	pass	VERB
fcis-10209	93	40	through	through	ADP
fcis-10209	93	41	a	a	DET
fcis-10209	93	42	3x3	3x3	NUM
fcis-10209	93	43	depth	depth	NOUN
fcis-10209	93	44	-	-	PUNCT
fcis-10209	93	45	wise	wise	ADJ
fcis-10209	93	46	convolution	convolution	NOUN
fcis-10209	93	47	.	.	PUNCT
fcis-10209	94	1	the	the	DET
fcis-10209	94	2	sdta	sdta	PROPN
fcis-10209	94	3	encoder	encoder	NOUN
fcis-10209	94	4	uses	use	VERB
fcis-10209	94	5	an	an	DET
fcis-10209	94	6	adaptive	adaptive	ADJ
fcis-10209	94	7	number	number	NOUN
fcis-10209	94	8	of	of	ADP
fcis-10209	94	9	subsets	subset	NOUN
fcis-10209	94	10	to	to	PART
fcis-10209	94	11	allow	allow	VERB
fcis-10209	94	12	flexibility	flexibility	NOUN
fcis-10209	94	13	in	in	ADP
fcis-10209	94	14	feature	feature	NOUN
fcis-10209	94	15	encoding	encoding	NOUN
fcis-10209	94	16	,	,	PUNCT
fcis-10209	94	17	and	and	CCONJ
fcis-10209	94	18	the	the	DET
fcis-10209	94	19	output	output	NOUN
fcis-10209	94	20	features	feature	NOUN
fcis-10209	94	21	of	of	ADP
fcis-10209	94	22	s	s	NOUN
fcis-10209	94	23	subsets	subset	NOUN
fcis-10209	94	24	are	be	AUX
fcis-10209	94	25	concatenated	concatenate	VERB
fcis-10209	94	26	to	to	PART
fcis-10209	94	27	obtain	obtain	VERB
fcis-10209	94	28	output	output	NOUN
fcis-10209	94	29	features	feature	NOUN
fcis-10209	94	30	with	with	ADP
fcis-10209	94	31	multiple	multiple	ADJ
fcis-10209	94	32	scales	scale	NOUN
fcis-10209	94	33	.	.	PUNCT
fcis-10209	95	1	this	this	DET
fcis-10209	95	2	module	module	NOUN
fcis-10209	95	3	tries	try	VERB
fcis-10209	95	4	to	to	PART
fcis-10209	95	5	learn	learn	VERB
fcis-10209	95	6	an	an	DET
fcis-10209	95	7	adaptive	adaptive	ADJ
fcis-10209	95	8	multiscale	multiscale	ADJ
fcis-10209	95	9	feature	feature	NOUN
fcis-10209	95	10	representation	representation	NOUN
fcis-10209	95	11	,	,	PUNCT
fcis-10209	95	12	encoding	encode	VERB
fcis-10209	95	13	different	different	ADJ
fcis-10209	95	14	spatial	spatial	ADJ
fcis-10209	95	15	levels	level	NOUN
fcis-10209	95	16	on	on	ADP
fcis-10209	95	17	the	the	DET
fcis-10209	95	18	input	input	NOUN
fcis-10209	95	19	image	image	NOUN
fcis-10209	95	20	and	and	CCONJ
fcis-10209	95	21	intuitively	intuitively	ADV
fcis-10209	95	22	encoding	encode	VERB
fcis-10209	95	23	the	the	DET
fcis-10209	95	24	global	global	ADJ
fcis-10209	95	25	image	image	NOUN
fcis-10209	95	26	representation	representation	NOUN
fcis-10209	95	27	.	.	PUNCT
fcis-10209	96	1	the	the	DET
fcis-10209	96	2	self	self	NOUN
fcis-10209	96	3	-	-	PUNCT
fcis-10209	96	4	attention	attention	NOUN
fcis-10209	96	5	calculation	calculation	NOUN
fcis-10209	96	6	module	module	NOUN
fcis-10209	96	7	is	be	AUX
fcis-10209	96	8	the	the	DET
fcis-10209	96	9	second	second	ADJ
fcis-10209	96	10	component	component	NOUN
fcis-10209	96	11	of	of	ADP
fcis-10209	96	12	the	the	DET
fcis-10209	96	13	sdta	sdta	PROPN
fcis-10209	96	14	encoder	encoder	NOUN
fcis-10209	96	15	,	,	PUNCT
fcis-10209	96	16	which	which	PRON
fcis-10209	96	17	calculates	calculate	VERB
fcis-10209	96	18	attention	attention	NOUN
fcis-10209	96	19	on	on	ADP
fcis-10209	96	20	the	the	DET
fcis-10209	96	21	input	input	NOUN
fcis-10209	96	22	features	feature	VERB
fcis-10209	96	23	to	to	PART
fcis-10209	96	24	obtain	obtain	VERB
fcis-10209	96	25	important	important	ADJ
fcis-10209	96	26	features	feature	NOUN
fcis-10209	96	27	.	.	PUNCT
fcis-10209	97	1	in	in	ADP
fcis-10209	97	2	this	this	DET
fcis-10209	97	3	module	module	NOUN
fcis-10209	97	4	,	,	PUNCT
fcis-10209	97	5	the	the	DET
fcis-10209	97	6	input	input	NOUN
fcis-10209	97	7	features	feature	NOUN
fcis-10209	97	8	are	be	AUX
fcis-10209	97	9	first	first	ADV
fcis-10209	97	10	transformed	transform	VERB
fcis-10209	97	11	into	into	ADP
fcis-10209	97	12	q	q	PROPN
fcis-10209	97	13	,	,	PUNCT
fcis-10209	97	14	k	k	NOUN
fcis-10209	97	15	,	,	PUNCT
fcis-10209	97	16	and	and	CCONJ
fcis-10209	97	17	v	v	ADP
fcis-10209	97	18	tensors	tensor	NOUN
fcis-10209	97	19	through	through	ADP
fcis-10209	97	20	three	three	NUM
fcis-10209	97	21	linear	linear	ADJ
fcis-10209	97	22	layers	layer	NOUN
fcis-10209	97	23	,	,	PUNCT
fcis-10209	97	24	and	and	CCONJ
fcis-10209	97	25	l2	l2	NOUN
fcis-10209	97	26	normalization	normalization	NOUN
fcis-10209	97	27	is	be	AUX
fcis-10209	97	28	performed	perform	VERB
fcis-10209	97	29	on	on	ADP
fcis-10209	97	30	the	the	DET
fcis-10209	97	31	q	q	NOUN
fcis-10209	97	32	and	and	CCONJ
fcis-10209	97	33	k	k	PROPN
fcis-10209	97	34	tensors	tensor	NOUN
fcis-10209	97	35	before	before	ADP
fcis-10209	97	36	calculating	calculate	VERB
fcis-10209	97	37	the	the	DET
fcis-10209	97	38	cross	cross	ADJ
fcis-10209	97	39	-	-	ADJ
fcis-10209	97	40	covariance	covariance	ADJ
fcis-10209	97	41	attention	attention	NOUN
fcis-10209	97	42	to	to	PART
fcis-10209	97	43	stabilize	stabilize	VERB
fcis-10209	97	44	the	the	DET
fcis-10209	97	45	training	training	NOUN
fcis-10209	97	46	.	.	PUNCT
fcis-10209	98	1	this	this	DET
fcis-10209	98	2	module	module	NOUN
fcis-10209	98	3	uses	use	VERB
fcis-10209	98	4	dot	dot	NOUN
fcis-10209	98	5	product	product	NOUN
fcis-10209	98	6	operation	operation	NOUN
fcis-10209	98	7	to	to	PART
fcis-10209	98	8	calculate	calculate	VERB
fcis-10209	98	9	the	the	DET
fcis-10209	98	10	dot	dot	NOUN
fcis-10209	98	11	product	product	NOUN
fcis-10209	98	12	between	between	ADP
fcis-10209	98	13	q	q	PROPN
fcis-10209	98	14	t	t	PROPN
fcis-10209	98	15	q	q	PROPN
fcis-10209	99	1	and	and	CCONJ
fcis-10209	99	2	k	k	PROPN
fcis-10209	99	3	k	k	PROPN
fcis-10209	99	4	t	t	PROPN
fcis-10209	99	5	in	in	ADP
fcis-10209	99	6	the	the	DET
fcis-10209	99	7	channel	channel	NOUN
fcis-10209	99	8	dimension	dimension	NOUN
fcis-10209	99	9	,	,	PUNCT
fcis-10209	99	10	resulting	result	VERB
fcis-10209	99	11	in	in	ADP
fcis-10209	99	12	an	an	DET
fcis-10209	99	13	attention	attention	NOUN
fcis-10209	99	14	matrix	matrix	NOUN
fcis-10209	99	15	of	of	ADP
fcis-10209	99	16	c	c	PROPN
fcis-10209	99	17	x	x	PROPN
fcis-10209	99	18	c.	c.	PROPN
fcis-10209	99	19	the	the	DET
fcis-10209	99	20	attention	attention	NOUN
fcis-10209	99	21	matrix	matrix	NOUN
fcis-10209	99	22	is	be	AUX
fcis-10209	99	23	further	far	ADV
fcis-10209	99	24	processed	process	VERB
fcis-10209	99	25	by	by	ADP
fcis-10209	99	26	softmax	softmax	NOUN
fcis-10209	99	27	and	and	CCONJ
fcis-10209	99	28	dot	dot	NOUN
fcis-10209	99	29	product	product	NOUN
fcis-10209	99	30	with	with	ADP
fcis-10209	99	31	the	the	DET
fcis-10209	99	32	v	v	NOUN
fcis-10209	99	33	tensor	tensor	NOUN
fcis-10209	99	34	to	to	PART
fcis-10209	99	35	calculate	calculate	VERB
fcis-10209	99	36	the	the	DET
fcis-10209	99	37	final	final	ADJ
fcis-10209	99	38	attention	attention	NOUN
fcis-10209	99	39	map	map	NOUN
fcis-10209	99	40	.	.	PUNCT
fcis-10209	100	1	finally	finally	ADV
fcis-10209	100	2	,	,	PUNCT
fcis-10209	100	3	two	two	NUM
fcis-10209	100	4	1x1	1x1	NUM
fcis-10209	100	5	point	point	ADV
fcis-10209	100	6	-	-	PUNCT
fcis-10209	100	7	wise	wise	ADJ
fcis-10209	100	8	75	75	NUM
fcis-10209	100	9	operations	operation	NOUN
fcis-10209	100	10	are	be	AUX
fcis-10209	100	11	used	use	VERB
fcis-10209	100	12	,	,	PUNCT
fcis-10209	100	13	with	with	ADP
fcis-10209	100	14	ln	ln	ADJ
fcis-10209	100	15	and	and	CCONJ
fcis-10209	100	16	gelu	gelu	ADJ
fcis-10209	100	17	activation	activation	NOUN
fcis-10209	100	18	layers	layer	NOUN
fcis-10209	100	19	to	to	PART
fcis-10209	100	20	produce	produce	VERB
fcis-10209	100	21	non	non	ADJ
fcis-10209	100	22	-	-	ADJ
fcis-10209	100	23	linear	linear	ADJ
fcis-10209	100	24	features	feature	NOUN
fcis-10209	100	25	.	.	PUNCT
fcis-10209	101	1	to	to	PART
fcis-10209	101	2	avoid	avoid	VERB
fcis-10209	101	3	complexity	complexity	NOUN
fcis-10209	101	4	problems	problem	NOUN
fcis-10209	101	5	,	,	PUNCT
fcis-10209	101	6	this	this	DET
fcis-10209	101	7	module	module	NOUN
fcis-10209	101	8	borrows	borrow	VERB
fcis-10209	101	9	from	from	ADP
fcis-10209	101	10	the	the	DET
fcis-10209	101	11	transpose	transpose	NOUN
fcis-10209	101	12	qk	qk	ADP
fcis-10209	101	13	attention	attention	NOUN
fcis-10209	101	14	feature	feature	NOUN
fcis-10209	101	15	map	map	NOUN
fcis-10209	101	16	and	and	CCONJ
fcis-10209	101	17	uses	use	VERB
fcis-10209	101	18	msa	msa	PROPN
fcis-10209	101	19	's	's	PART
fcis-10209	101	20	dot	dot	NOUN
fcis-10209	101	21	product	product	NOUN
fcis-10209	101	22	operation	operation	NOUN
fcis-10209	101	23	in	in	ADP
fcis-10209	101	24	the	the	DET
fcis-10209	101	25	channel	channel	NOUN
fcis-10209	101	26	dimension	dimension	NOUN
fcis-10209	101	27	to	to	PART
fcis-10209	101	28	achieve	achieve	VERB
fcis-10209	101	29	linear	linear	PROPN
fcis-10209	101	30	complexity	complexity	NOUN
fcis-10209	101	31	.	.	PUNCT
fcis-10209	102	1	fig	fig	NOUN
fcis-10209	102	2	6	6	NUM
fcis-10209	102	3	.	.	PUNCT
fcis-10209	103	1	the	the	DET
fcis-10209	103	2	stda	stda	NOUN
fcis-10209	103	3	encoder	encoder	NOUN
fcis-10209	103	4	structure	structure	NOUN
fcis-10209	103	5	diagram	diagram	PROPN
fcis-10209	103	6	4	4	NUM
fcis-10209	103	7	.	.	PUNCT
fcis-10209	103	8	results	result	NOUN
fcis-10209	103	9	and	and	CCONJ
fcis-10209	103	10	analysis	analysis	NOUN
fcis-10209	103	11	4.1	4.1	NUM
fcis-10209	103	12	.	.	PUNCT
fcis-10209	104	1	experimental	experimental	ADJ
fcis-10209	104	2	configuration	configuration	NOUN
fcis-10209	104	3	the	the	DET
fcis-10209	104	4	experiments	experiment	NOUN
fcis-10209	104	5	were	be	AUX
fcis-10209	104	6	conducted	conduct	VERB
fcis-10209	104	7	on	on	ADP
fcis-10209	104	8	a	a	DET
fcis-10209	104	9	windows	window	NOUN
fcis-10209	104	10	10	10	NUM
fcis-10209	104	11	operating	operating	NOUN
fcis-10209	104	12	system	system	NOUN
fcis-10209	104	13	,	,	PUNCT
fcis-10209	104	14	with	with	ADP
fcis-10209	104	15	an	an	DET
fcis-10209	104	16	amd	amd	ADJ
fcis-10209	104	17	ryzen	ryzen	ADJ
fcis-10209	104	18	7	7	NUM
fcis-10209	104	19	4800h	4800h	NUM
fcis-10209	104	20	cpu	cpu	NOUN
fcis-10209	104	21	and	and	CCONJ
fcis-10209	104	22	an	an	DET
fcis-10209	104	23	nvidia	nvidia	PROPN
fcis-10209	104	24	geforce	geforce	NOUN
fcis-10209	104	25	gtx	gtx	PROPN
fcis-10209	104	26	1650	1650	NUM
fcis-10209	104	27	gpu	gpu	X
fcis-10209	104	28	.	.	PUNCT
fcis-10209	105	1	the	the	DET
fcis-10209	105	2	programming	programming	NOUN
fcis-10209	105	3	environment	environment	NOUN
fcis-10209	105	4	used	use	VERB
fcis-10209	105	5	for	for	ADP
fcis-10209	105	6	training	training	NOUN
fcis-10209	105	7	was	be	AUX
fcis-10209	105	8	python	python	NOUN
fcis-10209	105	9	3.7	3.7	NUM
fcis-10209	105	10	and	and	CCONJ
fcis-10209	105	11	the	the	DET
fcis-10209	105	12	deep	deep	ADJ
fcis-10209	105	13	learning	learning	NOUN
fcis-10209	105	14	framework	framework	NOUN
fcis-10209	105	15	was	be	AUX
fcis-10209	105	16	pytorch	pytorch	NOUN
fcis-10209	105	17	1.8.0	1.8.0	NUM
fcis-10209	105	18	.	.	PUNCT
fcis-10209	106	1	4.2	4.2	NUM
fcis-10209	106	2	.	.	PUNCT
fcis-10209	107	1	training	training	NOUN
fcis-10209	107	2	settings	setting	NOUN
fcis-10209	107	3	the	the	DET
fcis-10209	107	4	sgd	sgd	PROPN
fcis-10209	107	5	optimizer	optimizer	NOUN
fcis-10209	107	6	was	be	AUX
fcis-10209	107	7	selected	select	VERB
fcis-10209	107	8	for	for	ADP
fcis-10209	107	9	model	model	NOUN
fcis-10209	107	10	training	training	NOUN
fcis-10209	107	11	,	,	PUNCT
fcis-10209	107	12	with	with	ADP
fcis-10209	107	13	a	a	DET
fcis-10209	107	14	learning	learn	VERB
fcis-10209	107	15	rate	rate	NOUN
fcis-10209	107	16	initialized	initialize	VERB
fcis-10209	107	17	to	to	ADP
fcis-10209	107	18	0.001	0.001	NUM
fcis-10209	107	19	,	,	PUNCT
fcis-10209	107	20	momentum	momentum	NOUN
fcis-10209	107	21	of	of	ADP
fcis-10209	107	22	0.9	0.9	NUM
fcis-10209	107	23	,	,	PUNCT
