id	sid	tid	token	lemma	pos
ajst-27534	1	1	academic	academic	ADJ
ajst-27534	1	2	journal	journal	NOUN
ajst-27534	1	3	of	of	ADP
ajst-27534	1	4	science	science	NOUN
ajst-27534	1	5	and	and	CCONJ
ajst-27534	1	6	technology	technology	NOUN
ajst-27534	1	7	issn	issn	NOUN
ajst-27534	1	8	:	:	PUNCT
ajst-27534	1	9	2771	2771	NUM
ajst-27534	1	10	-	-	SYM
ajst-27534	1	11	3032	3032	NUM
ajst-27534	1	12	|	|	NOUN
ajst-27534	1	13	vol	vol	NOUN
ajst-27534	1	14	.	.	PROPN
ajst-27534	1	15	13	13	NUM
ajst-27534	1	16	,	,	PUNCT
ajst-27534	1	17	no	no	INTJ
ajst-27534	1	18	.	.	NOUN
ajst-27534	1	19	2	2	NUM
ajst-27534	1	20	,	,	PUNCT
ajst-27534	1	21	2024	2024	NUM
ajst-27534	1	22	201	201	NUM
ajst-27534	1	23	implementation	implementation	NOUN
ajst-27534	1	24	of	of	ADP
ajst-27534	1	25	small	small	ADJ
ajst-27534	1	26	sample	sample	NOUN
ajst-27534	1	27	citrus	citrus	NOUN
ajst-27534	1	28	fruit	fruit	NOUN
ajst-27534	1	29	classification	classification	NOUN
ajst-27534	1	30	detection	detection	NOUN
ajst-27534	1	31	based	base	VERB
ajst-27534	1	32	on	on	ADP
ajst-27534	1	33	improved	improved	ADJ
ajst-27534	1	34	resnet18	resnet18	NOUN
ajst-27534	1	35	residual	residual	ADJ
ajst-27534	1	36	network	network	NOUN
ajst-27534	1	37	hai	hai	PROPN
ajst-27534	1	38	li1	li1	PROPN
ajst-27534	1	39	,	,	PUNCT
ajst-27534	1	40	hua	hua	PROPN
ajst-27534	1	41	qing1	qing1	PROPN
ajst-27534	1	42	,	,	PUNCT
ajst-27534	1	43	shihui	shihui	VERB
ajst-27534	1	44	wu1	wu1	NOUN
ajst-27534	1	45	,	,	PUNCT
ajst-27534	1	46	haiqing	haiqe	VERB
ajst-27534	1	47	bai1	bai1	NOUN
ajst-27534	1	48	,	,	PUNCT
ajst-27534	1	49	bo	bo	PROPN
ajst-27534	1	50	lin2	lin2	PROPN
ajst-27534	1	51	1chongqing	1chongqing	NUM
ajst-27534	1	52	college	college	NOUN
ajst-27534	1	53	of	of	ADP
ajst-27534	1	54	humanities	humanity	NOUN
ajst-27534	1	55	,	,	PUNCT
ajst-27534	1	56	science	science	NOUN
ajst-27534	1	57	and	and	CCONJ
ajst-27534	1	58	technology	technology	NOUN
ajst-27534	1	59	,	,	PUNCT
ajst-27534	1	60	chongqing	chongqe	VERB
ajst-27534	1	61	400000	400000	NUM
ajst-27534	1	62	,	,	PUNCT
ajst-27534	1	63	china	china	PROPN
ajst-27534	1	64	2isuzu	2isuzu	NUM
ajst-27534	1	65	(	(	PUNCT
ajst-27534	1	66	china	china	PROPN
ajst-27534	1	67	)	)	PUNCT
ajst-27534	1	68	engine	engine	PROPN
ajst-27534	1	69	co.	co.	PROPN
ajst-27534	1	70	,	,	PUNCT
ajst-27534	1	71	ltd	ltd	PROPN
ajst-27534	1	72	,	,	PUNCT
ajst-27534	1	73	chongqing	chongqe	VERB
ajst-27534	1	74	400000	400000	NUM
ajst-27534	1	75	,	,	PUNCT
ajst-27534	1	76	china	china	PROPN
ajst-27534	1	77	abstract	abstract	NOUN
ajst-27534	1	78	:	:	PUNCT
ajst-27534	1	79	citrus	citrus	NOUN
ajst-27534	1	80	is	be	AUX
ajst-27534	1	81	an	an	DET
ajst-27534	1	82	important	important	ADJ
ajst-27534	1	83	means	mean	NOUN
ajst-27534	1	84	of	of	ADP
ajst-27534	1	85	production	production	NOUN
ajst-27534	1	86	,	,	PUNCT
ajst-27534	1	87	and	and	CCONJ
ajst-27534	1	88	traditional	traditional	ADJ
ajst-27534	1	89	citrus	citrus	NOUN
ajst-27534	1	90	classification	classification	NOUN
ajst-27534	1	91	using	use	VERB
ajst-27534	1	92	manual	manual	ADJ
ajst-27534	1	93	labor	labor	NOUN
ajst-27534	1	94	is	be	AUX
ajst-27534	1	95	timeconsuming	timeconsuming	ADJ
ajst-27534	1	96	,	,	PUNCT
ajst-27534	1	97	labor	labor	NOUN
ajst-27534	1	98	-	-	PUNCT
ajst-27534	1	99	intensive	intensive	ADJ
ajst-27534	1	100	,	,	PUNCT
ajst-27534	1	101	and	and	CCONJ
ajst-27534	1	102	costly	costly	ADJ
ajst-27534	1	103	.	.	PUNCT
ajst-27534	2	1	this	this	DET
ajst-27534	2	2	article	article	NOUN
ajst-27534	2	3	proposes	propose	VERB
ajst-27534	2	4	a	a	DET
ajst-27534	2	5	new	new	ADJ
ajst-27534	2	6	residual	residual	ADJ
ajst-27534	2	7	block	block	NOUN
ajst-27534	2	8	structure	structure	NOUN
ajst-27534	2	9	based	base	VERB