fcis-10209	107	24	weight	weight	NOUN
fcis-10209	107	25	decay	decay	NOUN
fcis-10209	107	26	of	of	ADP
fcis-10209	107	27	5e-5	5e-5	NUM
fcis-10209	107	28	,	,	PUNCT
fcis-10209	107	29	batch_size	batch_size	NOUN
fcis-10209	107	30	of	of	ADP
fcis-10209	107	31	8	8	NUM
fcis-10209	107	32	,	,	PUNCT
fcis-10209	107	33	and	and	CCONJ
fcis-10209	107	34	100	100	NUM
fcis-10209	107	35	iterations	iteration	NOUN
fcis-10209	107	36	.	.	PUNCT
fcis-10209	108	1	additionally	additionally	ADV
fcis-10209	108	2	,	,	PUNCT
fcis-10209	108	3	the	the	DET
fcis-10209	108	4	cosine	cosine	NOUN
fcis-10209	108	5	annealing	anneal	VERB
fcis-10209	108	6	function	function	NOUN
fcis-10209	108	7	was	be	AUX
fcis-10209	108	8	applied	apply	VERB
fcis-10209	108	9	to	to	PART
fcis-10209	108	10	dynamically	dynamically	ADV
fcis-10209	108	11	adjust	adjust	VERB
fcis-10209	108	12	the	the	DET
fcis-10209	108	13	learning	learning	NOUN
fcis-10209	108	14	rate	rate	NOUN
fcis-10209	108	15	,	,	PUNCT
fcis-10209	108	16	with	with	SCONJ
fcis-10209	108	17	the	the	DET
fcis-10209	108	18	maximum	maximum	ADJ
fcis-10209	108	19	learning	learning	NOUN
fcis-10209	108	20	rate	rate	NOUN
fcis-10209	108	21	set	set	VERB
fcis-10209	108	22	to	to	ADP
fcis-10209	108	23	0.01	0.01	NUM
fcis-10209	108	24	.	.	PUNCT
fcis-10209	109	1	specifically	specifically	ADV
fcis-10209	109	2	,	,	PUNCT
fcis-10209	109	3	the	the	DET
fcis-10209	109	4	learning	learning	NOUN
fcis-10209	109	5	rate	rate	NOUN
fcis-10209	109	6	corresponding	correspond	VERB
fcis-10209	109	7	to	to	ADP
fcis-10209	109	8	the	the	DET
fcis-10209	109	9	current	current	ADJ
fcis-10209	109	10	epoch	epoch	NOUN
fcis-10209	109	11	was	be	AUX
fcis-10209	109	12	calculated	calculate	VERB
fcis-10209	109	13	based	base	VERB
fcis-10209	109	14	on	on	ADP
fcis-10209	109	15	the	the	DET
fcis-10209	109	16	shape	shape	NOUN
fcis-10209	109	17	of	of	ADP
fcis-10209	109	18	the	the	DET
fcis-10209	109	19	cosine	cosine	NOUN
fcis-10209	109	20	function	function	NOUN
fcis-10209	109	21	during	during	ADP
fcis-10209	109	22	each	each	DET
fcis-10209	109	23	iteration	iteration	NOUN
fcis-10209	109	24	,	,	PUNCT
fcis-10209	109	25	and	and	CCONJ
fcis-10209	109	26	then	then	ADV
fcis-10209	109	27	updated	update	VERB
fcis-10209	109	28	using	use	VERB
fcis-10209	109	29	the	the	DET
fcis-10209	109	30	scheduler	scheduler	NOUN
fcis-10209	109	31	.	.	PUNCT
fcis-10209	110	1	at	at	ADP
fcis-10209	110	2	the	the	DET
fcis-10209	110	3	end	end	NOUN
fcis-10209	110	4	of	of	ADP
fcis-10209	110	5	each	each	DET
fcis-10209	110	6	epoch	epoch	NOUN
fcis-10209	110	7	,	,	PUNCT
fcis-10209	110	8	the	the	DET
fcis-10209	110	9	relevant	relevant	ADJ
fcis-10209	110	10	indicators	indicator	NOUN
fcis-10209	110	11	for	for	ADP
fcis-10209	110	12	that	that	DET
fcis-10209	110	13	epoch	epoch	NOUN
fcis-10209	110	14	,	,	PUNCT
fcis-10209	110	15	including	include	VERB
fcis-10209	110	16	training	training	NOUN
fcis-10209	110	17	loss	loss	NOUN
fcis-10209	110	18	,	,	PUNCT
fcis-10209	110	19	training	training	NOUN
fcis-10209	110	20	accuracy	accuracy	NOUN
fcis-10209	110	21	,	,	PUNCT
fcis-10209	110	22	validation	validation	NOUN
fcis-10209	110	23	loss	loss	NOUN
fcis-10209	110	24	,	,	PUNCT
fcis-10209	110	25	validation	validation	NOUN
fcis-10209	110	26	accuracy	accuracy	NOUN
fcis-10209	110	27	,	,	PUNCT
fcis-10209	110	28	and	and	CCONJ
fcis-10209	110	29	current	current	ADJ
fcis-10209	110	30	learning	learning	NOUN
fcis-10209	110	31	rate	rate	NOUN
fcis-10209	110	32	,	,	PUNCT
fcis-10209	110	33	were	be	AUX
fcis-10209	110	34	recorded	record	VERB
fcis-10209	110	35	in	in	ADP
fcis-10209	110	36	tensorboard	tensorboard	NOUN
fcis-10209	110	37	for	for	ADP
fcis-10209	110	38	easy	easy	ADJ
fcis-10209	110	39	observation	observation	NOUN
fcis-10209	110	40	and	and	CCONJ
fcis-10209	110	41	comparison	comparison	NOUN
fcis-10209	110	42	of	of	ADP
fcis-10209	110	43	results	result	NOUN
fcis-10209	110	44	between	between	ADP
fcis-10209	110	45	different	different	ADJ
fcis-10209	110	46	experiments	experiment	NOUN
fcis-10209	110	47	.	.	PUNCT
fcis-10209	111	1	at	at	ADP
fcis-10209	111	2	the	the	DET
fcis-10209	111	3	same	same	ADJ
fcis-10209	111	4	time	time	NOUN
fcis-10209	111	5	,	,	PUNCT
fcis-10209	111	6	the	the	DET
fcis-10209	111	7	model	model	NOUN
fcis-10209	111	8	parameters	parameter	NOUN
fcis-10209	111	9	for	for	ADP
fcis-10209	111	10	the	the	DET
fcis-10209	111	11	current	current	ADJ
fcis-10209	111	12	epoch	epoch	NOUN
fcis-10209	111	13	were	be	AUX
fcis-10209	111	14	saved	save	VERB
fcis-10209	111	15	to	to	ADP
fcis-10209	111	16	a	a	DET
fcis-10209	111	17	designated	designate	VERB
fcis-10209	111	18	path	path	NOUN
fcis-10209	111	19	for	for	ADP
fcis-10209	111	20	future	future	ADJ
fcis-10209	111	21	testing	testing	NOUN
fcis-10209	111	22	or	or	CCONJ
fcis-10209	111	23	further	further	ADJ
fcis-10209	111	24	finetuning	finetuning	NOUN
fcis-10209	111	25	.	.	PUNCT
fcis-10209	112	1	4.3	4.3	NUM
fcis-10209	112	2	.	.	PUNCT
fcis-10209	112	3	performance	performance	NOUN
fcis-10209	112	4	evaluation	evaluation	NOUN
fcis-10209	112	5	metrics	metric	NOUN
fcis-10209	112	6	in	in	ADP
fcis-10209	112	7	this	this	DET
fcis-10209	112	8	paper	paper	NOUN
fcis-10209	112	9	,	,	PUNCT
fcis-10209	112	10	we	we	PRON
fcis-10209	112	11	used	use	VERB
fcis-10209	112	12	five	five	NUM
fcis-10209	112	13	specific	specific	ADJ
fcis-10209	112	14	metrics	metric	NOUN
fcis-10209	112	15	to	to	PART
fcis-10209	112	16	demonstrate	demonstrate	VERB
fcis-10209	112	17	the	the	DET
fcis-10209	112	18	performance	performance	NOUN
fcis-10209	112	19	of	of	ADP
fcis-10209	112	20	the	the	DET
fcis-10209	112	21	model	model	NOUN
fcis-10209	112	22	:	:	PUNCT
fcis-10209	112	23	accuracy	accuracy	NOUN
fcis-10209	112	24	,	,	PUNCT
fcis-10209	112	25	precision	precision	NOUN
fcis-10209	112	26	,	,	PUNCT
fcis-10209	112	27	recall	recall	NOUN
fcis-10209	112	28	,	,	PUNCT
fcis-10209	112	29	f1	f1	NOUN
fcis-10209	112	30	score	score	NOUN
fcis-10209	112	31	,	,	PUNCT
fcis-10209	112	32	and	and	CCONJ
fcis-10209	112	33	confusion	confusion	NOUN
fcis-10209	112	34	matrix	matrix	NOUN
fcis-10209	112	35	.	.	PUNCT
fcis-10209	113	1	in	in	ADP
fcis-10209	113	2	this	this	DET
fcis-10209	113	3	paper	paper	NOUN
fcis-10209	113	4	,	,	PUNCT
fcis-10209	113	5	assuming	assume	VERB
fcis-10209	113	6	tp	tp	NOUN
fcis-10209	113	7	and	and	CCONJ
fcis-10209	113	8	tn	tn	NOUN
fcis-10209	113	9	are	be	AUX
fcis-10209	113	10	the	the	DET
fcis-10209	113	11	number	number	NOUN
fcis-10209	113	12	of	of	ADP
fcis-10209	113	13	correctly	correctly	ADV
fcis-10209	113	14	identified	identify	VERB
fcis-10209	113	15	positive	positive	ADJ
fcis-10209	113	16	and	and	CCONJ
fcis-10209	113	17	negative	negative	ADJ
fcis-10209	113	18	samples	sample	NOUN
fcis-10209	113	19	,	,	PUNCT
fcis-10209	113	20	and	and	CCONJ
fcis-10209	113	21	fp	fp	NOUN
fcis-10209	113	22	and	and	CCONJ
fcis-10209	113	23	fn	fn	NOUN
fcis-10209	113	24	are	be	AUX
fcis-10209	113	25	the	the	DET
fcis-10209	113	26	number	number	NOUN
fcis-10209	113	27	of	of	ADP
fcis-10209	113	28	incorrectly	incorrectly	ADV
fcis-10209	113	29	identified	identify	VERB
fcis-10209	113	30	positive	positive	ADJ
fcis-10209	113	31	and	and	CCONJ
fcis-10209	113	32	negative	negative	ADJ
fcis-10209	113	33	samples	sample	NOUN
fcis-10209	113	34	,	,	PUNCT
fcis-10209	113	35	respectively	respectively	ADV
fcis-10209	113	36	,	,	PUNCT
fcis-10209	113	37	then	then	ADV
fcis-10209	113	38	:	:	PUNCT
fcis-10209	113	39	accuracy	accuracy	NOUN
fcis-10209	113	40	refers	refer	VERB
fcis-10209	113	41	to	to	ADP
fcis-10209	113	42	the	the	DET
fcis-10209	113	43	proportion	proportion	NOUN
fcis-10209	113	44	of	of	ADP
fcis-10209	113	45	samples	sample	NOUN
fcis-10209	113	46	correctly	correctly	ADV
fcis-10209	113	47	classified	classify	VERB
fcis-10209	113	48	by	by	ADP
fcis-10209	113	49	the	the	DET
fcis-10209	113	50	classifier	classifier	NOUN
fcis-10209	113	51	to	to	ADP
fcis-10209	113	52	the	the	DET
fcis-10209	113	53	total	total	ADJ
fcis-10209	113	54	number	number	NOUN
fcis-10209	113	55	of	of	ADP
fcis-10209	113	56	samples	sample	NOUN
fcis-10209	113	57	:	:	PUNCT
fcis-10209	113	58	(	(	PUNCT
fcis-10209	113	59	1	1	X
fcis-10209	113	60	)	)	PUNCT
fcis-10209	113	61	precision	precision	NOUN
fcis-10209	113	62	refers	refer	VERB
fcis-10209	113	63	to	to	ADP
fcis-10209	113	64	the	the	DET
fcis-10209	113	65	proportion	proportion	NOUN
fcis-10209	113	66	of	of	ADP
fcis-10209	113	67	actual	actual	ADJ
fcis-10209	113	68	positive	positive	ADJ
fcis-10209	113	69	samples	sample	NOUN
fcis-10209	113	70	among	among	ADP
fcis-10209	113	71	the	the	DET
fcis-10209	113	72	samples	sample	NOUN
fcis-10209	113	73	predicted	predict	VERB
fcis-10209	113	74	as	as	ADP
fcis-10209	113	75	positive	positive	ADJ
fcis-10209	113	76	by	by	ADP
fcis-10209	113	77	the	the	DET
fcis-10209	113	78	classifier	classifier	NOUN
fcis-10209	113	79	:	:	PUNCT
fcis-10209	113	80	(	(	PUNCT
fcis-10209	113	81	2	2	X
fcis-10209	113	82	)	)	PUNCT
fcis-10209	113	83	recall	recall	NOUN
fcis-10209	113	84	refers	refer	VERB
fcis-10209	113	85	to	to	ADP
fcis-10209	113	86	the	the	DET
fcis-10209	113	87	proportion	proportion	NOUN
fcis-10209	113	88	of	of	ADP
fcis-10209	113	89	samples	sample	NOUN
fcis-10209	113	90	correctly	correctly	ADV
fcis-10209	113	91	predicted	predict	VERB
fcis-10209	113	92	as	as	ADP
fcis-10209	113	93	positive	positive	ADJ
fcis-10209	113	94	by	by	ADP
fcis-10209	113	95	the	the	DET
fcis-10209	113	96	classifier	classifier	NOUN
fcis-10209	113	97	among	among	ADP
fcis-10209	113	98	all	all	DET
fcis-10209	113	99	true	true	ADJ
fcis-10209	113	100	positive	positive	ADJ
fcis-10209	113	101	samples	sample	NOUN
fcis-10209	113	102	:	:	PUNCT
fcis-10209	113	103	(	(	PUNCT
fcis-10209	113	104	3	3	X
fcis-10209	113	105	)	)	PUNCT
fcis-10209	113	106	f1	f1	NOUN
fcis-10209	113	107	score	score	NOUN
fcis-10209	113	108	is	be	AUX
fcis-10209	113	109	a	a	DET
fcis-10209	113	110	metric	metric	NOUN
fcis-10209	113	111	that	that	PRON
fcis-10209	113	112	considers	consider	VERB
fcis-10209	113	113	both	both	DET
fcis-10209	113	114	precision	precision	NOUN
fcis-10209	113	115	and	and	CCONJ
fcis-10209	113	116	recall	recall	NOUN
fcis-10209	113	117	,	,	PUNCT
fcis-10209	113	118	which	which	PRON
fcis-10209	113	119	is	be	AUX
fcis-10209	113	120	the	the	DET
fcis-10209	113	121	harmonic	harmonic	ADJ
fcis-10209	113	122	mean	mean	NOUN
fcis-10209	113	123	of	of	ADP
fcis-10209	113	124	precision	precision	NOUN
fcis-10209	113	125	and	and	CCONJ
fcis-10209	113	126	recall	recall	NOUN
fcis-10209	113	127	:	:	PUNCT
fcis-10209	113	128	1	1	NUM
fcis-10209	113	129	(	(	PUNCT
fcis-10209	113	130	4	4	NUM
fcis-10209	113	131	)	)	PUNCT
fcis-10209	113	132	the	the	DET
fcis-10209	113	133	confusion	confusion	NOUN
fcis-10209	113	134	matrix	matrix	NOUN
fcis-10209	113	135	can	can	AUX
fcis-10209	113	136	help	help	VERB
fcis-10209	113	137	gain	gain	VERB
fcis-10209	113	138	insight	insight	NOUN
fcis-10209	113	139	into	into	ADP
fcis-10209	113	140	the	the	DET
fcis-10209	113	141	model	model	NOUN
fcis-10209	113	142	's	's	PART
fcis-10209	113	143	performance	performance	NOUN
fcis-10209	113	144	on	on	ADP
fcis-10209	113	145	different	different	ADJ
fcis-10209	113	146	categories	category	NOUN
fcis-10209	113	147	and	and	CCONJ
fcis-10209	113	148	provide	provide	VERB
fcis-10209	113	149	some	some	DET
fcis-10209	113	150	reference	reference	NOUN
fcis-10209	113	151	for	for	ADP
fcis-10209	113	152	subsequent	subsequent	ADJ
fcis-10209	113	153	model	model	NOUN
fcis-10209	113	154	adjustments	adjustment	NOUN
fcis-10209	113	155	.	.	PUNCT
fcis-10209	114	1	4.4	4.4	NUM
fcis-10209	114	2	.	.	PUNCT
fcis-10209	115	1	experimental	experimental	ADJ
fcis-10209	115	2	results	result	NOUN
fcis-10209	115	3	and	and	CCONJ
fcis-10209	115	4	analysis	analysis	NOUN
fcis-10209	115	5	to	to	PART
fcis-10209	115	6	evaluate	evaluate	VERB
fcis-10209	115	7	the	the	DET
fcis-10209	115	8	accuracy	accuracy	NOUN
fcis-10209	115	9	of	of	ADP
fcis-10209	115	10	the	the	DET
fcis-10209	115	11	model	model	NOUN
fcis-10209	115	12	,	,	PUNCT
fcis-10209	115	13	the	the	DET
fcis-10209	115	14	cnntransformer	cnntransformer	NOUN
fcis-10209	115	15	hybrid	hybrid	NOUN
fcis-10209	115	16	model	model	NOUN
fcis-10209	115	17	was	be	AUX
fcis-10209	115	18	compared	compare	VERB
fcis-10209	115	19	with	with	ADP
fcis-10209	115	20	some	some	DET
fcis-10209	115	21	classical	classical	ADJ
fcis-10209	115	22	models	model	NOUN
fcis-10209	115	23	such	such	ADJ
fcis-10209	115	24	as	as	ADP
fcis-10209	115	25	vision	vision	NOUN
fcis-10209	115	26	transformer	transformer	NOUN
fcis-10209	115	27	,	,	PUNCT
fcis-10209	115	28	resnet50	resnet50	NOUN
fcis-10209	115	29	,	,	PUNCT
fcis-10209	115	30	alexnet	alexnet	NOUN
fcis-10209	115	31	,	,	PUNCT
fcis-10209	115	32	vgg16	vgg16	NOUN
fcis-10209	115	33	,	,	PUNCT
fcis-10209	115	34	etc	etc	X
fcis-10209	115	35	.	.	X
fcis-10209	115	36	in	in	ADP
fcis-10209	115	37	terms	term	NOUN
fcis-10209	115	38	of	of	ADP
fcis-10209	115	39	accuracy	accuracy	NOUN
fcis-10209	115	40	,	,	PUNCT
fcis-10209	115	41	as	as	SCONJ
fcis-10209	115	42	shown	show	VERB
fcis-10209	115	43	in	in	ADP
fcis-10209	115	44	table	table	NOUN
fcis-10209	115	45	1	1	NUM
fcis-10209	115	46	.	.	PUNCT
fcis-10209	115	47	table	table	NOUN
fcis-10209	115	48	1	1	NUM
fcis-10209	115	49	.	.	PUNCT
fcis-10209	115	50	comparison	comparison	NOUN
fcis-10209	115	51	of	of	ADP
fcis-10209	115	52	model	model	NOUN
fcis-10209	115	53	accuracy	accuracy	NOUN
fcis-10209	115	54	model	model	NOUN
fcis-10209	115	55	accuracy/%	accuracy/%	ADP
fcis-10209	115	56	vgg16	vgg16	PROPN
fcis-10209	115	57	86.21	86.21	NUM
fcis-10209	115	58	googlenet	googlenet	NOUN
fcis-10209	115	59	79.23	79.23	NUM
fcis-10209	115	60	alexnet	alexnet	NOUN
fcis-10209	115	61	80.09	80.09	NUM
fcis-10209	115	62	vit	vit	NOUN
fcis-10209	115	63	89.43	89.43	NUM
fcis-10209	115	64	cnn	cnn	PROPN
fcis-10209	115	65	-	-	PUNCT
fcis-10209	115	66	transformer	transformer	NOUN
fcis-10209	115	67	96.08	96.08	NUM
fcis-10209	115	68	based	base	VERB
fcis-10209	115	69	on	on	ADP
fcis-10209	115	70	table	table	NOUN
fcis-10209	115	71	1	1	NUM
fcis-10209	115	72	,	,	PUNCT
fcis-10209	115	73	it	it	PRON
fcis-10209	115	74	can	can	AUX
fcis-10209	115	75	be	be	AUX
fcis-10209	115	76	seen	see	VERB
fcis-10209	115	77	that	that	SCONJ
fcis-10209	115	78	the	the	DET
fcis-10209	115	79	use	use	NOUN
fcis-10209	115	80	of	of	ADP
fcis-10209	115	81	a	a	DET
fcis-10209	115	82	transformer	transformer	NOUN
fcis-10209	115	83	network	network	NOUN
fcis-10209	115	84	can	can	AUX
fcis-10209	115	85	bring	bring	VERB
fcis-10209	115	86	significant	significant	ADJ
fcis-10209	115	87	performance	performance	NOUN
fcis-10209	115	88	improvements	improvement	NOUN
fcis-10209	115	89	compared	compare	VERB
fcis-10209	115	90	to	to	ADP
fcis-10209	115	91	some	some	DET
fcis-10209	115	92	classical	classical	ADJ
fcis-10209	115	93	models	model	NOUN
fcis-10209	115	94	obtained	obtain	VERB
fcis-10209	115	95	by	by	ADP
fcis-10209	115	96	pre	pre	ADJ
fcis-10209	115	97	-	-	ADJ
fcis-10209	115	98	training	training	ADJ
fcis-10209	115	99	cnn	cnn	NOUN
fcis-10209	115	100	models	model	NOUN
fcis-10209	115	101	.	.	PUNCT
fcis-10209	116	1	due	due	ADP
fcis-10209	116	2	to	to	ADP
fcis-10209	116	3	the	the	DET
fcis-10209	116	4	stronger	strong	ADJ
fcis-10209	116	5	sequence	sequence	NOUN
fcis-10209	116	6	modeling	modeling	NOUN
fcis-10209	116	7	ability	ability	NOUN
fcis-10209	116	8	and	and	CCONJ
fcis-10209	116	9	the	the	DET
fcis-10209	116	10	ability	ability	NOUN
fcis-10209	116	11	to	to	PART
fcis-10209	116	12	capture	capture	VERB
fcis-10209	116	13	long	long	ADJ
fcis-10209	116	14	-	-	PUNCT
fcis-10209	116	15	range	range	NOUN
fcis-10209	116	16	dependencies	dependency	NOUN
fcis-10209	116	17	,	,	PUNCT
fcis-10209	116	18	the	the	DET
fcis-10209	116	19	hybrid	hybrid	NOUN
fcis-10209	116	20	model	model	NOUN
fcis-10209	116	21	and	and	CCONJ
fcis-10209	116	22	the	the	DET
fcis-10209	116	23	original	original	ADJ
fcis-10209	116	24	vit	vit	NOUN
fcis-10209	116	25	model	model	NOUN
fcis-10209	116	26	have	have	VERB
fcis-10209	116	27	higher	high	ADJ
fcis-10209	116	28	accuracy	accuracy	NOUN
fcis-10209	116	29	than	than	ADP
fcis-10209	116	30	other	other	ADJ
fcis-10209	116	31	models	model	NOUN
fcis-10209	116	32	.	.	PUNCT
fcis-10209	117	1	compared	compare	VERB
fcis-10209	117	2	with	with	ADP
fcis-10209	117	3	the	the	DET
fcis-10209	117	4	original	original	ADJ
fcis-10209	117	5	vit	vit	NOUN
fcis-10209	117	6	model	model	NOUN
fcis-10209	117	7	,	,	PUNCT
fcis-10209	117	8	the	the	DET
fcis-10209	117	9	cnn	cnn	PROPN
fcis-10209	117	10	-	-	PUNCT
fcis-10209	117	11	transformer	transformer	NOUN
fcis-10209	117	12	hybrid	hybrid	NOUN
fcis-10209	117	13	model	model	NOUN
fcis-10209	117	14	can	can	AUX
fcis-10209	117	15	better	well	ADV
fcis-10209	117	16	balance	balance	VERB
fcis-10209	117	17	local	local	ADJ
fcis-10209	117	18	and	and	CCONJ
fcis-10209	117	19	global	global	ADJ
fcis-10209	117	20	features	feature	NOUN
fcis-10209	117	21	,	,	PUNCT
fcis-10209	117	22	resulting	result	VERB
fcis-10209	117	23	in	in	ADP
fcis-10209	117	24	further	further	ADJ
fcis-10209	117	25	performance	performance	NOUN
fcis-10209	117	26	improvement	improvement	NOUN
fcis-10209	117	27	.	.	PUNCT
fcis-10209	118	1	fig7	fig7	PROPN
fcis-10209	118	2	.	.	PUNCT
fcis-10209	119	1	learning	learn	VERB
fcis-10209	119	2	rate	rate	NOUN
fcis-10209	119	3	curve	curve	PROPN
fcis-10209	119	4	fig	fig	NOUN
fcis-10209	119	5	8	8	NUM
fcis-10209	119	6	.	.	PUNCT
fcis-10209	120	1	accuracy	accuracy	NOUN
fcis-10209	120	2	curve	curve	NOUN
fcis-10209	120	3	on	on	ADP
fcis-10209	120	4	the	the	DET
fcis-10209	120	5	training	training	NOUN
fcis-10209	120	6	set	set	NOUN
fcis-10209	120	7	(	(	PUNCT
fcis-10209	120	8	left	left	ADJ
fcis-10209	120	9	)	)	PUNCT
fcis-10209	120	10	and	and	CCONJ
fcis-10209	120	11	validation	validation	NOUN
fcis-10209	120	12	set	set	NOUN
fcis-10209	120	13	(	(	PUNCT
fcis-10209	120	14	right	right	NOUN
fcis-10209	120	15	)	)	PUNCT
fcis-10209	120	16	to	to	PART
fcis-10209	120	17	demonstrate	demonstrate	VERB
fcis-10209	120	18	the	the	DET
fcis-10209	120	19	model	model	NOUN
fcis-10209	120	20	's	's	PART
fcis-10209	120	21	fitting	fitting	ADJ
fcis-10209	120	22	ability	ability	NOUN
fcis-10209	120	23	during	during	ADP
fcis-10209	120	24	the	the	DET
fcis-10209	120	25	training	training	NOUN
fcis-10209	120	26	process	process	NOUN
fcis-10209	120	27	and	and	CCONJ
fcis-10209	120	28	the	the	DET
fcis-10209	120	29	model	model	NOUN
fcis-10209	120	30	's	's	PART
fcis-10209	120	31	performance	performance	NOUN
fcis-10209	120	32	on	on	ADP
fcis-10209	120	33	unknown	unknown	ADJ
fcis-10209	120	34	data	datum	NOUN
fcis-10209	120	35	,	,	PUNCT
fcis-10209	120	36	the	the	DET
fcis-10209	120	37	learning	learning	NOUN
fcis-10209	120	38	rate	rate	NOUN
fcis-10209	120	39	curve	curve	NOUN
fcis-10209	120	40	(	(	PUNCT
fcis-10209	120	41	fig.7	fig.7	ADJ
fcis-10209	120	42	)	)	PUNCT
fcis-10209	120	43	,	,	PUNCT
fcis-10209	120	44	the	the	DET
fcis-10209	120	45	accuracy	accuracy	NOUN
fcis-10209	120	46	curve	curve	NOUN
fcis-10209	120	47	on	on	ADP
fcis-10209	120	48	the	the	DET
fcis-10209	120	49	training	training	NOUN
fcis-10209	120	50	set	set	NOUN
fcis-10209	120	51	and	and	CCONJ
fcis-10209	120	52	validation	validation	NOUN
fcis-10209	120	53	set	set	NOUN
fcis-10209	120	54	(	(	PUNCT
fcis-10209	120	55	fig.8	fig.8	PROPN
fcis-10209	120	56	)	)	PUNCT
fcis-10209	120	57	,	,	PUNCT
fcis-10209	120	58	and	and	CCONJ
fcis-10209	120	59	the	the	DET
fcis-10209	120	60	loss	loss	NOUN
fcis-10209	120	61	curve	curve	NOUN
fcis-10209	120	62	(	(	PUNCT
fcis-10209	120	63	fig.9	fig.9	PROPN
fcis-10209	120	64	)	)	PUNCT
fcis-10209	120	65	were	be	AUX
fcis-10209	120	66	plotted	plot	VERB
fcis-10209	120	67	.	.	PUNCT
fcis-10209	121	1	overall	overall	ADV
fcis-10209	121	2	,	,	PUNCT
fcis-10209	121	3	the	the	DET
fcis-10209	121	4	model	model	NOUN
fcis-10209	121	5	showed	show	VERB
fcis-10209	121	6	a	a	DET
fcis-10209	121	7	good	good	ADJ
fcis-10209	121	8	performance	performance	NOUN
fcis-10209	121	9	.	.	PUNCT
fcis-10209	122	1	on	on	ADP
fcis-10209	122	2	the	the	DET
fcis-10209	122	3	accuracy	accuracy	NOUN
fcis-10209	122	4	and	and	CCONJ
fcis-10209	122	5	loss	loss	NOUN
fcis-10209	122	6	curves	curve	NOUN
fcis-10209	122	7	of	of	ADP
fcis-10209	122	8	the	the	DET
fcis-10209	122	9	test	test	NOUN
fcis-10209	122	10	set	set	NOUN
fcis-10209	122	11	,	,	PUNCT
fcis-10209	122	12	there	there	PRON
fcis-10209	122	13	were	be	VERB
fcis-10209	122	14	slight	slight	ADJ
fcis-10209	122	15	fluctuations	fluctuation	NOUN
fcis-10209	122	16	and	and	CCONJ
fcis-10209	122	17	instability	instability	NOUN
fcis-10209	122	18	,	,	PUNCT
fcis-10209	122	19	which	which	PRON
fcis-10209	122	20	may	may	AUX
fcis-10209	122	21	be	be	AUX
fcis-10209	122	22	76	76	NUM
fcis-10209	122	23	due	due	ADJ
fcis-10209	122	24	to	to	ADP
fcis-10209	122	25	the	the	DET
fcis-10209	122	26	presence	presence	NOUN
fcis-10209	122	27	of	of	ADP
fcis-10209	122	28	noisy	noisy	ADJ
fcis-10209	122	29	or	or	CCONJ
fcis-10209	122	30	irregular	irregular	ADJ
fcis-10209	122	31	data	datum	NOUN
fcis-10209	122	32	in	in	ADP
fcis-10209	122	33	the	the	DET
fcis-10209	122	34	test	test	NOUN
fcis-10209	122	35	set	set	NOUN
fcis-10209	122	36	that	that	PRON
fcis-10209	122	37	are	be	AUX
fcis-10209	122	38	different	different	ADJ
fcis-10209	122	39	from	from	ADP
fcis-10209	122	40	the	the	DET
fcis-10209	122	41	data	datum	NOUN
fcis-10209	122	42	in	in	ADP
fcis-10209	122	43	the	the	DET
fcis-10209	122	44	training	training	NOUN
fcis-10209	122	45	set	set	NOUN
fcis-10209	122	46	,	,	PUNCT
fcis-10209	122	47	including	include	VERB
fcis-10209	122	48	incorrectly	incorrectly	ADV
fcis-10209	122	49	labeled	label	VERB
fcis-10209	122	50	data	datum	NOUN
fcis-10209	122	51	or	or	CCONJ
fcis-10209	122	52	missing	miss	VERB
fcis-10209	122	53	data	datum	NOUN
fcis-10209	122	54	,	,	PUNCT
fcis-10209	122	55	but	but	CCONJ
fcis-10209	122	56	the	the	DET
fcis-10209	122	57	fluctuation	fluctuation	NOUN
fcis-10209	122	58	range	range	NOUN
fcis-10209	122	59	is	be	AUX
fcis-10209	122	60	still	still	ADV
fcis-10209	122	61	acceptable	acceptable	ADJ
fcis-10209	122	62	.	.	PUNCT
fcis-10209	123	1	fig	fig	NOUN
fcis-10209	123	2	9	9	NUM
fcis-10209	123	3	.	.	PUNCT
fcis-10209	124	1	loss	loss	NOUN
fcis-10209	124	2	curve	curve	NOUN
fcis-10209	124	3	on	on	ADP
fcis-10209	124	4	the	the	DET
fcis-10209	124	5	training	training	NOUN
fcis-10209	124	6	set	set	NOUN
fcis-10209	124	7	(	(	PUNCT
fcis-10209	124	8	left	left	ADJ
fcis-10209	124	9	)	)	PUNCT
fcis-10209	124	10	and	and	CCONJ
fcis-10209	124	11	validation	validation	NOUN
fcis-10209	124	12	set	set	NOUN
fcis-10209	124	13	(	(	PUNCT
fcis-10209	124	14	right	right	ADJ
fcis-10209	124	15	)	)	PUNCT
fcis-10209	124	16	4.5	4.5	NUM
fcis-10209	124	17	.	.	PUNCT
fcis-10209	125	1	model	model	NOUN
fcis-10209	125	2	performance	performance	NOUN
fcis-10209	125	3	evaluation	evaluation	NOUN
fcis-10209	125	4	fig	fig	NOUN
fcis-10209	125	5	10	10	NUM
fcis-10209	125	6	.	.	PUNCT
fcis-10209	126	1	confusion	confusion	NOUN
fcis-10209	126	2	matrix	matrix	NOUN