ajst-27534	2	10	on	on	ADP
ajst-27534	2	11	the	the	DET
ajst-27534	2	12	introduction	introduction	NOUN
ajst-27534	2	13	of	of	ADP
ajst-27534	2	14	pre	pre	ADJ
ajst-27534	2	15	activated	activate	VERB
ajst-27534	2	16	residual	residual	ADJ
ajst-27534	2	17	block	block	NOUN
ajst-27534	2	18	mode	mode	NOUN
ajst-27534	2	19	to	to	PART
ajst-27534	2	20	build	build	VERB
ajst-27534	2	21	the	the	DET
ajst-27534	2	22	resnet18	resnet18	NOUN
ajst-27534	2	23	network	network	NOUN
ajst-27534	2	24	model	model	NOUN
ajst-27534	2	25	,	,	PUNCT
ajst-27534	2	26	establish	establish	VERB
ajst-27534	2	27	a	a	DET
ajst-27534	2	28	small	small	ADJ
ajst-27534	2	29	sample	sample	NOUN
ajst-27534	2	30	database	database	NOUN
ajst-27534	2	31	of	of	ADP
ajst-27534	2	32	4000	4000	NUM
ajst-27534	2	33	citrus	citrus	NOUN
ajst-27534	2	34	fruits	fruit	NOUN
ajst-27534	2	35	in	in	ADP
ajst-27534	2	36	four	four	NUM
ajst-27534	2	37	categories	category	NOUN
ajst-27534	2	38	and	and	CCONJ
ajst-27534	2	39	use	use	VERB
ajst-27534	2	40	this	this	DET
ajst-27534	2	41	network	network	NOUN
ajst-27534	2	42	model	model	NOUN
ajst-27534	2	43	for	for	ADP
ajst-27534	2	44	classification	classification	NOUN
ajst-27534	2	45	detection	detection	NOUN
ajst-27534	2	46	.	.	PUNCT
ajst-27534	3	1	the	the	DET
ajst-27534	3	2	average	average	ADJ
ajst-27534	3	3	accuracy	accuracy	NOUN
ajst-27534	3	4	of	of	ADP
ajst-27534	3	5	the	the	DET
ajst-27534	3	6	final	final	ADJ
ajst-27534	3	7	classification	classification	NOUN
ajst-27534	3	8	detection	detection	NOUN
ajst-27534	3	9	of	of	ADP
ajst-27534	3	10	resnet18	resnet18	NOUN
ajst-27534	3	11	after	after	ADP
ajst-27534	3	12	optimization	optimization	NOUN
ajst-27534	3	13	reached	reach	VERB
ajst-27534	3	14	93.764	93.764	NUM
ajst-27534	3	15	%	%	NOUN
ajst-27534	3	16	,	,	PUNCT
ajst-27534	3	17	the	the	DET
ajst-27534	3	18	training	training	NOUN
ajst-27534	3	19	time	time	NOUN
ajst-27534	3	20	of	of	ADP
ajst-27534	3	21	the	the	DET
ajst-27534	3	22	model	model	NOUN
ajst-27534	3	23	was	be	AUX
ajst-27534	3	24	reduced	reduce	VERB
ajst-27534	3	25	by	by	ADP
ajst-27534	3	26	41.47	41.47	NUM
ajst-27534	3	27	%	%	NOUN
ajst-27534	3	28	,	,	PUNCT
ajst-27534	3	29	and	and	CCONJ
ajst-27534	3	30	the	the	DET
ajst-27534	3	31	improved	improved	ADJ
ajst-27534	3	32	model	model	NOUN
ajst-27534	3	33	showed	show	VERB
ajst-27534	3	34	improvements	improvement	NOUN
ajst-27534	3	35	in	in	ADP
ajst-27534	3	36	both	both	DET
ajst-27534	3	37	training	train	VERB
ajst-27534	3	38	speed	speed	NOUN
ajst-27534	3	39	and	and	CCONJ
ajst-27534	3	40	accuracy	accuracy	NOUN
ajst-27534	3	41	.	.	PUNCT
ajst-27534	4	1	keywords	keyword	NOUN
ajst-27534	4	2	:	:	PUNCT
ajst-27534	4	3	resnet	resnet	NOUN
ajst-27534	4	4	;	;	PUNCT
ajst-27534	4	5	classification	classification	NOUN
ajst-27534	4	6	detection	detection	NOUN
ajst-27534	4	7	.	.	PUNCT
ajst-27534	5	1	1	1	X
ajst-27534	5	2	.	.	X
ajst-27534	5	3	introduction	introduction	NOUN
ajst-27534	5	4	citrus	citrus	NOUN
ajst-27534	5	5	,	,	PUNCT
ajst-27534	5	6	as	as	ADP
ajst-27534	5	7	one	one	NUM
ajst-27534	5	8	of	of	ADP
ajst-27534	5	9	china	china	PROPN
ajst-27534	5	10	's	's	PART
ajst-27534	5	11	major	major	ADJ
ajst-27534	5	12	economic	economic	ADJ
ajst-27534	5	13	crops	crop	NOUN
ajst-27534	5	14	and	and	CCONJ
ajst-27534	5	15	export	export	NOUN
ajst-27534	5	16	fruits	fruit	NOUN
ajst-27534	5	17	,	,	PUNCT
ajst-27534	5	18	surpassed	surpass	VERB
ajst-27534	5	19	apples	apple	NOUN
ajst-27534	5	20	for	for	ADP
ajst-27534	5	21	the	the	DET
ajst-27534	5	22	first	first	ADJ
ajst-27534	5	23	time	time	NOUN
ajst-27534	5	24	to	to	PART
ajst-27534	5	25	become	become	VERB
ajst-27534	5	26	the	the	DET
ajst-27534	5	27	largest	large	ADJ
ajst-27534	5	28	fruit	fruit	NOUN