fcis-10209	126	3	of	of	ADP
fcis-10209	126	4	the	the	DET
fcis-10209	126	5	original	original	ADJ
fcis-10209	126	6	vit	vit	NOUN
fcis-10209	126	7	(	(	PUNCT
fcis-10209	126	8	left	left	ADJ
fcis-10209	126	9	)	)	PUNCT
fcis-10209	126	10	and	and	CCONJ
fcis-10209	126	11	the	the	DET
fcis-10209	126	12	hybrid	hybrid	ADJ
fcis-10209	126	13	model	model	NOUN
fcis-10209	126	14	(	(	PUNCT
fcis-10209	126	15	right	right	ADJ
fcis-10209	126	16	)	)	PUNCT
fcis-10209	126	17	fig	fig	NOUN
fcis-10209	126	18	11	11	NUM
fcis-10209	126	19	.	.	PUNCT
fcis-10209	127	1	class	class	NOUN
fcis-10209	127	2	7	7	NUM
fcis-10209	127	3	is	be	AUX
fcis-10209	127	4	recognized	recognize	VERB
fcis-10209	127	5	as	as	ADP
fcis-10209	127	6	class	class	NOUN
fcis-10209	127	7	0	0	NUM
fcis-10209	127	8	the	the	DET
fcis-10209	127	9	confusion	confusion	NOUN
fcis-10209	127	10	matrix	matrix	NOUN
fcis-10209	127	11	is	be	AUX
fcis-10209	127	12	a	a	DET
fcis-10209	127	13	tool	tool	NOUN
fcis-10209	127	14	used	use	VERB
fcis-10209	127	15	to	to	PART
fcis-10209	127	16	evaluate	evaluate	VERB
fcis-10209	127	17	the	the	DET
fcis-10209	127	18	performance	performance	NOUN
fcis-10209	127	19	of	of	ADP
fcis-10209	127	20	classification	classification	NOUN
fcis-10209	127	21	models	model	NOUN
fcis-10209	127	22	.	.	PUNCT
fcis-10209	128	1	it	it	PRON
fcis-10209	128	2	can	can	AUX
fcis-10209	128	3	help	help	VERB
fcis-10209	128	4	understand	understand	VERB
fcis-10209	128	5	the	the	DET
fcis-10209	128	6	model	model	NOUN
fcis-10209	128	7	's	's	PART
fcis-10209	128	8	performance	performance	NOUN
fcis-10209	128	9	on	on	ADP
fcis-10209	128	10	different	different	ADJ
fcis-10209	128	11	categories	category	NOUN
fcis-10209	128	12	,	,	PUNCT
fcis-10209	128	13	and	and	CCONJ
fcis-10209	128	14	quantify	quantify	VERB
fcis-10209	128	15	and	and	CCONJ
fcis-10209	128	16	visualize	visualize	VERB
fcis-10209	128	17	its	its	PRON
fcis-10209	128	18	classification	classification	NOUN
fcis-10209	128	19	performance	performance	NOUN
fcis-10209	128	20	.	.	PUNCT
fcis-10209	129	1	here	here	ADV
fcis-10209	129	2	,	,	PUNCT
fcis-10209	129	3	a	a	DET
fcis-10209	129	4	comparison	comparison	NOUN
fcis-10209	129	5	was	be	AUX
fcis-10209	129	6	made	make	VERB
fcis-10209	129	7	between	between	ADP
fcis-10209	129	8	the	the	DET
fcis-10209	129	9	confusion	confusion	NOUN
fcis-10209	129	10	matrices	matrix	NOUN
fcis-10209	129	11	of	of	ADP
fcis-10209	129	12	the	the	DET
fcis-10209	129	13	original	original	ADJ
fcis-10209	129	14	vit	vit	ADJ
fcis-10209	129	15	model	model	NOUN
fcis-10209	129	16	and	and	CCONJ
fcis-10209	129	17	the	the	DET
fcis-10209	129	18	cnn	cnn	PROPN
fcis-10209	129	19	-	-	PUNCT
fcis-10209	129	20	transformer	transformer	NOUN
fcis-10209	129	21	hybrid	hybrid	NOUN
fcis-10209	129	22	model	model	NOUN
fcis-10209	129	23	as	as	SCONJ
fcis-10209	129	24	shown	show	VERB
fcis-10209	129	25	in	in	ADP
fcis-10209	129	26	fig.10	fig.10	PRON
fcis-10209	129	27	.	.	PUNCT
fcis-10209	130	1	the	the	DET
fcis-10209	130	2	horizontal	horizontal	ADJ
fcis-10209	130	3	axis	axis	NOUN
fcis-10209	130	4	represents	represent	VERB
fcis-10209	130	5	the	the	DET
fcis-10209	130	6	predicted	predict	VERB
fcis-10209	130	7	categories	category	NOUN
fcis-10209	130	8	on	on	ADP
fcis-10209	130	9	the	the	DET
fcis-10209	130	10	test	test	NOUN
fcis-10209	130	11	set	set	NOUN
fcis-10209	130	12	,	,	PUNCT
fcis-10209	130	13	while	while	SCONJ
fcis-10209	130	14	the	the	DET
fcis-10209	130	15	vertical	vertical	ADJ
fcis-10209	130	16	axis	axis	NOUN
fcis-10209	130	17	represents	represent	VERB
fcis-10209	130	18	the	the	DET
fcis-10209	130	19	true	true	ADJ
fcis-10209	130	20	categories	category	NOUN
fcis-10209	130	21	of	of	ADP
fcis-10209	130	22	the	the	DET
fcis-10209	130	23	test	test	NOUN
fcis-10209	130	24	set	set	NOUN
fcis-10209	130	25	images	image	NOUN
fcis-10209	130	26	.	.	PUNCT
fcis-10209	131	1	the	the	DET
fcis-10209	131	2	numbers	number	NOUN
fcis-10209	131	3	on	on	ADP
fcis-10209	131	4	the	the	DET
fcis-10209	131	5	main	main	ADJ
fcis-10209	131	6	diagonal	diagonal	NOUN
fcis-10209	131	7	represent	represent	VERB
fcis-10209	131	8	the	the	DET
fcis-10209	131	9	number	number	NOUN
fcis-10209	131	10	of	of	ADP
fcis-10209	131	11	correct	correct	ADJ
fcis-10209	131	12	predictions	prediction	NOUN
fcis-10209	131	13	by	by	ADP
fcis-10209	131	14	the	the	DET
fcis-10209	131	15	model	model	NOUN
fcis-10209	131	16	,	,	PUNCT
fcis-10209	131	17	while	while	SCONJ
fcis-10209	131	18	the	the	DET
fcis-10209	131	19	others	other	NOUN
fcis-10209	131	20	are	be	AUX
fcis-10209	131	21	the	the	DET
fcis-10209	131	22	numbers	number	NOUN
fcis-10209	131	23	of	of	ADP
fcis-10209	131	24	incorrect	incorrect	ADJ
fcis-10209	131	25	predictions	prediction	NOUN
fcis-10209	131	26	.	.	PUNCT
fcis-10209	132	1	in	in	ADP
fcis-10209	132	2	the	the	DET
fcis-10209	132	3	original	original	ADJ
fcis-10209	132	4	dataset	dataset	NOUN
fcis-10209	132	5	,	,	PUNCT
fcis-10209	132	6	there	there	PRON
fcis-10209	132	7	are	be	VERB
fcis-10209	132	8	8	8	NUM
fcis-10209	132	9	categories	category	NOUN
fcis-10209	132	10	of	of	ADP
fcis-10209	132	11	weeds	weed	NOUN
fcis-10209	132	12	,	,	PUNCT
fcis-10209	132	13	labeled	label	VERB
fcis-10209	132	14	from	from	ADP
fcis-10209	132	15	0	0	NUM
fcis-10209	132	16	to	to	ADP
fcis-10209	132	17	7	7	NUM
fcis-10209	132	18	,	,	PUNCT
fcis-10209	132	19	and	and	CCONJ
fcis-10209	132	20	the	the	DET
fcis-10209	132	21	negative	negative	ADJ
fcis-10209	132	22	class	class	NOUN
fcis-10209	132	23	is	be	AUX
fcis-10209	132	24	labeled	label	VERB
fcis-10209	132	25	as	as	ADP
fcis-10209	132	26	8	8	NUM
fcis-10209	132	27	.	.	PUNCT
fcis-10209	133	1	the	the	DET
fcis-10209	133	2	sample	sample	NOUN
fcis-10209	133	3	distribution	distribution	NOUN
fcis-10209	133	4	is	be	AUX
fcis-10209	133	5	uneven	uneven	ADJ
fcis-10209	133	6	,	,	PUNCT
fcis-10209	133	7	so	so	SCONJ
fcis-10209	133	8	there	there	PRON
fcis-10209	133	9	are	be	VERB
fcis-10209	133	10	more	more	ADJ
fcis-10209	133	11	cases	case	NOUN
fcis-10209	133	12	where	where	SCONJ
fcis-10209	133	13	the	the	DET
fcis-10209	133	14	eighth	eighth	ADJ
fcis-10209	133	15	category	category	NOUN
fcis-10209	133	16	is	be	AUX
fcis-10209	133	17	misclassified	misclassifie	VERB
fcis-10209	133	18	as	as	ADP
fcis-10209	133	19	other	other	ADJ
fcis-10209	133	20	categories	category	NOUN
fcis-10209	133	21	or	or	CCONJ
fcis-10209	133	22	other	other	ADJ
fcis-10209	133	23	categories	category	NOUN
fcis-10209	133	24	are	be	AUX
fcis-10209	133	25	misclassified	misclassifie	VERB
fcis-10209	133	26	as	as	ADP
fcis-10209	133	27	the	the	DET
fcis-10209	133	28	eighth	eighth	ADJ
fcis-10209	133	29	category	category	NOUN
fcis-10209	133	30	.	.	PUNCT
fcis-10209	134	1	however	however	ADV
fcis-10209	134	2	,	,	PUNCT
fcis-10209	134	3	considering	consider	VERB
fcis-10209	134	4	that	that	SCONJ
fcis-10209	134	5	the	the	DET
fcis-10209	134	6	number	number	NOUN
fcis-10209	134	7	of	of	ADP
fcis-10209	134	8	images	image	NOUN
fcis-10209	134	9	in	in	ADP
fcis-10209	134	10	the	the	DET
fcis-10209	134	11	eighth	eighth	ADJ
fcis-10209	134	12	category	category	NOUN
fcis-10209	134	13	is	be	AUX
fcis-10209	134	14	almost	almost	ADV
fcis-10209	134	15	8	8	NUM
fcis-10209	134	16	times	time	NOUN
fcis-10209	134	17	that	that	DET
fcis-10209	134	18	of	of	ADP
fcis-10209	134	19	other	other	ADJ
fcis-10209	134	20	categories	category	NOUN
fcis-10209	134	21	,	,	PUNCT
fcis-10209	134	22	this	this	DET
fcis-10209	134	23	error	error	NOUN
fcis-10209	134	24	rate	rate	NOUN
fcis-10209	134	25	can	can	AUX
fcis-10209	134	26	be	be	AUX
fcis-10209	134	27	accepted	accept	VERB
fcis-10209	134	28	.	.	PUNCT
fcis-10209	135	1	the	the	DET
fcis-10209	135	2	hybrid	hybrid	NOUN
fcis-10209	135	3	model	model	NOUN
fcis-10209	135	4	repeatedly	repeatedly	ADV
fcis-10209	135	5	misidentified	misidentifie	VERB
fcis-10209	135	6	images	image	NOUN
fcis-10209	135	7	of	of	ADP
fcis-10209	135	8	category	category	NOUN
fcis-10209	135	9	7	7	NUM
fcis-10209	135	10	as	as	ADP
fcis-10209	135	11	images	image	NOUN
fcis-10209	135	12	of	of	ADP
fcis-10209	135	13	category	category	NOUN
fcis-10209	135	14	0	0	NUM
fcis-10209	135	15	.	.	PUNCT
fcis-10209	136	1	based	base	VERB
fcis-10209	136	2	on	on	ADP
fcis-10209	136	3	the	the	DET
fcis-10209	136	4	analysis	analysis	NOUN
fcis-10209	136	5	of	of	ADP
fcis-10209	136	6	the	the	DET
fcis-10209	136	7	images	image	NOUN
fcis-10209	136	8	as	as	SCONJ
fcis-10209	136	9	shown	show	VERB
fcis-10209	136	10	in	in	ADP
fcis-10209	136	11	fig.11	fig.11	PROPN
fcis-10209	136	12	,	,	PUNCT
fcis-10209	136	13	it	it	PRON
fcis-10209	136	14	is	be	AUX
fcis-10209	136	15	speculated	speculate	VERB
fcis-10209	136	16	in	in	ADP
fcis-10209	136	17	this	this	DET
fcis-10209	136	18	study	study	NOUN
fcis-10209	136	19	that	that	SCONJ
fcis-10209	136	20	this	this	PRON
fcis-10209	136	21	is	be	AUX
fcis-10209	136	22	due	due	ADJ
fcis-10209	136	23	to	to	ADP
fcis-10209	136	24	poor	poor	ADJ
fcis-10209	136	25	lighting	lighting	NOUN
fcis-10209	136	26	and	and	CCONJ
fcis-10209	136	27	a	a	DET
fcis-10209	136	28	similarity	similarity	NOUN
fcis-10209	136	29	in	in	ADP
fcis-10209	136	30	the	the	DET
fcis-10209	136	31	morphology	morphology	NOUN
fcis-10209	136	32	of	of	ADP
fcis-10209	136	33	the	the	DET
fcis-10209	136	34	two	two	NUM
fcis-10209	136	35	types	type	NOUN
fcis-10209	136	36	of	of	ADP
fcis-10209	136	37	weeds	weed	NOUN
fcis-10209	136	38	,	,	PUNCT
fcis-10209	136	39	leading	lead	VERB
fcis-10209	136	40	to	to	ADP
fcis-10209	136	41	difficult	difficult	ADJ
fcis-10209	136	42	recognition	recognition	NOUN
fcis-10209	136	43	.	.	PUNCT
fcis-10209	137	1	overall	overall	ADJ
fcis-10209	137	2	,	,	PUNCT
fcis-10209	137	3	according	accord	VERB
fcis-10209	137	4	to	to	ADP
fcis-10209	137	5	the	the	DET
fcis-10209	137	6	analysis	analysis	NOUN
fcis-10209	137	7	of	of	ADP
fcis-10209	137	8	the	the	DET
fcis-10209	137	9	confusion	confusion	NOUN
fcis-10209	137	10	matrix	matrix	NOUN
fcis-10209	137	11	,	,	PUNCT
fcis-10209	137	12	the	the	DET
fcis-10209	137	13	hybrid	hybrid	NOUN
fcis-10209	137	14	model	model	NOUN
fcis-10209	137	15	has	have	VERB
fcis-10209	137	16	good	good	ADJ
fcis-10209	137	17	performance	performance	NOUN
fcis-10209	137	18	on	on	ADP
fcis-10209	137	19	this	this	DET
fcis-10209	137	20	dataset	dataset	NOUN
fcis-10209	137	21	,	,	PUNCT
fcis-10209	137	22	and	and	CCONJ
fcis-10209	137	23	has	have	AUX
fcis-10209	137	24	made	make	VERB
fcis-10209	137	25	significant	significant	ADJ
fcis-10209	137	26	improvements	improvement	NOUN
fcis-10209	137	27	in	in	ADP
fcis-10209	137	28	performance	performance	NOUN
fcis-10209	137	29	compared	compare	VERB
fcis-10209	137	30	to	to	ADP
fcis-10209	137	31	the	the	DET
fcis-10209	137	32	original	original	ADJ
fcis-10209	137	33	vit	vit	ADJ
fcis-10209	137	34	model	model	NOUN
fcis-10209	137	35	.	.	PUNCT
fcis-10209	138	1	5	5	X
fcis-10209	138	2	.	.	X
fcis-10209	138	3	conclusion	conclusion	NOUN
fcis-10209	138	4	to	to	PART
fcis-10209	138	5	simultaneously	simultaneously	ADV
fcis-10209	138	6	capture	capture	VERB
fcis-10209	138	7	local	local	ADJ
fcis-10209	138	8	and	and	CCONJ
fcis-10209	138	9	global	global	ADJ
fcis-10209	138	10	information	information	NOUN
fcis-10209	138	11	in	in	ADP
fcis-10209	138	12	the	the	DET
fcis-10209	138	13	feature	feature	NOUN
fcis-10209	138	14	extraction	extraction	NOUN
fcis-10209	138	15	and	and	CCONJ
fcis-10209	138	16	classification	classification	NOUN
fcis-10209	138	17	process	process	NOUN
fcis-10209	138	18	of	of	ADP
fcis-10209	138	19	weed	weed	NOUN
fcis-10209	138	20	recognition	recognition	NOUN