ajst-27534	5	29	in	in	ADP
ajst-27534	5	30	china	china	PROPN
ajst-27534	5	31	in	in	ADP
ajst-27534	5	32	2018	2018	NUM
ajst-27534	5	33	.	.	PUNCT
ajst-27534	6	1	in	in	ADP
ajst-27534	6	2	2019	2019	NUM
ajst-27534	6	3	,	,	PUNCT
ajst-27534	6	4	the	the	DET
ajst-27534	6	5	citrus	citrus	NOUN
ajst-27534	6	6	area	area	NOUN
ajst-27534	6	7	in	in	ADP
ajst-27534	6	8	china	china	PROPN
ajst-27534	6	9	was	be	AUX
ajst-27534	6	10	41.72	41.72	NUM
ajst-27534	6	11	million	million	NUM
ajst-27534	6	12	mu	mu	NOUN
ajst-27534	6	13	,	,	PUNCT
ajst-27534	6	14	with	with	ADP
ajst-27534	6	15	a	a	DET
ajst-27534	6	16	yield	yield	NOUN
ajst-27534	6	17	of	of	ADP
ajst-27534	6	18	45.84	45.84	NUM
ajst-27534	6	19	million	million	NUM
ajst-27534	6	20	tons	ton	NOUN
ajst-27534	6	21	,	,	PUNCT
ajst-27534	6	22	both	both	CCONJ
ajst-27534	6	23	in	in	ADP
ajst-27534	6	24	terms	term	NOUN
ajst-27534	6	25	of	of	ADP
ajst-27534	6	26	area	area	NOUN
ajst-27534	6	27	and	and	CCONJ
ajst-27534	6	28	quantity	quantity	NOUN
ajst-27534	6	29	,	,	PUNCT
ajst-27534	6	30	it	it	PRON
ajst-27534	6	31	ranks	rank	VERB
ajst-27534	6	32	first	first	ADV
ajst-27534	6	33	in	in	ADP
ajst-27534	6	34	the	the	DET
ajst-27534	6	35	world	world	NOUN
ajst-27534	6	36	,	,	PUNCT
ajst-27534	6	37	and	and	CCONJ
ajst-27534	6	38	chinese	chinese	ADJ
ajst-27534	6	39	citrus	citrus	NOUN
ajst-27534	6	40	accounts	account	VERB
ajst-27534	6	41	for	for	ADP
ajst-27534	6	42	nearly	nearly	ADV
ajst-27534	6	43	one	one	NUM
ajst-27534	6	44	-	-	PUNCT
ajst-27534	6	45	third	third	NOUN
ajst-27534	6	46	of	of	ADP
ajst-27534	6	47	the	the	DET
ajst-27534	6	48	world	world	NOUN
ajst-27534	6	49	's	's	PART
ajst-27534	6	50	citrus	citrus	NOUN
ajst-27534	6	51	[	[	X
ajst-27534	6	52	1	1	NUM
ajst-27534	6	53	]	]	PUNCT
ajst-27534	6	54	.	.	PUNCT
ajst-27534	7	1	therefore	therefore	ADV
ajst-27534	7	2	,	,	PUNCT
ajst-27534	7	3	achieving	achieve	VERB
ajst-27534	7	4	classification	classification	NOUN
ajst-27534	7	5	and	and	CCONJ
ajst-27534	7	6	detection	detection	NOUN
ajst-27534	7	7	of	of	ADP
ajst-27534	7	8	citrus	citrus	NOUN
ajst-27534	7	9	fruits	fruit	NOUN
ajst-27534	7	10	is	be	AUX
ajst-27534	7	11	of	of	ADP
ajst-27534	7	12	great	great	ADJ
ajst-27534	7	13	significance	significance	NOUN
ajst-27534	7	14	.	.	PUNCT
ajst-27534	8	1	shi	shi	PROPN
ajst-27534	8	2	yu	yu	PROPN
ajst-27534	8	3	et	et	PROPN
ajst-27534	8	4	al	al	PROPN
ajst-27534	8	5	.	.	PROPN
ajst-27534	8	6	proposed	propose	VERB
ajst-27534	8	7	the	the	DET
ajst-27534	8	8	e	e	NOUN
ajst-27534	8	9	-	-	ADJ
ajst-27534	8	10	res2net	res2net	ADJ
ajst-27534	8	11	model	model	NOUN
ajst-27534	8	12	for	for	ADP
ajst-27534	8	13	the	the	DET
ajst-27534	8	14	classification	classification	NOUN
ajst-27534	8	15	of	of	ADP
ajst-27534	8	16	citrus	citrus	NOUN
ajst-27534	8	17	leaves	leave	NOUN
ajst-27534	8	18	,	,	PUNCT
ajst-27534	8	19	by	by	ADP
ajst-27534	8	20	combining	combine	VERB
ajst-27534	8	21	the	the	DET
ajst-27534	8	22	se	se	PROPN
ajst-27534	8	23	attention	attention	NOUN
ajst-27534	8	24	mechanism	mechanism	NOUN
ajst-27534	8	25	with	with	ADP
ajst-27534	8	26	the	the	DET
ajst-27534	8	27	res2net	res2net	NOUN
ajst-27534	8	28	proposed	propose	VERB
ajst-27534	8	29	by	by	ADP
ajst-27534	8	30	gaos	gaos	PROPN
ajst-27534	8	31	et	et	PROPN
ajst-27534	8	32	al	al	PROPN
ajst-27534	8	33	.	.	PROPN
ajst-27534	8	34	based	base	VERB
ajst-27534	8	35	on	on	ADP
ajst-27534	8	36	res	re	NOUN
ajst-27534	8	37	net	net	NOUN
ajst-27534	8	38	,	,	PUNCT
ajst-27534	8	39	the	the	DET
ajst-27534	8	40	model	model	NOUN
ajst-27534	8	41	quality	quality	NOUN
ajst-27534	8	42	was	be	AUX
ajst-27534	8	43	improved	improve	VERB
ajst-27534	8	44	,	,	PUNCT
ajst-27534	8	45	and	and	CCONJ
ajst-27534	8	46	the	the	DET
ajst-27534	8	47	final	final	ADJ