fcis-10209	138	21	,	,	PUNCT
fcis-10209	138	22	this	this	DET
fcis-10209	138	23	paper	paper	NOUN
fcis-10209	138	24	combines	combine	VERB
fcis-10209	138	25	the	the	DET
fcis-10209	138	26	advantages	advantage	NOUN
fcis-10209	138	27	of	of	ADP
fcis-10209	138	28	traditional	traditional	ADJ
fcis-10209	138	29	cnn	cnn	PROPN
fcis-10209	138	30	models	model	NOUN
fcis-10209	138	31	and	and	CCONJ
fcis-10209	138	32	vision	vision	NOUN
fcis-10209	138	33	transformer	transformer	NOUN
fcis-10209	138	34	models	model	NOUN
fcis-10209	138	35	to	to	PART
fcis-10209	138	36	design	design	VERB
fcis-10209	138	37	a	a	DET
fcis-10209	138	38	more	more	ADV
fcis-10209	138	39	suitable	suitable	ADJ
fcis-10209	138	40	cnn	cnn	ADJ
fcis-10209	138	41	-	-	PUNCT
fcis-10209	138	42	transformer	transformer	ADJ
fcis-10209	138	43	hybrid	hybrid	NOUN
fcis-10209	138	44	model	model	NOUN
fcis-10209	138	45	for	for	ADP
fcis-10209	138	46	weed	weed	NOUN
fcis-10209	138	47	recognition	recognition	NOUN
fcis-10209	138	48	.	.	PUNCT
fcis-10209	139	1	based	base	VERB
fcis-10209	139	2	on	on	ADP
fcis-10209	139	3	a	a	DET
fcis-10209	139	4	four	four	NUM
fcis-10209	139	5	-	-	PUNCT
fcis-10209	139	6	stage	stage	NOUN
fcis-10209	139	7	cnn	cnn	NOUN
fcis-10209	139	8	structure	structure	NOUN
fcis-10209	139	9	,	,	PUNCT
fcis-10209	139	10	each	each	DET
fcis-10209	139	11	stage	stage	NOUN
fcis-10209	139	12	uses	use	VERB
fcis-10209	139	13	stacked	stack	VERB
fcis-10209	139	14	convblocks	convblock	NOUN
fcis-10209	139	15	to	to	PART
fcis-10209	139	16	perform	perform	VERB
fcis-10209	139	17	local	local	ADJ
fcis-10209	139	18	feature	feature	NOUN
fcis-10209	139	19	extraction	extraction	NOUN
fcis-10209	139	20	,	,	PUNCT
fcis-10209	139	21	and	and	CCONJ
fcis-10209	139	22	a	a	DET
fcis-10209	139	23	transblock	transblock	NOUN
fcis-10209	139	24	is	be	AUX
fcis-10209	139	25	added	add	VERB
fcis-10209	139	26	at	at	ADP
fcis-10209	139	27	the	the	DET
fcis-10209	139	28	end	end	NOUN
fcis-10209	139	29	of	of	ADP
fcis-10209	139	30	each	each	DET
fcis-10209	139	31	stage	stage	NOUN
fcis-10209	139	32	for	for	ADP
fcis-10209	139	33	global	global	ADJ
fcis-10209	139	34	feature	feature	NOUN
fcis-10209	139	35	extraction	extraction	NOUN
fcis-10209	139	36	.	.	PUNCT
fcis-10209	140	1	in	in	ADP
fcis-10209	140	2	the	the	DET
fcis-10209	140	3	transblock	transblock	NOUN
fcis-10209	140	4	,	,	PUNCT
fcis-10209	140	5	an	an	DET
fcis-10209	140	6	stda	stda	VERB
fcis-10209	140	7	attention	attention	NOUN
fcis-10209	140	8	mechanism	mechanism	NOUN
fcis-10209	140	9	is	be	AUX
fcis-10209	140	10	introduced	introduce	VERB
fcis-10209	140	11	,	,	PUNCT
fcis-10209	140	12	which	which	PRON
fcis-10209	140	13	efficiently	efficiently	ADV
fcis-10209	140	14	combines	combine	VERB
fcis-10209	140	15	the	the	DET
fcis-10209	140	16	cnn	cnn	NOUN
fcis-10209	140	17	and	and	CCONJ
fcis-10209	140	18	vit	vit	NOUN
fcis-10209	140	19	models	model	NOUN
fcis-10209	140	20	to	to	PART
fcis-10209	140	21	minimize	minimize	VERB
fcis-10209	140	22	the	the	DET
fcis-10209	140	23	impact	impact	NOUN
fcis-10209	140	24	of	of	ADP
fcis-10209	140	25	the	the	DET
fcis-10209	140	26	slower	slow	ADJ
fcis-10209	140	27	inference	inference	NOUN
fcis-10209	140	28	speed	speed	NOUN
fcis-10209	140	29	caused	cause	VERB
fcis-10209	140	30	by	by	ADP
fcis-10209	140	31	the	the	DET
fcis-10209	140	32	self	self	NOUN
fcis-10209	140	33	-	-	PUNCT
fcis-10209	140	34	attention	attention	NOUN
fcis-10209	140	35	calculation	calculation	NOUN
fcis-10209	140	36	method	method	NOUN
fcis-10209	140	37	.	.	PUNCT
fcis-10209	141	1	experimental	experimental	ADJ
fcis-10209	141	2	results	result	NOUN
fcis-10209	141	3	show	show	VERB
fcis-10209	141	4	that	that	SCONJ
fcis-10209	141	5	the	the	DET
fcis-10209	141	6	hybrid	hybrid	NOUN
fcis-10209	141	7	model	model	NOUN
fcis-10209	141	8	has	have	VERB
fcis-10209	141	9	high	high	ADJ
fcis-10209	141	10	accuracy	accuracy	NOUN
fcis-10209	141	11	,	,	PUNCT
fcis-10209	141	12	precision	precision	NOUN
fcis-10209	141	13	,	,	PUNCT
fcis-10209	141	14	recall	recall	NOUN
fcis-10209	141	15	,	,	PUNCT
fcis-10209	141	16	and	and	CCONJ
fcis-10209	141	17	f1	f1	PROPN
fcis-10209	141	18	score	score	NOUN
fcis-10209	141	19	.	.	PUNCT
fcis-10209	142	1	this	this	DET
fcis-10209	142	2	model	model	NOUN
fcis-10209	142	3	overcomes	overcome	VERB
fcis-10209	142	4	the	the	DET
fcis-10209	142	5	limitations	limitation	NOUN
fcis-10209	142	6	of	of	ADP
fcis-10209	142	7	cnn	cnn	PROPN
fcis-10209	142	8	in	in	ADP
fcis-10209	142	9	modeling	model	VERB
fcis-10209	142	10	global	global	ADJ
fcis-10209	142	11	information	information	NOUN
fcis-10209	142	12	and	and	CCONJ
fcis-10209	142	13	does	do	AUX
fcis-10209	142	14	not	not	PART
fcis-10209	142	15	require	require	VERB
fcis-10209	142	16	the	the	DET
fcis-10209	142	17	massive	massive	ADJ
fcis-10209	142	18	computation	computation	NOUN
fcis-10209	142	19	of	of	ADP
fcis-10209	142	20	the	the	DET
fcis-10209	142	21	vit	vit	ADJ
fcis-10209	142	22	model	model	NOUN
fcis-10209	142	23	,	,	PUNCT
fcis-10209	142	24	making	make	VERB
fcis-10209	142	25	it	it	PRON
fcis-10209	142	26	more	more	ADV
fcis-10209	142	27	suitable	suitable	ADJ
fcis-10209	142	28	for	for	ADP
fcis-10209	142	29	deployment	deployment	NOUN
fcis-10209	142	30	on	on	ADP
fcis-10209	142	31	edge	edge	NOUN
fcis-10209	142	32	devices	device	NOUN
fcis-10209	142	33	for	for	ADP
fcis-10209	142	34	weed	weed	NOUN
fcis-10209	142	35	recognition	recognition	NOUN
fcis-10209	142	36	.	.	PUNCT
fcis-10209	143	1	however	however	ADV
fcis-10209	143	2	,	,	PUNCT
fcis-10209	143	3	there	there	PRON
fcis-10209	143	4	are	be	VERB
fcis-10209	143	5	still	still	ADV
fcis-10209	143	6	challenges	challenge	NOUN
fcis-10209	143	7	in	in	ADP
fcis-10209	143	8	practical	practical	ADJ
fcis-10209	143	9	applications	application	NOUN
fcis-10209	143	10	,	,	PUNCT
fcis-10209	143	11	such	such	ADJ
fcis-10209	143	12	as	as	ADP
fcis-10209	143	13	the	the	DET
fcis-10209	143	14	diversity	diversity	NOUN
fcis-10209	143	15	and	and	CCONJ
fcis-10209	143	16	quantity	quantity	NOUN
fcis-10209	143	17	of	of	ADP
fcis-10209	143	18	weed	weed	NOUN
fcis-10209	143	19	species	specie	NOUN
fcis-10209	143	20	increasing	increase	VERB
fcis-10209	143	21	the	the	DET
fcis-10209	143	22	difficulty	difficulty	NOUN
fcis-10209	143	23	of	of	ADP
fcis-10209	143	24	model	model	NOUN
fcis-10209	143	25	training	training	NOUN
fcis-10209	143	26	,	,	PUNCT
fcis-10209	143	27	requiring	require	VERB
fcis-10209	143	28	more	more	ADJ
fcis-10209	143	29	computing	compute	VERB
fcis-10209	143	30	resources	resource	NOUN
fcis-10209	143	31	and	and	CCONJ
fcis-10209	143	32	memory	memory	NOUN
fcis-10209	143	33	space	space	NOUN
fcis-10209	143	34	to	to	PART
fcis-10209	143	35	ensure	ensure	VERB
fcis-10209	143	36	the	the	DET
fcis-10209	143	37	efficient	efficient	ADJ
fcis-10209	143	38	operation	operation	NOUN
fcis-10209	143	39	of	of	ADP
fcis-10209	143	40	the	the	DET
fcis-10209	143	41	model	model	NOUN
fcis-10209	143	42	,	,	PUNCT
fcis-10209	143	43	and	and	CCONJ
fcis-10209	143	44	different	different	ADJ
fcis-10209	143	45	shooting	shooting	NOUN
fcis-10209	143	46	angles	angle	NOUN
fcis-10209	143	47	and	and	CCONJ
fcis-10209	143	48	lighting	lighting	NOUN
fcis-10209	143	49	conditions	condition	NOUN
fcis-10209	143	50	in	in	ADP
fcis-10209	143	51	different	different	ADJ
fcis-10209	143	52	environments	environment	NOUN
fcis-10209	143	53	may	may	AUX
fcis-10209	143	54	also	also	ADV
fcis-10209	143	55	affect	affect	VERB
fcis-10209	143	56	the	the	DET
fcis-10209	143	57	stability	stability	NOUN
fcis-10209	143	58	and	and	CCONJ
fcis-10209	143	59	accuracy	accuracy	NOUN
fcis-10209	143	60	of	of	ADP
fcis-10209	143	61	the	the	DET
fcis-10209	143	62	model	model	NOUN
fcis-10209	143	63	.	.	PUNCT
fcis-10209	144	1	future	future	ADJ
fcis-10209	144	2	research	research	NOUN
fcis-10209	144	3	can	can	AUX
fcis-10209	144	4	further	far	ADV
fcis-10209	144	5	optimize	optimize	VERB
fcis-10209	144	6	the	the	DET
fcis-10209	144	7	structure	structure	NOUN
fcis-10209	144	8	and	and	CCONJ
fcis-10209	144	9	parameter	parameter	NOUN
fcis-10209	144	10	settings	setting	NOUN
fcis-10209	144	11	of	of	ADP
fcis-10209	144	12	the	the	DET
fcis-10209	144	13	hybrid	hybrid	ADJ
fcis-10209	144	14	model	model	NOUN
fcis-10209	144	15	,	,	PUNCT
fcis-10209	144	16	explore	explore	VERB
fcis-10209	144	17	more	more	ADV
fcis-10209	144	18	efficient	efficient	ADJ
fcis-10209	144	19	computing	computing	NOUN
fcis-10209	144	20	methods	method	NOUN
fcis-10209	144	21	and	and	CCONJ
fcis-10209	144	22	algorithms	algorithm	NOUN
fcis-10209	144	23	,	,	PUNCT
fcis-10209	144	24	and	and	CCONJ
fcis-10209	144	25	combine	combine	VERB
fcis-10209	144	26	the	the	DET
fcis-10209	144	27	model	model	NOUN
fcis-10209	144	28	with	with	ADP
fcis-10209	144	29	other	other	ADJ
fcis-10209	144	30	weed	weed	NOUN
fcis-10209	144	31	classifiers	classifier	NOUN
fcis-10209	144	32	and	and	CCONJ
fcis-10209	144	33	agricultural	agricultural	ADJ
fcis-10209	144	34	technologies	technology	NOUN
fcis-10209	144	35	to	to	PART
fcis-10209	144	36	improve	improve	VERB
fcis-10209	144	37	the	the	DET
fcis-10209	144	38	efficiency	efficiency	NOUN
fcis-10209	144	39	and	and	CCONJ
fcis-10209	144	40	sustainability	sustainability	NOUN
fcis-10209	144	41	of	of	ADP
fcis-10209	144	42	agricultural	agricultural	ADJ
fcis-10209	144	43	production	production	NOUN
fcis-10209	144	44	.	.	PUNCT
fcis-10209	145	1	references	reference	NOUN
fcis-10209	145	2	[	[	X
fcis-10209	145	3	1	1	NUM
fcis-10209	145	4	]	]	X
fcis-10209	145	5	haq	haq	PROPN
fcis-10209	145	6	,	,	PUNCT
fcis-10209	145	7	m.	m.	NOUN
fcis-10209	145	8	a.	a.	NOUN
fcis-10209	145	9	(	(	PUNCT
fcis-10209	145	10	2022	2022	NUM
fcis-10209	145	11	)	)	PUNCT
fcis-10209	145	12	.	.	PUNCT
fcis-10209	146	1	cnn	cnn	PROPN
fcis-10209	146	2	based	base	VERB
fcis-10209	146	3	automated	automate	VERB
fcis-10209	146	4	weed	weed	NOUN
fcis-10209	146	5	detection	detection	NOUN
fcis-10209	146	6	system	system	NOUN
fcis-10209	146	7	using	use	VERB
fcis-10209	146	8	uav	uav	PROPN
fcis-10209	146	9	imagery	imagery	NOUN
fcis-10209	146	10	.	.	PUNCT
fcis-10209	147	1	computer	computer	NOUN
fcis-10209	147	2	systems	system	NOUN
fcis-10209	147	3	science	science	NOUN
fcis-10209	147	4	and	and	CCONJ
fcis-10209	147	5	engineering	engineering	NOUN
fcis-10209	147	6	,	,	PUNCT
fcis-10209	147	7	42(2	42(2	PROPN
fcis-10209	147	8	)	)	PUNCT
fcis-10209	147	9	,	,	PUNCT
fcis-10209	147	10	837	837	NUM
fcis-10209	147	11	-	-	SYM
fcis-10209	147	12	849	849	NUM
fcis-10209	147	13	.	.	PUNCT
fcis-10209	148	1	[	[	X
fcis-10209	148	2	2	2	NUM
fcis-10209	148	3	]	]	X
fcis-10209	148	4	razfar	razfar	ADV
fcis-10209	148	5	,	,	PUNCT
fcis-10209	148	6	n.	n.	ADJ
fcis-10209	148	7	,	,	PUNCT
fcis-10209	148	8	true	true	ADJ
fcis-10209	148	9	,	,	PUNCT
fcis-10209	148	10	j.	j.	PROPN
fcis-10209	148	11	,	,	PUNCT
fcis-10209	148	12	bassiouny	bassiouny	PROPN
fcis-10209	148	13	,	,	PUNCT
fcis-10209	148	14	r.	r.	PROPN
fcis-10209	148	15	,	,	PUNCT
fcis-10209	148	16	venkatesh	venkatesh	PROPN
fcis-10209	148	17	,	,	PUNCT
fcis-10209	148	18	v.	v.	PROPN
fcis-10209	148	19	,	,	PUNCT
fcis-10209	148	20	&	&	CCONJ
fcis-10209	148	21	kashef	kashef	PROPN
fcis-10209	148	22	,	,	PUNCT
fcis-10209	148	23	r.	r.	PROPN
fcis-10209	148	24	(	(	PUNCT
fcis-10209	148	25	2021	2021	NUM
fcis-10209	148	26	)	)	PUNCT
fcis-10209	148	27	.	.	PUNCT
fcis-10209	149	1	weed	weed	NOUN
fcis-10209	149	2	detection	detection	NOUN
fcis-10209	149	3	in	in	ADP
fcis-10209	149	4	soybean	soybean	NOUN
fcis-10209	149	5	crops	crop	NOUN
fcis-10209	149	6	using	use	VERB
fcis-10209	149	7	custom	custom	NOUN
fcis-10209	149	8	lightweight	lightweight	ADJ
fcis-10209	149	9	deep	deep	ADJ
fcis-10209	149	10	learning	learning	NOUN
fcis-10209	149	11	models	model	NOUN
fcis-10209	149	12	.	.	PUNCT
fcis-10209	150	1	computers	computer	NOUN
fcis-10209	150	2	and	and	CCONJ
fcis-10209	150	3	electronics	electronic	NOUN
fcis-10209	150	4	in	in	ADP
fcis-10209	150	5	agriculture	agriculture	NOUN
fcis-10209	150	6	,	,	PUNCT
fcis-10209	150	7	185	185	NUM
fcis-10209	150	8	,	,	PUNCT
fcis-10209	150	9	106016	106016	NUM
fcis-10209	150	10	.	.	PUNCT
fcis-10209	151	1	[	[	X
fcis-10209	151	2	3	3	NUM
fcis-10209	151	3	]	]	SYM
fcis-10209	151	4	wu	wu	PROPN
fcis-10209	151	5	,	,	PUNCT
fcis-10209	151	6	z.	z.	PROPN
fcis-10209	151	7	,	,	PUNCT
fcis-10209	151	8	chen	chen	PROPN
fcis-10209	151	9	,	,	PUNCT
fcis-10209	151	10	y.	y.	PROPN
fcis-10209	151	11	,	,	PUNCT
fcis-10209	151	12	zhao	zhao	PROPN
fcis-10209	151	13	,	,	PUNCT
fcis-10209	151	14	b.	b.	PROPN
fcis-10209	151	15	,	,	PUNCT
fcis-10209	151	16	kang	kang	PROPN
fcis-10209	151	17	,	,	PUNCT
fcis-10209	151	18	x.	x.	PROPN
fcis-10209	151	19	,	,	PUNCT
fcis-10209	151	20	&	&	CCONJ
fcis-10209	151	21	ding	ding	PROPN
fcis-10209	151	22	,	,	PUNCT
fcis-10209	151	23	y.	y.	NOUN
fcis-10209	151	24	(	(	PUNCT
fcis-10209	151	25	2021	2021	NUM
fcis-10209	151	26	)	)	PUNCT
fcis-10209	151	27	.	.	PUNCT
fcis-10209	152	1	review	review	NOUN
fcis-10209	152	2	of	of	ADP
fcis-10209	152	3	weed	weed	NOUN
fcis-10209	152	4	detection	detection	NOUN
fcis-10209	152	5	methods	method	NOUN
fcis-10209	152	6	based	base	VERB
fcis-10209	152	7	on	on	ADP
fcis-10209	152	8	computer	computer	NOUN
fcis-10209	152	9	vision	vision	NOUN
fcis-10209	152	10	.	.	PUNCT
fcis-10209	153	1	frontiers	frontier	NOUN
fcis-10209	153	2	in	in	ADP
fcis-10209	153	3	plant	plant	NOUN
fcis-10209	153	4	science	science	NOUN
fcis-10209	153	5	,	,	PUNCT
fcis-10209	153	6	12	12	NUM
fcis-10209	153	7	,	,	PUNCT
fcis-10209	153	8	634505	634505	NUM
fcis-10209	153	9	.	.	PUNCT
fcis-10209	154	1	[	[	X
fcis-10209	154	2	4	4	NUM
fcis-10209	154	3	]	]	SYM
fcis-10209	154	4	su	su	PROPN
fcis-10209	154	5	,	,	PUNCT
fcis-10209	154	6	w.-h	w.-h	NOUN
fcis-10209	154	7	.	.	PUNCT
fcis-10209	155	1	(	(	PUNCT
fcis-10209	155	2	2021	2021	NUM
fcis-10209	155	3	)	)	PUNCT
fcis-10209	155	4	.	.	PUNCT
fcis-10209	156	1	advanced	advanced	ADJ
fcis-10209	156	2	machine	machine	NOUN
fcis-10209	156	3	learning	learn	VERB
fcis-10209	156	4	in	in	ADP
fcis-10209	156	5	point	point	NOUN
fcis-10209	156	6	spectroscopy	spectroscopy	NOUN
fcis-10209	156	7	,	,	PUNCT
fcis-10209	156	8	rgband	rgband	NOUN
fcis-10209	156	9	hyperspectral	hyperspectral	NOUN
fcis-10209	156	10	-	-	PUNCT
fcis-10209	156	11	imaging	imaging	NOUN
fcis-10209	156	12	for	for	ADP
fcis-10209	156	13	automatic	automatic	ADJ
fcis-10209	156	14	discriminations	discrimination	NOUN
fcis-10209	156	15	of	of	ADP
fcis-10209	156	16	crops	crop	NOUN
fcis-10209	156	17	and	and	CCONJ
fcis-10209	156	18	weeds	weed	NOUN
fcis-10209	156	19	:	:	PUNCT
fcis-10209	156	20	a	a	DET
fcis-10209	156	21	review	review	NOUN
fcis-10209	156	22	.	.	PUNCT
fcis-10209	157	1	sensors	sensor	NOUN
fcis-10209	157	2	,	,	PUNCT
fcis-10209	157	3	21(14	21(14	NUM
fcis-10209	157	4	)	)	PUNCT
fcis-10209	157	5	,	,	PUNCT
fcis-10209	157	6	4707	4707	NUM
fcis-10209	157	7	.	.	PUNCT
fcis-10209	158	1	[	[	X
fcis-10209	158	2	5	5	NUM
fcis-10209	158	3	]	]	SYM
fcis-10209	158	4	tao	tao	PROPN
fcis-10209	158	5	,	,	PUNCT
fcis-10209	158	6	t.	t.	PROPN
fcis-10209	158	7	,	,	PUNCT
fcis-10209	158	8	&	&	CCONJ
fcis-10209	158	9	wei	wei	PROPN
fcis-10209	158	10	,	,	PUNCT
fcis-10209	158	11	x.	x.	NOUN
fcis-10209	158	12	(	(	PUNCT
fcis-10209	158	13	2020	2020	NUM
fcis-10209	158	14	)	)	PUNCT
fcis-10209	158	15	.	.	PUNCT
fcis-10209	159	1	a	a	DET
fcis-10209	159	2	hybrid	hybrid	ADJ
fcis-10209	159	3	cnn	cnn	PROPN
fcis-10209	159	4	-	-	PUNCT
fcis-10209	159	5	svm	svm	ADJ
fcis-10209	159	6	classifier	classifier	NOUN
fcis-10209	159	7	for	for	ADP
fcis-10209	159	8	weed	weed	NOUN
fcis-10209	159	9	recognition	recognition	NOUN
fcis-10209	159	10	in	in	ADP
fcis-10209	159	11	winter	winter	NOUN
fcis-10209	159	12	rape	rape	NOUN
fcis-10209	159	13	field	field	NOUN
fcis-10209	159	14	.	.	PUNCT
fcis-10209	160	1	precision	precision	NOUN
fcis-10209	160	2	agriculture	agriculture	NOUN
fcis-10209	160	3	,	,	PUNCT
fcis-10209	160	4	21(1	21(1	NUM
fcis-10209	160	5	)	)	PUNCT
fcis-10209	160	6	,	,	PUNCT
fcis-10209	160	7	26	26	NUM
fcis-10209	160	8	-	-	SYM
fcis-10209	160	9	37	37	NUM
fcis-10209	160	10	.	.	PUNCT
fcis-10209	161	1	[	[	X
fcis-10209	161	2	6	6	NUM
fcis-10209	161	3	]	]	X
fcis-10209	161	4	cristóbal	cristóbal	PROPN
fcis-10209	161	5	,	,	PUNCT
fcis-10209	161	6	j.	j.	PROPN
fcis-10209	161	7	,	,	PUNCT
fcis-10209	161	8	moreda	moreda	PROPN
fcis-10209	161	9	,	,	PUNCT
fcis-10209	161	10	g.	g.	PROPN
fcis-10209	161	11	p.	p.	PROPN
fcis-10209	161	12	,	,	PUNCT
fcis-10209	161	13	&	&	CCONJ
fcis-10209	161	14	muñoz	muñoz	PROPN
fcis-10209	161	15	-	-	PUNCT
fcis-10209	161	16	rodríguez	rodríguez	PROPN
fcis-10209	161	17	,	,	PUNCT
fcis-10209	161	18	m.	m.	NOUN
fcis-10209	161	19	(	(	PUNCT
fcis-10209	161	20	2006	2006	NUM
fcis-10209	161	21	)	)	PUNCT
fcis-10209	161	22	.	.	PUNCT
fcis-10209	162	1	support	support	NOUN
fcis-10209	162	2	vector	vector	NOUN
fcis-10209	162	3	machine	machine	NOUN
fcis-10209	162	4	classification	classification	NOUN
fcis-10209	162	5	to	to	PART
fcis-10209	162	6	localize	localize	VERB
fcis-10209	162	7	weeds	weed	NOUN
fcis-10209	162	8	in	in	ADP
fcis-10209	162	9	images	image	NOUN
fcis-10209	162	10	of	of	ADP
fcis-10209	162	11	a	a	DET
fcis-10209	162	12	maize	maize	NOUN
fcis-10209	162	13	field	field	NOUN
fcis-10209	162	14	.	.	PUNCT
fcis-10209	163	1	spanish	spanish	ADJ
fcis-10209	163	2	journal	journal	NOUN
fcis-10209	163	3	of	of	ADP
fcis-10209	163	4	agricultural	agricultural	ADJ
fcis-10209	163	5	research	research	NOUN
fcis-10209	163	6	,	,	PUNCT
fcis-10209	163	7	4(4	4(4	NUM
fcis-10209	163	8	)	)	PUNCT
fcis-10209	163	9	,	,	PUNCT
fcis-10209	163	10	433	433	NUM
fcis-10209	163	11	-	-	SYM
fcis-10209	163	12	444	444	NUM
fcis-10209	163	13	.	.	PUNCT
fcis-10209	163	14	77	77	NUM
fcis-10209	164	1	[	[	X
fcis-10209	164	2	7	7	NUM
fcis-10209	164	3	]	]	PUNCT
fcis-10209	164	4	lottes	lotte	NOUN
fcis-10209	164	5	,	,	PUNCT
fcis-10209	164	6	p.	p.	NOUN
fcis-10209	164	7	,	,	PUNCT
fcis-10209	164	8	&	&	CCONJ
fcis-10209	164	9	simard	simard	PROPN
fcis-10209	164	10	,	,	PUNCT
fcis-10209	164	11	p.	p.	NOUN
fcis-10209	164	12	(	(	PUNCT
fcis-10209	164	13	2010	2010	NUM
fcis-10209	164	14	)	)	PUNCT
fcis-10209	164	15	.	.	PUNCT
fcis-10209	165	1	weed	weed	VERB
fcis-10209	165	2	species	specie	NOUN
fcis-10209	165	3	detection	detection	NOUN
fcis-10209	165	4	using	use	VERB
fcis-10209	165	5	dense	dense	ADJ
fcis-10209	165	6	stereo	stereo	NOUN
fcis-10209	165	7	vision	vision	NOUN
fcis-10209	165	8	.	.	PUNCT
fcis-10209	166	1	proceedings	proceeding	NOUN
fcis-10209	166	2	of	of	ADP
fcis-10209	166	3	the	the	DET
fcis-10209	166	4	17th	17th	ADJ
fcis-10209	166	5	international	international	ADJ
fcis-10209	166	6	conference	conference	NOUN
fcis-10209	166	7	on	on	ADP
fcis-10209	166	8	image	image	NOUN
fcis-10209	166	9	processing	processing	NOUN
fcis-10209	166	10	(	(	PUNCT
fcis-10209	166	11	icip	icip	PROPN
fcis-10209	166	12	)	)	PUNCT
fcis-10209	166	13	,	,	PUNCT
fcis-10209	166	14	1337	1337	NUM
fcis-10209	166	15	-	-	SYM
fcis-10209	166	16	1340	1340	NUM
fcis-10209	166	17	.	.	PUNCT
fcis-10209	167	1	zhang	zhang	PROPN
fcis-10209	167	2	,	,	PUNCT
fcis-10209	167	3	m.	m.	NOUN
fcis-10209	167	4	,	,	PUNCT
fcis-10209	167	5	liu	liu	PROPN
fcis-10209	167	6	,	,	PUNCT
fcis-10209	167	7	h.	h.	PROPN
fcis-10209	167	8	,	,	PUNCT
fcis-10209	167	9	dong	dong	PROPN
fcis-10209	167	10	,	,	PUNCT
fcis-10209	167	11	t.	t.	PROPN
fcis-10209	167	12	,	,	PUNCT
fcis-10209	167	13	&	&	CCONJ
fcis-10209	167	14	slaughter	slaughter	PROPN
fcis-10209	167	15	,	,	PUNCT
fcis-10209	167	16	d.	d.	PROPN
fcis-10209	167	17	c.	c.	PROPN
fcis-10209	167	18	(	(	PUNCT
fcis-10209	167	19	2015	2015	NUM
fcis-10209	167	20	)	)	PUNCT
fcis-10209	167	21	.	.	PUNCT
fcis-10209	168	1	automated	automate	VERB
fcis-10209	168	2	weed	weed	NOUN
fcis-10209	168	3	identification	identification	NOUN
fcis-10209	168	4	using	use	VERB
fcis-10209	168	5	an	an	DET
fcis-10209	168	6	ensemble	ensemble	NOUN
fcis-10209	168	7	of	of	ADP
fcis-10209	168	8	optimized	optimize	VERB
fcis-10209	168	9	segmentation	segmentation	NOUN
fcis-10209	168	10	and	and	CCONJ
fcis-10209	168	11	classification	classification	NOUN
fcis-10209	168	12	methods	method	NOUN
fcis-10209	168	13	.	.	PUNCT
fcis-10209	169	1	computers	computer	NOUN
fcis-10209	169	2	and	and	CCONJ
fcis-10209	169	3	electronics	electronic	NOUN
fcis-10209	169	4	in	in	ADP
fcis-10209	169	5	agriculture	agriculture	NOUN
fcis-10209	169	6	,	,	PUNCT
fcis-10209	169	7	116	116	NUM
fcis-10209	169	8	,	,	PUNCT
fcis-10209	169	9	225	225	NUM
fcis-10209	169	10	-	-	SYM
fcis-10209	169	11	232	232	NUM
fcis-10209	169	12	.	.	PUNCT
fcis-10209	170	1	[	[	X
fcis-10209	170	2	8	8	NUM
fcis-10209	170	3	]	]	SYM
fcis-10209	170	4	hu	hu	PROPN
fcis-10209	170	5	,	,	PUNCT
fcis-10209	170	6	k.	k.	PROPN
fcis-10209	170	7	,	,	PUNCT
fcis-10209	170	8	wang	wang	PROPN
fcis-10209	170	9	,	,	PUNCT
fcis-10209	170	10	z.	z.	PROPN
fcis-10209	170	11	,	,	PUNCT
fcis-10209	170	12	coleman	coleman	PROPN
fcis-10209	170	13	,	,	PUNCT
fcis-10209	170	14	g.	g.	PROPN
fcis-10209	170	15	,	,	PUNCT
fcis-10209	170	16	bender	bender	PROPN
fcis-10209	170	17	,	,	PUNCT
fcis-10209	170	18	a.	a.	PROPN
fcis-10209	170	19	,	,	PUNCT
fcis-10209	170	20	yao	yao	PROPN
fcis-10209	170	21	,	,	PUNCT
fcis-10209	170	22	t.	t.	PROPN
fcis-10209	170	23	,	,	PUNCT
fcis-10209	170	24	zeng	zeng	PROPN
fcis-10209	170	25	,	,	PUNCT
fcis-10209	170	26	s.	s.	PROPN
fcis-10209	170	27	,	,	PUNCT
fcis-10209	170	28	song	song	NOUN
fcis-10209	170	29	,	,	PUNCT
fcis-10209	170	30	d.	d.	PROPN
fcis-10209	170	31	,	,	PUNCT
fcis-10209	170	32	schumann	schumann	PROPN
fcis-10209	170	33	,	,	PUNCT
fcis-10209	170	34	a.	a.	PROPN
fcis-10209	170	35	,	,	PUNCT
fcis-10209	170	36	&	&	CCONJ
fcis-10209	170	37	walsh	walsh	PROPN
fcis-10209	170	38	,	,	PUNCT
fcis-10209	170	39	m.	m.	NOUN
fcis-10209	170	40	(	(	PUNCT
fcis-10209	170	41	2020	2020	NUM
fcis-10209	170	42	)	)	PUNCT
fcis-10209	170	43	.	.	PUNCT
fcis-10209	171	1	deep	deep	ADJ
fcis-10209	171	2	learning	learning	NOUN
fcis-10209	171	3	techniques	technique	NOUN
fcis-10209	171	4	for	for	ADP
fcis-10209	171	5	in	in	ADP
fcis-10209	171	6	-	-	PUNCT
fcis-10209	171	7	crop	crop	NOUN
fcis-10209	171	8	weed	weed	NOUN
fcis-10209	171	9	identification	identification	NOUN
fcis-10209	171	10	:	:	PUNCT
fcis-10209	171	11	a	a	DET
fcis-10209	171	12	review	review	NOUN
fcis-10209	171	13	.	.	PUNCT
fcis-10209	172	1	computers	computer	NOUN
fcis-10209	172	2	and	and	CCONJ
fcis-10209	172	3	electronics	electronic	NOUN
fcis-10209	172	4	in	in	ADP
fcis-10209	172	5	agriculture	agriculture	NOUN
fcis-10209	172	6	,	,	PUNCT
fcis-10209	172	7	178	178	NUM
fcis-10209	172	8	,	,	PUNCT
fcis-10209	172	9	105715	105715	NUM
fcis-10209	172	10	.	.	PUNCT
fcis-10209	173	1	[	[	X
fcis-10209	173	2	9	9	NUM
fcis-10209	173	3	]	]	X
fcis-10209	173	4	espejo	espejo	PROPN
fcis-10209	173	5	-	-	PUNCT
fcis-10209	173	6	garcia	garcia	PROPN
fcis-10209	173	7	,	,	PUNCT
fcis-10209	173	8	b.	b.	PROPN
fcis-10209	173	9	,	,	PUNCT
fcis-10209	173	10	mylonas	mylona	NOUN
fcis-10209	173	11	,	,	PUNCT
fcis-10209	173	12	n.	n.	NOUN
fcis-10209	173	13	,	,	PUNCT