ajst-27534	8	48	classification	classification	NOUN
ajst-27534	8	49	accuracy	accuracy	NOUN
ajst-27534	8	50	reached	reach	VERB
ajst-27534	8	51	86.04	86.04	NUM
ajst-27534	8	52	%	%	NOUN
ajst-27534	8	53	.	.	PUNCT
ajst-27534	9	1	yue	yue	PROPN
ajst-27534	9	2	yang	yang	PROPN
ajst-27534	9	3	,	,	PUNCT
ajst-27534	9	4	zhang	zhang	PROPN
ajst-27534	9	5	wei	wei	PROPN
ajst-27534	9	6	and	and	CCONJ
ajst-27534	9	7	others	other	NOUN
ajst-27534	9	8	designed	design	VERB
ajst-27534	9	9	an	an	DET
ajst-27534	9	10	adaptive	adaptive	ADJ
ajst-27534	9	11	vgg16	vgg16	NOUN
ajst-27534	9	12	network	network	NOUN
ajst-27534	9	13	(	(	PUNCT
ajst-27534	9	14	advgg16	advgg16	NOUN
ajst-27534	9	15	)	)	PUNCT
ajst-27534	9	16	based	base	VERB
ajst-27534	9	17	on	on	ADP
ajst-27534	9	18	the	the	DET
ajst-27534	9	19	breast	breast	NOUN
ajst-27534	9	20	cancer	cancer	NOUN
ajst-27534	9	21	risk	risk	NOUN
ajst-27534	9	22	prediction	prediction	NOUN
ajst-27534	9	23	problem	problem	NOUN
ajst-27534	9	24	and	and	CCONJ
ajst-27534	9	25	used	use	VERB
ajst-27534	9	26	the	the	DET
ajst-27534	9	27	cross	cross	NOUN
ajst-27534	9	28	validation	validation	NOUN
ajst-27534	9	29	mechanism	mechanism	NOUN
ajst-27534	9	30	to	to	PART
ajst-27534	9	31	predict	predict	VERB
ajst-27534	9	32	the	the	DET
ajst-27534	9	33	risk	risk	NOUN
ajst-27534	9	34	of	of	ADP
ajst-27534	9	35	breast	breast	NOUN
ajst-27534	9	36	cancer	cancer	NOUN
ajst-27534	9	37	data	datum	NOUN
ajst-27534	9	38	set	set	VERB
ajst-27534	9	39	[	[	PUNCT
ajst-27534	9	40	2	2	NUM
ajst-27534	9	41	-	-	SYM
ajst-27534	9	42	5	5	NUM
ajst-27534	9	43	]	]	PUNCT
ajst-27534	9	44	.	.	PUNCT
ajst-27534	10	1	jiang	jiang	PROPN
ajst-27534	10	2	jinhui	jinhui	PROPN
ajst-27534	10	3	adopted	adopt	VERB
ajst-27534	10	4	the	the	DET
ajst-27534	10	5	shushang2	shushang2	NOUN
ajst-27534	10	6	citrus	citrus	NOUN
ajst-27534	10	7	method	method	NOUN
ajst-27534	10	8	that	that	PRON
ajst-27534	10	9	combines	combine	VERB
ajst-27534	10	10	sam	sam	PROPN
ajst-27534	10	11	and	and	CCONJ
ajst-27534	10	12	vgg16	vgg16	PROPN
ajst-27534	10	13	,	,	PUNCT
ajst-27534	10	14	which	which	PRON
ajst-27534	10	15	effectively	effectively	ADV
ajst-27534	10	16	saves	save	VERB
ajst-27534	10	17	time	time	NOUN
ajst-27534	10	18	on	on	ADP
ajst-27534	10	19	dataset	dataset	ADJ
ajst-27534	10	20	creation	creation	NOUN
ajst-27534	10	21	and	and	CCONJ
ajst-27534	10	22	model	model	NOUN
ajst-27534	10	23	training	training	NOUN
ajst-27534	10	24	[	[	X
ajst-27534	10	25	6	6	NUM
ajst-27534	10	26	]	]	PUNCT
ajst-27534	10	27	.	.	PUNCT
ajst-27534	11	1	however	however	ADV
ajst-27534	11	2	,	,	PUNCT
ajst-27534	11	3	the	the	DET
ajst-27534	11	4	neural	neural	ADJ
ajst-27534	11	5	network	network	NOUN
ajst-27534	11	6	models	model	NOUN
ajst-27534	11	7	used	use	VERB
ajst-27534	11	8	in	in	ADP
ajst-27534	11	9	the	the	DET
ajst-27534	11	10	appeals	appeal	NOUN
ajst-27534	11	11	are	be	AUX
ajst-27534	11	12	relatively	relatively	ADV
ajst-27534	11	13	complex	complex	ADJ
ajst-27534	11	14	,	,	PUNCT
ajst-27534	11	15	due	due	ADP
ajst-27534	11	16	to	to	ADP
ajst-27534	11	17	the	the	DET
ajst-27534	11	18	large	large	ADJ
ajst-27534	11	19	number	number	NOUN
ajst-27534	11	20	of	of	ADP
ajst-27534	11	21	parameters	parameter	NOUN
ajst-27534	11	22	,	,	PUNCT
ajst-27534	11	23	this	this	DET
ajst-27534	11	24	article	article	NOUN
ajst-27534	11	25	only	only	ADV
ajst-27534	11	26	improves	improve	VERB
ajst-27534	11	27	the	the	DET
ajst-27534	11	28	residual	residual	ADJ
ajst-27534	11	29	block	block	NOUN
ajst-27534	11	30	structure	structure	NOUN
ajst-27534	11	31	of	of	ADP
ajst-27534	11	32	resnet	resnet	NOUN
ajst-27534	11	33	,	,	PUNCT
ajst-27534	11	34	performing	perform	VERB
ajst-27534	11	35	certain	certain	ADJ
ajst-27534	11	36	optimizations	optimization	NOUN
ajst-27534	11	37	on	on	ADP
ajst-27534	11	38	residual	residual	ADJ
ajst-27534	11	39	blocks	block	NOUN
ajst-27534	11	40	that	that	PRON
ajst-27534	11	41	have	have	AUX
ajst-27534	11	42	been	be	AUX
ajst-27534	11	43	pre	pre	AUX
ajst-27534	11	44	activated	activate	VERB
ajst-27534	11	45	,	,	PUNCT
ajst-27534	11	46	the	the	DET
ajst-27534	11	47	accuracy	accuracy	NOUN
ajst-27534	11	48	of	of	ADP
ajst-27534	11	49	citrus	citrus	NOUN
ajst-27534	11	50	classification	classification	NOUN
ajst-27534	11	51	is	be	AUX
ajst-27534	11	52	significantly	significantly	ADV
ajst-27534	11	53	improved	improve	VERB
ajst-27534	11	54	while	while	SCONJ
ajst-27534	11	55	keeping	keep	VERB
ajst-27534	11	56	the	the	DET
ajst-27534	11	57	number	number	NOUN
ajst-27534	11	58	of	of	ADP
ajst-27534	11	59	network	network	NOUN
ajst-27534	11	60	parameters	parameter	NOUN
ajst-27534	11	61	basically	basically	ADV
ajst-27534	11	62	unchanged	unchanged	ADJ
ajst-27534	11	63	.	.	PUNCT
ajst-27534	12	1	2	2	X
ajst-27534	12	2	.	.	NUM
ajst-27534	12	3	optimized	optimize	VERB
ajst-27534	12	4	resnet	resnet	NOUN
ajst-27534	12	5	network	network	NOUN
ajst-27534	12	6	model	model	NOUN
ajst-27534	12	7	traditional	traditional	ADJ
ajst-27534	12	8	convolutional	convolutional	ADJ
ajst-27534	12	9	neural	neural	ADJ
ajst-27534	12	10	networks	network	NOUN
ajst-27534	12	11	such	such	ADJ
ajst-27534	12	12	as	as	ADP
ajst-27534	12	13	lenet	lenet	NOUN
ajst-27534	12	14	,	,	PUNCT
ajst-27534	12	15	alexnet	alexnet	ADJ
ajst-27534	12	16	,	,	PUNCT
ajst-27534	12	17	vgg	vgg	NOUN
ajst-27534	12	18	,	,	PUNCT
ajst-27534	12	19	and	and	CCONJ
ajst-27534	12	20	googlenet	googlenet	NOUN
ajst-27534	12	21	all	all	PRON
ajst-27534	12	22	suffer	suffer	VERB
ajst-27534	12	23	from	from	ADP
ajst-27534	12	24	degradation	degradation	NOUN
ajst-27534	12	25	as	as	SCONJ
ajst-27534	12	26	the	the	DET
ajst-27534	12	27	number	number	NOUN
ajst-27534	12	28	of	of	ADP
ajst-27534	12	29	layers	layer	NOUN
ajst-27534	12	30	increases	increase	VERB
ajst-27534	12	31	after	after	ADP
ajst-27534	12	32	reaching	reach	VERB
ajst-27534	12	33	a	a	DET
ajst-27534	12	34	certain	certain	ADJ
ajst-27534	12	35	level	level	NOUN
ajst-27534	12	36	.	.	PUNCT
ajst-27534	13	1	but	but	CCONJ
ajst-27534	13	2	with	with	ADP
ajst-27534	13	3	the	the	DET
ajst-27534	13	4	release	release	NOUN
ajst-27534	13	5	of	of	ADP
ajst-27534	13	6	resnet	resnet	NOUN
ajst-27534	13	7	in	in	ADP
ajst-27534	13	8	2015	2015	NUM
ajst-27534	13	9	,	,	PUNCT
ajst-27534	13	10	resnet	resnet	NOUN
ajst-27534	13	11	was	be	AUX
ajst-27534	13	12	the	the	DET
ajst-27534	13	13	first	first	ADJ
ajst-27534	13	14	to	to	PART
ajst-27534	13	15	propose	propose	VERB
ajst-27534	13	16	the	the	DET
ajst-27534	13	17	concept	concept	NOUN
ajst-27534	13	18	of	of	ADP
ajst-27534	13	19	residual	residual	ADJ
ajst-27534	13	20	connections	connection	NOUN
ajst-27534	13	21	,	,	PUNCT
ajst-27534	13	22	allow	allow	VERB
ajst-27534	13	23	data	datum	NOUN
ajst-27534	13	24	to	to	PART
ajst-27534	13	25	enter	enter	VERB
ajst-27534	13	26	certain	certain	ADJ
ajst-27534	13	27	network	network	NOUN
ajst-27534	13	28	levels	level	NOUN
ajst-27534	13	29	and	and	CCONJ
ajst-27534	13	30	directly	directly	ADV
ajst-27534	13	31	input	input	VERB
ajst-27534	13	32	to	to	ADP
ajst-27534	13	33	deeper	deep	ADJ
ajst-27534	13	34	network	network	NOUN
ajst-27534	13	35	levels	level	NOUN
ajst-27534	13	36	.	.	PUNCT
ajst-27534	14	1	its	its	PRON
ajst-27534	14	2	topology	topology	NOUN
ajst-27534	14	3	structure	structure	NOUN
ajst-27534	14	4	can	can	AUX
ajst-27534	14	5	efficiently	efficiently	ADV
ajst-27534	14	6	train	train	VERB
ajst-27534	14	7	very	very	ADV
ajst-27534	14	8	deep	deep	ADJ