fcis-10209	173	14	athanasakos	athanasakos	PROPN
fcis-10209	173	15	,	,	PUNCT
fcis-10209	173	16	l.	l.	PROPN
fcis-10209	173	17	,	,	PUNCT
fcis-10209	173	18	fountas	founta	NOUN
fcis-10209	173	19	,	,	PUNCT
fcis-10209	173	20	s.	s.	PROPN
fcis-10209	173	21	,	,	PUNCT
fcis-10209	173	22	&	&	CCONJ
fcis-10209	173	23	vasilakoglou	vasilakoglou	PROPN
fcis-10209	173	24	,	,	PUNCT
fcis-10209	173	25	i.	i.	PROPN
fcis-10209	173	26	(	(	PUNCT
fcis-10209	173	27	2019	2019	NUM
fcis-10209	173	28	)	)	PUNCT
fcis-10209	173	29	.	.	PUNCT
fcis-10209	174	1	towards	towards	ADP
fcis-10209	174	2	weeds	weed	NOUN
fcis-10209	174	3	identification	identification	NOUN
fcis-10209	174	4	assistance	assistance	NOUN
fcis-10209	174	5	through	through	ADP
fcis-10209	174	6	transfer	transfer	NOUN
fcis-10209	174	7	learning	learning	NOUN
fcis-10209	174	8	.	.	PUNCT
fcis-10209	175	1	computers	computer	NOUN
fcis-10209	175	2	and	and	CCONJ
fcis-10209	175	3	electronics	electronic	NOUN
fcis-10209	175	4	in	in	ADP
fcis-10209	175	5	agriculture	agriculture	NOUN
fcis-10209	175	6	,	,	PUNCT
fcis-10209	175	7	162	162	NUM
fcis-10209	175	8	,	,	PUNCT
fcis-10209	175	9	183	183	NUM
fcis-10209	175	10	-	-	SYM
fcis-10209	175	11	193	193	NUM
fcis-10209	175	12	.	.	PUNCT
fcis-10209	176	1	[	[	X
fcis-10209	176	2	10	10	NUM
fcis-10209	176	3	]	]	X
fcis-10209	176	4	wang	wang	PROPN
fcis-10209	176	5	,	,	PUNCT
fcis-10209	176	6	y.	y.	PROPN
fcis-10209	176	7	,	,	PUNCT
fcis-10209	176	8	liu	liu	PROPN
fcis-10209	176	9	,	,	PUNCT
fcis-10209	176	10	h.	h.	PROPN
fcis-10209	176	11	,	,	PUNCT
fcis-10209	176	12	wang	wang	PROPN
fcis-10209	176	13	,	,	PUNCT
fcis-10209	176	14	d.	d.	PROPN
fcis-10209	176	15	,	,	PUNCT
fcis-10209	176	16	&	&	CCONJ
fcis-10209	176	17	liu	liu	PROPN
fcis-10209	176	18	,	,	PUNCT
fcis-10209	176	19	d.	d.	PROPN
fcis-10209	176	20	(	(	PUNCT
fcis-10209	176	21	2020	2020	NUM
fcis-10209	176	22	)	)	PUNCT
fcis-10209	176	23	.	.	PUNCT
fcis-10209	177	1	image	image	NOUN
fcis-10209	177	2	processing	processing	NOUN
fcis-10209	177	3	in	in	ADP
fcis-10209	177	4	fault	fault	NOUN
fcis-10209	177	5	identification	identification	NOUN
fcis-10209	177	6	for	for	ADP
fcis-10209	177	7	power	power	NOUN
fcis-10209	177	8	equipment	equipment	NOUN
fcis-10209	177	9	based	base	VERB
fcis-10209	177	10	on	on	ADP
fcis-10209	177	11	improved	improve	VERB
fcis-10209	177	12	super	super	ADJ
fcis-10209	177	13	green	green	ADJ
fcis-10209	177	14	algorithm	algorithm	NOUN
fcis-10209	177	15	.	.	PUNCT
fcis-10209	178	1	measurement	measurement	NOUN
fcis-10209	178	2	,	,	PUNCT
fcis-10209	178	3	154	154	NUM
fcis-10209	178	4	,	,	PUNCT
fcis-10209	178	5	107503	107503	NUM
fcis-10209	178	6	.	.	PUNCT
fcis-10209	179	1	xiao	xiao	PROPN
fcis-10209	179	2	,	,	PUNCT
fcis-10209	179	3	l.	l.	PROPN
fcis-10209	179	4	,	,	PUNCT
fcis-10209	179	5	ouyang	ouyang	PROPN
fcis-10209	179	6	,	,	PUNCT
fcis-10209	179	7	h.	h.	PROPN
fcis-10209	179	8	,	,	PUNCT
fcis-10209	179	9	&	&	CCONJ
fcis-10209	179	10	fan	fan	PROPN
fcis-10209	179	11	,	,	PUNCT
fcis-10209	179	12	c.	c.	PROPN
fcis-10209	179	13	(	(	PUNCT
fcis-10209	179	14	2019	2019	NUM
fcis-10209	179	15	)	)	PUNCT
fcis-10209	179	16	.	.	PUNCT
fcis-10209	180	1	an	an	DET
fcis-10209	180	2	improved	improved	ADJ
fcis-10209	180	3	otsu	otsu	NOUN
fcis-10209	180	4	method	method	NOUN
fcis-10209	180	5	for	for	ADP
fcis-10209	180	6	threshold	threshold	NOUN
fcis-10209	180	7	segmentation	segmentation	NOUN
fcis-10209	180	8	based	base	VERB
fcis-10209	180	9	on	on	ADP
fcis-10209	180	10	set	set	VERB
fcis-10209	180	11	mapping	mapping	NOUN
fcis-10209	180	12	and	and	CCONJ
fcis-10209	180	13	trapezoid	trapezoid	ADJ
fcis-10209	180	14	region	region	NOUN
fcis-10209	180	15	intercept	intercept	NOUN
fcis-10209	180	16	histogram	histogram	NOUN
fcis-10209	180	17	.	.	PUNCT
fcis-10209	181	1	optik	optik	PROPN
fcis-10209	181	2	,	,	PUNCT
fcis-10209	181	3	180	180	NUM
fcis-10209	181	4	,	,	PUNCT
fcis-10209	181	5	718	718	NUM
fcis-10209	181	6	-	-	SYM
fcis-10209	181	7	726	726	NUM
fcis-10209	181	8	.	.	PUNCT
fcis-10209	182	1	jiang	jiang	PROPN
fcis-10209	182	2	,	,	PUNCT
fcis-10209	182	3	h.	h.	PROPN
fcis-10209	182	4	,	,	PUNCT
fcis-10209	182	5	zhang	zhang	PROPN
fcis-10209	182	6	,	,	PUNCT
fcis-10209	182	7	c.	c.	PROPN
fcis-10209	182	8	,	,	PUNCT
fcis-10209	182	9	qiao	qiao	PROPN
fcis-10209	182	10	,	,	PUNCT
fcis-10209	182	11	y.	y.	PROPN
fcis-10209	182	12	,	,	PUNCT
fcis-10209	182	13	zhang	zhang	PROPN
fcis-10209	182	14	,	,	PUNCT
fcis-10209	182	15	z.	z.	PROPN
fcis-10209	182	16	,	,	PUNCT
fcis-10209	182	17	zhang	zhang	PROPN
fcis-10209	182	18	,	,	PUNCT
fcis-10209	182	19	w.	w.	PROPN
fcis-10209	182	20	,	,	PUNCT
fcis-10209	182	21	&	&	CCONJ
fcis-10209	182	22	song	song	PROPN
fcis-10209	182	23	,	,	PUNCT
fcis-10209	182	24	c.	c.	PROPN
fcis-10209	182	25	(	(	PUNCT
fcis-10209	182	26	2020	2020	NUM
fcis-10209	182	27	)	)	PUNCT
fcis-10209	182	28	.	.	PUNCT
fcis-10209	183	1	cnn	cnn	PROPN
fcis-10209	183	2	feature	feature	VERB
fcis-10209	183	3	based	base	VERB
fcis-10209	183	4	graph	graph	NOUN
fcis-10209	183	5	convolutional	convolutional	ADJ
fcis-10209	183	6	network	network	NOUN
fcis-10209	183	7	for	for	ADP
fcis-10209	183	8	weed	weed	NOUN
fcis-10209	183	9	and	and	CCONJ
fcis-10209	183	10	crop	crop	NOUN
fcis-10209	183	11	recognition	recognition	NOUN
fcis-10209	183	12	in	in	ADP
fcis-10209	183	13	smart	smart	ADJ
fcis-10209	183	14	farming	farming	NOUN
fcis-10209	183	15	.	.	PUNCT
fcis-10209	184	1	computers	computer	NOUN
fcis-10209	184	2	and	and	CCONJ
fcis-10209	184	3	electronics	electronic	NOUN
fcis-10209	184	4	in	in	ADP
fcis-10209	184	5	agriculture	agriculture	NOUN
fcis-10209	184	6	,	,	PUNCT
fcis-10209	184	7	178	178	NUM
fcis-10209	184	8	,	,	PUNCT
fcis-10209	184	9	105797	105797	NUM
fcis-10209	184	10	.	.	PUNCT
fcis-10209	185	1	[	[	X
fcis-10209	185	2	11	11	NUM
fcis-10209	185	3	]	]	X
fcis-10209	185	4	shah	shah	NOUN
fcis-10209	185	5	,	,	PUNCT
fcis-10209	185	6	d.	d.	PROPN
fcis-10209	185	7	,	,	PUNCT
fcis-10209	185	8	gupta	gupta	PROPN
fcis-10209	185	9	,	,	PUNCT
fcis-10209	185	10	r.	r.	PROPN
fcis-10209	185	11	,	,	PUNCT
fcis-10209	185	12	patel	patel	PROPN
fcis-10209	185	13	,	,	PUNCT
fcis-10209	185	14	k.	k.	PROPN
fcis-10209	185	15	,	,	PUNCT
fcis-10209	185	16	jariwala	jariwala	PROPN
fcis-10209	185	17	,	,	PUNCT
fcis-10209	185	18	d.	d.	PROPN
fcis-10209	185	19	,	,	PUNCT
fcis-10209	185	20	&	&	CCONJ
fcis-10209	185	21	kanani	kanani	PROPN
fcis-10209	185	22	,	,	PUNCT
fcis-10209	185	23	j.	j.	PROPN
fcis-10209	185	24	(	(	PUNCT
fcis-10209	185	25	2021	2021	NUM
fcis-10209	185	26	)	)	PUNCT
fcis-10209	185	27	.	.	PUNCT
fcis-10209	186	1	deep	deep	ADJ
fcis-10209	186	2	learning	learning	NOUN
fcis-10209	186	3	based	base	VERB
fcis-10209	186	4	pest	pest	NOUN
fcis-10209	186	5	classification	classification	NOUN
fcis-10209	186	6	in	in	ADP
fcis-10209	186	7	soybean	soybean	NOUN
fcis-10209	186	8	crop	crop	NOUN
fcis-10209	186	9	using	use	VERB
fcis-10209	186	10	residual	residual	ADJ
fcis-10209	186	11	network-50	network-50	PROPN
fcis-10209	186	12	.	.	PROPN
fcis-10209	186	13	2021	2021	NUM
fcis-10209	186	14	international	international	ADJ
fcis-10209	186	15	conference	conference	NOUN
fcis-10209	186	16	on	on	ADP
fcis-10209	186	17	inventive	inventive	ADJ
fcis-10209	186	18	systems	system	NOUN
fcis-10209	186	19	and	and	CCONJ
fcis-10209	186	20	control	control	NOUN
fcis-10209	186	21	(	(	PUNCT
fcis-10209	186	22	icisc	icisc	NOUN
fcis-10209	186	23	)	)	PUNCT
fcis-10209	186	24	,	,	PUNCT
fcis-10209	186	25	95	95	NUM
fcis-10209	186	26	-	-	SYM
fcis-10209	186	27	99	99	NUM
fcis-10209	186	28	.	.	PUNCT
fcis-10209	187	1	[	[	X
fcis-10209	187	2	12	12	NUM
fcis-10209	187	3	]	]	X
fcis-10209	187	4	zhan	zhan	PROPN
fcis-10209	187	5	,	,	PUNCT
fcis-10209	187	6	z.	z.	PROPN
fcis-10209	187	7	,	,	PUNCT
fcis-10209	187	8	li	li	PROPN
fcis-10209	187	9	,	,	PUNCT
fcis-10209	187	10	y.	y.	PROPN
fcis-10209	187	11	,	,	PUNCT
fcis-10209	187	12	liu	liu	PROPN
fcis-10209	187	13	,	,	PUNCT
fcis-10209	187	14	f.	f.	PROPN
fcis-10209	187	15	,	,	PUNCT
fcis-10209	187	16	zhang	zhang	PROPN
fcis-10209	187	17	,	,	PUNCT
fcis-10209	187	18	h.	h.	PROPN
fcis-10209	187	19	,	,	PUNCT
fcis-10209	187	20	&	&	CCONJ
fcis-10209	187	21	xu	xu	PROPN
fcis-10209	187	22	,	,	PUNCT
fcis-10209	187	23	l.	l.	PROPN
fcis-10209	187	24	(	(	PUNCT
fcis-10209	187	25	2021	2021	NUM
fcis-10209	187	26	)	)	PUNCT
fcis-10209	187	27	.	.	PUNCT
fcis-10209	188	1	semantic	semantic	ADJ
fcis-10209	188	2	segmentation	segmentation	NOUN
fcis-10209	188	3	of	of	ADP
fcis-10209	188	4	agricultural	agricultural	ADJ
fcis-10209	188	5	crops	crop	NOUN
fcis-10209	188	6	using	use	VERB
fcis-10209	188	7	a	a	DET
fcis-10209	188	8	fully	fully	ADV
fcis-10209	188	9	convolutional	convolutional	ADJ
fcis-10209	188	10	neural	neural	ADJ
fcis-10209	188	11	network	network	NOUN
fcis-10209	188	12	with	with	ADP
fcis-10209	188	13	dilated	dilated	ADJ
fcis-10209	188	14	convolutions	convolution	NOUN
fcis-10209	188	15	.	.	PUNCT
fcis-10209	189	1	remote	remote	ADJ
fcis-10209	189	2	sensing	sensing	NOUN
fcis-10209	189	3	,	,	PUNCT
fcis-10209	189	4	13(2	13(2	PROPN
fcis-10209	189	5	)	)	PUNCT
fcis-10209	189	6	,	,	PUNCT
fcis-10209	189	7	285	285	NUM
fcis-10209	189	8	.	.	PUNCT
fcis-10209	190	1	[	[	X
fcis-10209	190	2	13	13	NUM
fcis-10209	190	3	]	]	X
fcis-10209	190	4	zhu	zhu	PROPN
fcis-10209	190	5	,	,	PUNCT
fcis-10209	190	6	b.	b.	PROPN
fcis-10209	190	7	,	,	PUNCT
fcis-10209	190	8	zhen	zhen	PROPN
fcis-10209	190	9	,	,	PUNCT
fcis-10209	190	10	f.	f.	PROPN
fcis-10209	190	11	,	,	PUNCT
fcis-10209	190	12	li	li	PROPN
fcis-10209	190	13	,	,	PUNCT
fcis-10209	190	14	s.	s.	PROPN
fcis-10209	190	15	,	,	PUNCT
fcis-10209	190	16	hu	hu	PROPN
fcis-10209	190	17	,	,	PUNCT
fcis-10209	190	18	j.	j.	PROPN
fcis-10209	190	19	,	,	PUNCT
fcis-10209	190	20	&	&	CCONJ
fcis-10209	190	21	liu	liu	PROPN
fcis-10209	190	22	,	,	PUNCT
fcis-10209	190	23	y.	y.	PROPN
fcis-10209	190	24	(	(	PUNCT
fcis-10209	190	25	2021	2021	NUM
fcis-10209	190	26	)	)	PUNCT
fcis-10209	190	27	.	.	PUNCT
fcis-10209	191	1	crop	crop	NOUN
fcis-10209	191	2	segmentation	segmentation	NOUN
fcis-10209	191	3	with	with	ADP
fcis-10209	191	4	an	an	DET
fcis-10209	191	5	improved	improved	ADJ
fcis-10209	191	6	u	u	ADJ
fcis-10209	191	7	-	-	ADJ
fcis-10209	191	8	net	net	ADJ
fcis-10209	191	9	model	model	NOUN
fcis-10209	191	10	based	base	VERB
fcis-10209	191	11	on	on	ADP
fcis-10209	191	12	selfattention	selfattention	NOUN
fcis-10209	191	13	mechanism	mechanism	NOUN
fcis-10209	191	14	.	.	PUNCT
fcis-10209	192	1	remote	remote	ADJ
fcis-10209	192	2	sensing	sensing	NOUN
fcis-10209	192	3	,	,	PUNCT
fcis-10209	192	4	13(9	13(9	NUM
fcis-10209	192	5	)	)	PUNCT
fcis-10209	192	6	,	,	PUNCT
fcis-10209	192	7	1857	1857	NUM
fcis-10209	192	8	.	.	PUNCT
fcis-10209	193	1	doi	doi	NOUN
fcis-10209	193	2	:	:	PUNCT
fcis-10209	193	3	10.3390	10.3390	NUM
fcis-10209	193	4	/	/	SYM
fcis-10209	193	5	rs13091857	rs13091857	NOUN
fcis-10209	193	6	.	.	PUNCT
fcis-10209	194	1	[	[	X
fcis-10209	194	2	14	14	NUM
fcis-10209	194	3	]	]	SYM
fcis-10209	194	4	li	li	PROPN
fcis-10209	194	5	,	,	PUNCT
fcis-10209	194	6	j.	j.	PROPN
fcis-10209	194	7	,	,	PUNCT
fcis-10209	194	8	xia	xia	PROPN
fcis-10209	194	9	,	,	PUNCT
fcis-10209	194	10	x.	x.	PROPN
fcis-10209	194	11	,	,	PUNCT
fcis-10209	194	12	li	li	PROPN
fcis-10209	194	13	,	,	PUNCT
fcis-10209	194	14	w.	w.	PROPN
fcis-10209	194	15	,	,	PUNCT
fcis-10209	194	16	li	li	PROPN
fcis-10209	194	17	,	,	PUNCT
fcis-10209	194	18	h.	h.	PROPN
fcis-10209	194	19	,	,	PUNCT
fcis-10209	194	20	wang	wang	PROPN
fcis-10209	194	21	,	,	PUNCT
fcis-10209	194	22	x.	x.	PROPN
fcis-10209	194	23	,	,	PUNCT
fcis-10209	194	24	xiao	xiao	PROPN
fcis-10209	194	25	,	,	PUNCT
fcis-10209	194	26	x.	x.	PROPN
fcis-10209	194	27	,	,	PUNCT
fcis-10209	194	28	wang	wang	PROPN
fcis-10209	194	29	,	,	PUNCT
fcis-10209	194	30	r.	r.	PROPN
fcis-10209	194	31	,	,	PUNCT
fcis-10209	194	32	zheng	zheng	PROPN
fcis-10209	194	33	,	,	PUNCT
fcis-10209	194	34	m.	m.	NOUN
fcis-10209	194	35	,	,	PUNCT
fcis-10209	194	36	&	&	CCONJ
fcis-10209	194	37	pan	pan	PROPN
fcis-10209	194	38	,	,	PUNCT
fcis-10209	194	39	x.	x.	NOUN
fcis-10209	194	40	(	(	PUNCT
fcis-10209	194	41	2022	2022	NUM
fcis-10209	194	42	)	)	PUNCT
fcis-10209	194	43	.	.	PUNCT
fcis-10209	195	1	next	next	ADJ
fcis-10209	195	2	-	-	PUNCT
fcis-10209	195	3	vit	vit	NOUN
fcis-10209	195	4	:	:	PUNCT
fcis-10209	195	5	next	next	ADJ
fcis-10209	195	6	generation	generation	NOUN
fcis-10209	195	7	vision	vision	NOUN
fcis-10209	195	8	transformer	transformer	NOUN
fcis-10209	195	9	for	for	ADP
fcis-10209	195	10	efficient	efficient	ADJ
fcis-10209	195	11	deployment	deployment	NOUN
fcis-10209	195	12	in	in	ADP
fcis-10209	195	13	realistic	realistic	ADJ
fcis-10209	195	14	industrial	industrial	ADJ
fcis-10209	195	15	scenarios	scenario	NOUN
fcis-10209	195	16	.	.	PUNCT
fcis-10209	196	1	ieee	ieee	NOUN
fcis-10209	196	2	/	/	SYM
fcis-10209	196	3	cvf	cvf	NOUN
fcis-10209	196	4	international	international	ADJ
fcis-10209	196	5	conference	conference	NOUN
fcis-10209	196	6	on	on	ADP
fcis-10209	196	7	computer	computer	NOUN
fcis-10209	196	8	vision	vision	NOUN
fcis-10209	196	9	and	and	CCONJ
fcis-10209	196	10	pattern	pattern	NOUN
fcis-10209	196	11	recognition	recognition	NOUN
fcis-10209	196	12	(	(	PUNCT
fcis-10209	196	13	cvpr	cvpr	NOUN
fcis-10209	196	14	)	)	PUNCT
fcis-10209	196	15	workshops	workshop	NOUN
fcis-10209	196	16	.	.	PUNCT
fcis-10209	197	1	[	[	X
fcis-10209	197	2	15	15	NUM
fcis-10209	197	3	]	]	X
fcis-10209	197	4	maaz	maaz	PROPN
fcis-10209	197	5	,	,	PUNCT
fcis-10209	197	6	m.	m.	NOUN
fcis-10209	197	7	,	,	PUNCT
fcis-10209	197	8	shaker	shaker	NOUN
fcis-10209	197	9	,	,	PUNCT
fcis-10209	197	10	a.	a.	NOUN
fcis-10209	197	11	,	,	PUNCT
fcis-10209	197	12	cholakkal	cholakkal	NOUN
fcis-10209	197	13	,	,	PUNCT
fcis-10209	197	14	h.	h.	PROPN
fcis-10209	197	15	,	,	PUNCT
fcis-10209	197	16	khan	khan	PROPN
fcis-10209	197	17	,	,	PUNCT
fcis-10209	197	18	s.	s.	PROPN
fcis-10209	197	19	,	,	PUNCT
fcis-10209	197	20	zamir	zamir	PROPN
fcis-10209	197	21	,	,	PUNCT
fcis-10209	197	22	s.	s.	PROPN
fcis-10209	197	23	w.	w.	PROPN
fcis-10209	197	24	,	,	PUNCT
fcis-10209	197	25	anwer	anwer	NOUN
fcis-10209	197	26	,	,	PUNCT
fcis-10209	197	27	r.	r.	PROPN
fcis-10209	197	28	m.	m.	PROPN
fcis-10209	197	29	,	,	PUNCT
fcis-10209	197	30	&	&	CCONJ
fcis-10209	197	31	khan	khan	PROPN
fcis-10209	197	32	,	,	PUNCT
fcis-10209	197	33	f.	f.	PROPN
fcis-10209	197	34	s.	s.	PROPN
fcis-10209	197	35	(	(	PUNCT
fcis-10209	197	36	2023	2023	NUM
fcis-10209	197	37	)	)	PUNCT
fcis-10209	197	38	.	.	PUNCT
fcis-10209	198	1	edgenext	edgenext	NOUN
fcis-10209	198	2	:	:	PUNCT
fcis-10209	198	3	efficiently	efficiently	ADV
fcis-10209	198	4	amalgamated	amalgamate	VERB
fcis-10209	198	5	cnn	cnn	PROPN
fcis-10209	198	6	-	-	PUNCT
fcis-10209	198	7	transformer	transformer	NOUN
fcis-10209	198	8	architecture	architecture	NOUN
fcis-10209	198	9	for	for	ADP
fcis-10209	198	10	mobile	mobile	ADJ
fcis-10209	198	11	vision	vision	NOUN
fcis-10209	198	12	applications	application	NOUN
fcis-10209	198	13	.	.	PUNCT
fcis-10209	199	1	in	in	ADP
fcis-10209	199	2	l.	l.	PROPN
fcis-10209	199	3	karlinsky	karlinsky	PROPN
fcis-10209	199	4	,	,	PUNCT
fcis-10209	199	5	t.	t.	PROPN
fcis-10209	199	6	michaeli	michaeli	PROPN
fcis-10209	199	7	,	,	PUNCT
fcis-10209	199	8	&	&	CCONJ
fcis-10209	199	9	k.	k.	PROPN
fcis-10209	199	10	nishino	nishino	PROPN
fcis-10209	199	11	(	(	PUNCT
fcis-10209	199	12	eds	ed	NOUN
fcis-10209	199	13	.	.	PUNCT
fcis-10209	199	14	)	)	PUNCT
fcis-10209	199	15	,	,	PUNCT
fcis-10209	199	16	computer	computer	NOUN
fcis-10209	199	17	vision	vision	NOUN
fcis-10209	199	18	–	–	PUNCT
fcis-10209	199	19	eccv	eccv	ADJ
fcis-10209	199	20	2022	2022	NUM
fcis-10209	199	21	workshops	workshop	NOUN
fcis-10209	199	22	.	.	PUNCT
fcis-10209	200	1	eccv	eccv	ADJ
fcis-10209	200	2	2022	2022	NUM
fcis-10209	200	3	.	.	PUNCT
fcis-10209	201	1	lecture	lecture	NOUN
fcis-10209	201	2	notes	note	NOUN
fcis-10209	201	3	in	in	ADP
fcis-10209	201	4	computer	computer	NOUN
fcis-10209	201	5	science	science	NOUN
fcis-10209	201	6	,	,	PUNCT
fcis-10209	201	7	vol	vol	NOUN
fcis-10209	201	8	13807	13807	NUM
fcis-10209	201	9	.	.	PUNCT
fcis-10209	202	1	springer	springer	NOUN
fcis-10209	202	2	,	,	PUNCT
fcis-10209	202	3	cham	cham	PROPN
fcis-10209	202	4	.	.	PUNCT
fcis-10209	203	1	[	[	X
fcis-10209	203	2	16	16	NUM
fcis-10209	203	3	]	]	X
fcis-10209	203	4	olsen	olsen	PROPN
fcis-10209	203	5	,	,	PUNCT
fcis-10209	203	6	a.	a.	PROPN
fcis-10209	203	7	,	,	PUNCT
fcis-10209	203	8	konvalina	konvalina	PROPN
fcis-10209	203	9	,	,	PUNCT
fcis-10209	203	10	d.	d.	PROPN
fcis-10209	203	11	a.	a.	PROPN
fcis-10209	203	12	,	,	PUNCT
fcis-10209	203	13	philippa	philippa	PROPN
fcis-10209	203	14	,	,	PUNCT
fcis-10209	203	15	b.	b.	PROPN
fcis-10209	203	16	,	,	PUNCT
fcis-10209	203	17	et	et	PROPN
fcis-10209	203	18	al	al	PROPN
fcis-10209	203	19	.	.	PROPN
fcis-10209	204	1	(	(	PUNCT
fcis-10209	204	2	2019	2019	NUM
fcis-10209	204	3	)	)	PUNCT
fcis-10209	204	4	.	.	PUNCT
fcis-10209	205	1	deepweeds	deepweed	NOUN
fcis-10209	205	2	:	:	PUNCT
fcis-10209	205	3	a	a	DET
fcis-10209	205	4	multiclass	multiclass	ADJ
fcis-10209	205	5	weed	weed	NOUN
fcis-10209	205	6	species	species	NOUN
fcis-10209	205	7	image	image	NOUN
fcis-10209	205	8	dataset	dataset	VERB
fcis-10209	205	9	for	for	ADP
fcis-10209	205	10	deep	deep	ADJ
fcis-10209	205	11	learning	learning	NOUN
fcis-10209	205	12	.	.	PUNCT
fcis-10209	206	1	scientific	scientific	ADJ
fcis-10209	206	2	reports	report	NOUN
fcis-10209	206	3	,	,	PUNCT
fcis-10209	206	4	9(1	9(1	NUM
fcis-10209	206	5	)	)	PUNCT
fcis-10209	206	6	,	,	PUNCT
fcis-10209	206	7	1	1	NUM
fcis-10209	206	8	-	-	SYM
fcis-10209	206	9	12	12	NUM
fcis-10209	206	10	.	.	PUNCT
fcis-10209	207	1	[	[	X
fcis-10209	207	2	17	17	NUM
fcis-10209	207	3	]	]	X
fcis-10209	207	4	dosovitskiy	dosovitskiy	NOUN
fcis-10209	207	5	,	,	PUNCT
fcis-10209	207	6	a.	a.	PROPN
fcis-10209	207	7	,	,	PUNCT
fcis-10209	207	8	beyer	beyer	PROPN
fcis-10209	207	9	,	,	PUNCT
fcis-10209	207	10	l.	l.	PROPN
fcis-10209	207	11	,	,	PUNCT
fcis-10209	207	12	kolesnikov	kolesnikov	PROPN
fcis-10209	207	13	,	,	PUNCT
fcis-10209	207	14	a.	a.	NOUN
fcis-10209	207	15	,	,	PUNCT
fcis-10209	207	16	weissenborn	weissenborn	ADJ
fcis-10209	207	17	,	,	PUNCT
fcis-10209	207	18	d.	d.	PROPN
fcis-10209	207	19	,	,	PUNCT
fcis-10209	207	20	zhai	zhai	PROPN
fcis-10209	207	21	,	,	PUNCT
fcis-10209	207	22	x.	x.	PROPN
fcis-10209	207	23	,	,	PUNCT
fcis-10209	207	24	unterthiner	unterthiner	NOUN
fcis-10209	207	25	,	,	PUNCT
fcis-10209	207	26	t.	t.	PROPN
fcis-10209	207	27	,	,	PUNCT
fcis-10209	207	28	...	...	PUNCT
fcis-10209	207	29	&	&	CCONJ
fcis-10209	207	30	houlsby	houlsby	ADJ
fcis-10209	207	31	,	,	PUNCT
fcis-10209	207	32	n.	n.	NOUN
fcis-10209	207	33	(	(	PUNCT
fcis-10209	207	34	2021	2021	NUM
fcis-10209	207	35	)	)	PUNCT
fcis-10209	207	36	.	.	PUNCT
fcis-10209	208	1	an	an	DET
fcis-10209	208	2	image	image	NOUN
fcis-10209	208	3	is	be	AUX
fcis-10209	208	4	worth	worth	ADJ
fcis-10209	208	5	16x16	16x16	NUM
fcis-10209	208	6	words	word	NOUN
fcis-10209	208	7	:	:	PUNCT
fcis-10209	208	8	transformers	transformer	NOUN
fcis-10209	208	9	for	for	ADP
fcis-10209	208	10	image	image	NOUN
fcis-10209	208	11	recognition	recognition	NOUN
fcis-10209	208	12	at	at	ADP
fcis-10209	208	13	scale	scale	NOUN
fcis-10209	208	14	.	.	PUNCT
fcis-10209	209	1	arxiv	arxiv	PROPN
fcis-10209	209	2	preprint	preprint	VERB
fcis-10209	209	3	arxiv:2010.11929	arxiv:2010.11929	NOUN
fcis-10209	209	4	.	.	PUNCT
fcis-10209	210	1	[	[	X
fcis-10209	210	2	18	18	NUM
fcis-10209	210	3	]	]	X
fcis-10209	210	4	wang	wang	PROPN
fcis-10209	210	5	,	,	PUNCT
fcis-10209	210	6	x.	x.	PROPN
fcis-10209	210	7	,	,	PUNCT
fcis-10209	210	8	chan	chan	PROPN
fcis-10209	210	9	,	,	PUNCT
fcis-10209	210	10	k.	k.	PROPN
fcis-10209	210	11	,	,	PUNCT
fcis-10209	210	12	gao	gao	PROPN
fcis-10209	210	13	,	,	PUNCT
fcis-10209	210	14	y.	y.	PROPN
fcis-10209	210	15	,	,	PUNCT
fcis-10209	210	16	&	&	CCONJ
fcis-10209	210	17	yang	yang	PROPN
fcis-10209	210	18	,	,	PUNCT
fcis-10209	210	19	m.	m.	PROPN
fcis-10209	210	20	h.	h.	PROPN
fcis-10209	210	21	(	(	PUNCT
fcis-10209	210	22	2020	2020	NUM
fcis-10209	210	23	)	)	PUNCT
fcis-10209	210	24	.	.	PUNCT
fcis-10209	211	1	training	train	VERB
fcis-10209	211	2	attention	attention	NOUN
fcis-10209	211	3	networks	network	NOUN
fcis-10209	211	4	for	for	ADP
fcis-10209	211	5	high	high	ADJ
fcis-10209	211	6	-	-	PUNCT
fcis-10209	211	7	resolution	resolution	NOUN
fcis-10209	211	8	image	image	NOUN
fcis-10209	211	9	processing	processing	NOUN
fcis-10209	211	10	:	:	PUNCT
fcis-10209	211	11	an	an	DET
fcis-10209	211	12	empirical	empirical	ADJ
fcis-10209	211	13	study	study	NOUN
fcis-10209	211	14	.	.	PUNCT
fcis-10209	212	1	proceedings	proceeding	NOUN
fcis-10209	212	2	of	of	ADP
fcis-10209	212	3	the	the	DET
fcis-10209	212	4	ieee	ieee	NOUN
fcis-10209	212	5	/	/	SYM
fcis-10209	212	6	cvf	cvf	NOUN
fcis-10209	212	7	conference	conference	NOUN
fcis-10209	212	8	on	on	ADP
fcis-10209	212	9	computer	computer	NOUN
fcis-10209	212	10	vision	vision	NOUN
fcis-10209	212	11	and	and	CCONJ
fcis-10209	212	12	pattern	pattern	NOUN
fcis-10209	212	13	recognition	recognition	NOUN
fcis-10209	212	14	(	(	PUNCT
fcis-10209	212	15	cvpr	cvpr	NOUN
fcis-10209	212	16	)	)	PUNCT
fcis-10209	212	17	,	,	PUNCT
fcis-10209	212	18	1043210441	1043210441	NUM
fcis-10209	212	19	.	.	PUNCT
fcis-10209	213	1	[	[	X
fcis-10209	213	2	19	19	NUM
fcis-10209	213	3	]	]	X
fcis-10209	213	4	kolesnikov	kolesnikov	PROPN
fcis-10209	213	5	,	,	PUNCT
fcis-10209	213	6	a.	a.	PROPN
fcis-10209	213	7	,	,	PUNCT
fcis-10209	213	8	lampert	lampert	PROPN
fcis-10209	213	9	,	,	PUNCT
fcis-10209	213	10	c.	c.	PROPN
fcis-10209	213	11	h.	h.	PROPN
fcis-10209	213	12	,	,	PUNCT
fcis-10209	213	13	&	&	CCONJ
fcis-10209	213	14	abdelsalam	abdelsalam	PROPN
fcis-10209	213	15	,	,	PUNCT
fcis-10209	213	16	m.	m.	NOUN
fcis-10209	213	17	(	(	PUNCT
fcis-10209	213	18	2021	2021	NUM
fcis-10209	213	19	)	)	PUNCT
fcis-10209	213	20	.	.	PUNCT
fcis-10209	214	1	improved	improved	ADJ
fcis-10209	214	2	convolutions	convolution	NOUN
fcis-10209	214	3	via	via	ADP
fcis-10209	214	4	hebbian	hebbian	ADJ
fcis-10209	214	5	block	block	NOUN
fcis-10209	214	6	-	-	PUNCT
fcis-10209	214	7	sparse	sparse	ADJ
fcis-10209	214	8	kernel	kernel	NOUN
fcis-10209	214	9	and	and	CCONJ
fcis-10209	214	10	convex	convex	VERB
fcis-10209	214	11	optimization	optimization	NOUN
fcis-10209	214	12	.	.	PUNCT
fcis-10209	215	1	proceedings	proceeding	NOUN
fcis-10209	215	2	of	of	ADP
fcis-10209	215	3	the	the	DET
fcis-10209	215	4	ieee	ieee	NOUN
fcis-10209	215	5	/	/	SYM
fcis-10209	215	6	cvf	cvf	NOUN
fcis-10209	215	7	conference	conference	NOUN
fcis-10209	215	8	on	on	ADP
fcis-10209	215	9	computer	computer	NOUN
fcis-10209	215	10	vision	vision	NOUN
fcis-10209	215	11	and	and	CCONJ
fcis-10209	215	12	pattern	pattern	NOUN
fcis-10209	215	13	recognition	recognition	NOUN
fcis-10209	215	14	(	(	PUNCT
fcis-10209	215	15	cvpr	cvpr	NOUN
fcis-10209	215	16	)	)	PUNCT
fcis-10209	215	17	,	,	PUNCT
fcis-10209	215	18	8144	8144	NUM
fcis-10209	215	19	-	-	SYM
fcis-10209	215	20	8153	8153	NUM
fcis-10209	215	21	.	.	PUNCT