ajst-27534	14	9	neural	neural	ADJ
ajst-27534	14	10	networks	network	NOUN
ajst-27534	14	11	[	[	X
ajst-27534	14	12	7].the	7].the	DET
ajst-27534	14	13	resnet18	resnet18	NOUN
ajst-27534	14	14	network	network	NOUN
ajst-27534	14	15	structure	structure	NOUN
ajst-27534	14	16	is	be	AUX
ajst-27534	14	17	as	as	SCONJ
ajst-27534	14	18	follows	follow	VERB
ajst-27534	14	19	:	:	PUNCT
ajst-27534	14	20	resnet18	resnet18	NOUN
ajst-27534	14	21	has	have	VERB
ajst-27534	14	22	a	a	DET
ajst-27534	14	23	total	total	NOUN
ajst-27534	14	24	of	of	ADP
ajst-27534	14	25	18	18	NUM
ajst-27534	14	26	layers	layer	NOUN
ajst-27534	14	27	,	,	PUNCT
ajst-27534	14	28	with	with	ADP
ajst-27534	14	29	each	each	DET
ajst-27534	14	30	convolutional	convolutional	ADJ
ajst-27534	14	31	block	block	NOUN
ajst-27534	14	32	followed	follow	VERB
ajst-27534	14	33	by	by	ADP
ajst-27534	14	34	a	a	DET
ajst-27534	14	35	batch	batch	NOUN
ajst-27534	14	36	normalization	normalization	NOUN
ajst-27534	14	37	and	and	CCONJ
ajst-27534	14	38	a	a	DET
ajst-27534	14	39	relu	relu	NOUN
ajst-27534	14	40	activation	activation	NOUN
ajst-27534	14	41	function	function	NOUN
ajst-27534	14	42	.	.	PUNCT
ajst-27534	15	1	finally	finally	ADV
ajst-27534	15	2	,	,	PUNCT
ajst-27534	15	3	citrus	citrus	ADJ
ajst-27534	15	4	classification	classification	NOUN
ajst-27534	15	5	was	be	AUX
ajst-27534	15	6	performed	perform	VERB
ajst-27534	15	7	using	use	VERB
ajst-27534	15	8	the	the	DET
ajst-27534	15	9	softmax	softmax	NOUN
ajst-27534	15	10	function	function	NOUN
ajst-27534	15	11	after	after	ADP
ajst-27534	15	12	global	global	ADJ
ajst-27534	15	13	average	average	ADJ
ajst-27534	15	14	pooling	pooling	NOUN
ajst-27534	15	15	and	and	CCONJ
ajst-27534	15	16	input	input	NOUN
ajst-27534	15	17	citrus	citrus	NOUN
ajst-27534	15	18	graphics	graphic	NOUN
ajst-27534	15	19	7ⅹ7	7ⅹ7	NUM
ajst-27534	15	20	conv1	conv1	NOUN
ajst-27534	15	21	/	/	SYM
ajst-27534	15	22	stride	stride	NOUN
ajst-27534	15	23	2	2	NUM
ajst-27534	15	24	max	max	NOUN
ajst-27534	15	25	pooling	pool	VERB
ajst-27534	15	26	3ⅹ3/	3ⅹ3/	NUM
ajst-27534	15	27	stride	stride	NOUN
ajst-27534	15	28	2	2	NUM
ajst-27534	15	29	[	[	X
ajst-27534	15	30	3ⅹ3	3ⅹ3	NUM
ajst-27534	15	31	,	,	PUNCT
ajst-27534	15	32	64	64	NUM
ajst-27534	15	33	ⅹ2	ⅹ2	NOUN
ajst-27534	15	34	,	,	PUNCT
ajst-27534	15	35	conv2	conv2	NOUN
ajst-27534	15	36	3ⅹ3	3ⅹ3	NUM
ajst-27534	15	37	,	,	PUNCT
ajst-27534	15	38	64	64	NUM
ajst-27534	15	39	]	]	PUNCT
ajst-27534	16	1	[	[	X
ajst-27534	16	2	3ⅹ3	3ⅹ3	NUM
ajst-27534	16	3	,	,	PUNCT
ajst-27534	16	4	128	128	NUM
ajst-27534	16	5	ⅹ2	ⅹ2	NOUN
ajst-27534	16	6	,	,	PUNCT
ajst-27534	16	7	conv3	conv3	NOUN
ajst-27534	16	8	3ⅹ3	3ⅹ3	NUM
ajst-27534	16	9	,	,	PUNCT
ajst-27534	16	10	128	128	NUM
ajst-27534	16	11	]	]	PUNCT
ajst-27534	17	1	[	[	X
ajst-27534	17	2	3ⅹ3	3ⅹ3	NUM
ajst-27534	17	3	,	,	PUNCT
ajst-27534	17	4	512	512	NUM
ajst-27534	17	5	ⅹ2	ⅹ2	NOUN
ajst-27534	17	6	,	,	PUNCT
ajst-27534	17	7	conv5	conv5	NOUN
ajst-27534	17	8	3ⅹ3	3ⅹ3	NUM
ajst-27534	17	9	,	,	PUNCT
ajst-27534	17	10	512	512	NUM
ajst-27534	17	11	]	]	PUNCT
ajst-27534	17	12	average	average	ADJ
ajst-27534	17	13	pool	pool	NOUN
ajst-27534	17	14	1000	1000	NUM
ajst-27534	17	15	-	-	PUNCT
ajst-27534	17	16	dfc	dfc	NOUN
ajst-27534	17	17	202	202	NUM
ajst-27534	17	18	flattening	flattening	NOUN
ajst-27534	17	19	.	.	PUNCT
ajst-27534	18	1	2.1	2.1	NUM
ajst-27534	18	2	.	.	PUNCT
ajst-27534	18	3	improvement	improvement	NOUN
ajst-27534	18	4	of	of	ADP
ajst-27534	18	5	incomplete	incomplete	ADJ
ajst-27534	18	6	blocks	block	NOUN
ajst-27534	18	7	in	in	ADP
ajst-27534	18	8	2022	2022	NUM
ajst-27534	18	9	,	,	PUNCT
ajst-27534	18	10	xia	xia	PROPN
ajst-27534	18	11	minggui	minggui	PROPN
ajst-27534	18	12	,	,	PUNCT
ajst-27534	18	13	tian	tian	ADJ
ajst-27534	18	14	ruijun	ruijun	NOUN
ajst-27534	18	15	,	,	PUNCT
ajst-27534	18	16	and	and	CCONJ
ajst-27534	18	17	others	other	NOUN
ajst-27534	18	18	used	use	VERB
ajst-27534	18	19	residual	residual	ADJ
ajst-27534	18	20	units	unit	NOUN
ajst-27534	18	21	with	with	ADP
ajst-27534	18	22	pre	pre	VERB
ajst-27534	18	23	activated	activate	VERB
ajst-27534	18	24	structures	structure	NOUN
ajst-27534	18	25	to	to	PART
ajst-27534	18	26	solve	solve	VERB
ajst-27534	18	27	the	the	DET
ajst-27534	18	28	potential	potential	ADJ
ajst-27534	18	29	problems	problem	NOUN
ajst-27534	18	30	of	of	ADP
ajst-27534	18	31	gradient	gradient	NOUN
ajst-27534	18	32	vanishing	vanishing	NOUN
ajst-27534	18	33	and	and	CCONJ
ajst-27534	18	34	exploding	explode	VERB
ajst-27534	18	35	in	in	ADP
ajst-27534	18	36	the	the	DET
ajst-27534	18	37	original	original	ADJ
ajst-27534	18	38	residual	residual	ADJ
ajst-27534	18	39	units	unit	NOUN
ajst-27534	18	40	,	,	PUNCT
ajst-27534	18	41	and	and	CCONJ
ajst-27534	18	42	ultimately	ultimately	ADV
ajst-27534	18	43	achieved	achieve	VERB
ajst-27534	18	44	good	good	ADJ
ajst-27534	18	45	results	result	NOUN
ajst-27534	18	46	.	.	PUNCT
ajst-27534	19	1	its	its	PRON
ajst-27534	19	2	structure	structure	NOUN
ajst-27534	19	3	is	be	AUX
ajst-27534	19	4	as	as	SCONJ
ajst-27534	19	5	follows	follow	VERB
ajst-27534	19	6	:	:	PUNCT
ajst-27534	19	7	but	but	CCONJ
ajst-27534	19	8	we	we	PRON
ajst-27534	19	9	believe	believe	VERB
ajst-27534	19	10	that	that	SCONJ
ajst-27534	19	11	batch	batch	NOUN
ajst-27534	19	12	normalization	normalization	NOUN
ajst-27534	19	13	should	should	AUX
ajst-27534	19	14	still	still	ADV
ajst-27534	19	15	be	be	AUX
ajst-27534	19	16	performed	perform	VERB
ajst-27534	19	17	after	after	ADP
ajst-27534	19	18	the	the	DET
ajst-27534	19	19	second	second	ADJ
ajst-27534	19	20	convolution	convolution	NOUN
ajst-27534	19	21	,	,	PUNCT
ajst-27534	19	22	because	because	SCONJ
ajst-27534	19	23	the	the	DET
ajst-27534	19	24	next	next	ADJ
ajst-27534	19	25	stage	stage	NOUN
ajst-27534	19	26	is	be	AUX
ajst-27534	19	27	to	to	PART
ajst-27534	19	28	add	add	VERB
ajst-27534	19	29	the	the	DET
ajst-27534	19	30	original	original	ADJ
ajst-27534	19	31	input	input	NOUN
ajst-27534	19	32	image	image	NOUN
ajst-27534	19	33	to	to	ADP
ajst-27534	19	34	the	the	DET
ajst-27534	19	35	computed	computed	ADJ
ajst-27534	19	36	result	result	NOUN
ajst-27534	19	37	,	,	PUNCT
ajst-27534	19	38	and	and	CCONJ
ajst-27534	19	39	previous	previous	ADJ
ajst-27534	19	40	networks	network	NOUN
ajst-27534	19	41	have	have	AUX
ajst-27534	19	42	used	use	VERB
ajst-27534	19	43	batch	batch	NOUN
ajst-27534	19	44	normalization	normalization	NOUN
ajst-27534	19	45	after	after	ADP
ajst-27534	19	46	convolution	convolution	NOUN
ajst-27534	19	47	,	,	PUNCT
ajst-27534	19	48	so	so	CCONJ
ajst-27534	19	49	the	the	DET
ajst-27534	19	50	original	original	ADJ
ajst-27534	19	51	image	image	NOUN
ajst-27534	19	52	entering	enter	VERB
ajst-27534	19	53	the	the	DET
ajst-27534	19	54	residual	residual	ADJ
ajst-27534	19	55	block	block	NOUN
ajst-27534	19	56	must	must	AUX
ajst-27534	19	57	also	also	ADV
ajst-27534	19	58	be	be	AUX
ajst-27534	19	59	the	the	DET
ajst-27534	19	60	result	result	NOUN
ajst-27534	19	61	of	of	ADP
ajst-27534	19	62	batch	batch	NOUN
ajst-27534	19	63	normalization	normalization	NOUN
ajst-27534	19	64	.	.	PUNCT
ajst-27534	20	1	performing	perform	VERB
ajst-27534	20	2	batch	batch	NOUN
ajst-27534	20	3	normalization	normalization	NOUN
ajst-27534	20	4	after	after	SCONJ
ajst-27534	20	5	the	the	DET
ajst-27534	20	6	second	second	ADJ
ajst-27534	20	7	convolution	convolution	NOUN
ajst-27534	20	8	will	will	AUX
ajst-27534	20	9	make	make	VERB
ajst-27534	20	10	their	their	PRON
ajst-27534	20	11	data	datum	NOUN
ajst-27534	20	12	distributions	distribution	NOUN
ajst-27534	20	13	the	the	DET
ajst-27534	20	14	same	same	ADJ
ajst-27534	20	15	,	,	PUNCT
ajst-27534	20	16	and	and	CCONJ
ajst-27534	20	17	adding	add	VERB
ajst-27534	20	18	them	they	PRON
ajst-27534	20	19	together	together	ADV
ajst-27534	20	20	can	can	AUX
ajst-27534	20	21	better	well	ADV
ajst-27534	20	22	highlight	highlight	VERB
ajst-27534	20	23	the	the	DET
ajst-27534	20	24	role	role	NOUN
ajst-27534	20	25	of	of	ADP
ajst-27534	20	26	residual	residual	ADJ
ajst-27534	20	27	blocks	block	NOUN
ajst-27534	20	28	.	.	PUNCT
ajst-27534	21	1	then	then	ADV
ajst-27534	21	2	,	,	PUNCT
ajst-27534	21	3	applying	apply	VERB
ajst-27534	21	4	activation	activation	NOUN
ajst-27534	21	5	functions	function	NOUN
ajst-27534	21	6	to	to	ADP
ajst-27534	21	7	the	the	DET
ajst-27534	21	8	added	add	VERB
ajst-27534	21	9	results	result	NOUN
ajst-27534	21	10	can	can	AUX
ajst-27534	21	11	improve	improve	VERB
ajst-27534	21	12	the	the	DET
ajst-27534	21	13	network	network	NOUN
ajst-27534	21	14	's	's	PART
ajst-27534	21	15	expressive	expressive	ADJ
ajst-27534	21	16	power	power	NOUN
ajst-27534	21	17	.	.	PUNCT
ajst-27534	22	1	so	so	ADV
ajst-27534	22	2	we	we	PRON
ajst-27534	22	3	propose	propose	VERB
ajst-27534	22	4	the	the	DET
ajst-27534	22	5	following	follow	VERB
ajst-27534	22	6	novel	novel	ADJ
ajst-27534	22	7	residual	residual	ADJ
ajst-27534	22	8	block	block	NOUN
ajst-27534	22	9	structure	structure	NOUN
ajst-27534	22	10	:	:	PUNCT
ajst-27534	22	11	2.2	2.2	NUM
ajst-27534	22	12	.	.	PUNCT
ajst-27534	23	1	comparison	comparison	NOUN
ajst-27534	23	2	of	of	ADP
ajst-27534	23	3	residual	residual	ADJ
ajst-27534	23	4	block	block	NOUN
ajst-27534	23	5	improvement	improvement	NOUN
ajst-27534	23	6	experiments	experiment	NOUN
ajst-27534	23	7	using	use	VERB
ajst-27534	23	8	the	the	DET
ajst-27534	23	9	original	original	ADJ
ajst-27534	23	10	residual	residual	ADJ
ajst-27534	23	11	block	block	NOUN
ajst-27534	23	12	,	,	PUNCT
ajst-27534	23	13	introducing	introduce	VERB
ajst-27534	23	14	pre	pre	VERB
ajst-27534	23	15	activated	activate	VERB
ajst-27534	23	16	residual	residual	ADJ
ajst-27534	23	17	blocks	block	NOUN
ajst-27534	23	18	and	and	CCONJ
ajst-27534	23	19	the	the	DET
ajst-27534	23	20	new	new	ADJ
ajst-27534	23	21	residual	residual	ADJ
ajst-27534	23	22	block	block	NOUN
ajst-27534	23	23	proposed	propose	VERB
ajst-27534	23	24	in	in	ADP
ajst-27534	23	25	this	this	DET
ajst-27534	23	26	paper	paper	NOUN
ajst-27534	23	27	for	for	ADP
ajst-27534	23	28	10	10	NUM
ajst-27534	23	29	repeated	repeat	VERB
ajst-27534	23	30	experiments	experiment	NOUN
ajst-27534	23	31	,	,	PUNCT
ajst-27534	23	32	the	the	DET
ajst-27534	23	33	experimental	experimental	ADJ
ajst-27534	23	34	results	result	NOUN
ajst-27534	23	35	are	be	AUX
ajst-27534	23	36	as	as	SCONJ
ajst-27534	23	37	follows	follow	VERB
ajst-27534	23	38	:	:	PUNCT
ajst-27534	23	39	result	result	VERB
ajst-27534	23	40	analysis	analysis	NOUN
ajst-27534	23	41	:	:	PUNCT
ajst-27534	23	42	the	the	DET
ajst-27534	23	43	average	average	ADJ
ajst-27534	23	44	accuracy	accuracy	NOUN
ajst-27534	23	45	of	of	ADP
ajst-27534	23	46	10	10	NUM
ajst-27534	23	47	experiments	experiment	NOUN
ajst-27534	23	48	with	with	ADP
ajst-27534	23	49	the	the	DET
ajst-27534	23	50	introduction	introduction	NOUN
ajst-27534	23	51	of	of	ADP
ajst-27534	23	52	pre	pre	ADJ
ajst-27534	23	53	activated	activate	VERB
ajst-27534	23	54	residual	residual	ADJ
ajst-27534	23	55	structure	structure	NOUN
ajst-27534	23	56	improved	improve	VERB
ajst-27534	23	57	to	to	ADP
ajst-27534	23	58	0.93764	0.93764	NUM
ajst-27534	23	59	,	,	PUNCT
ajst-27534	23	60	while	while	SCONJ
ajst-27534	23	61	the	the	DET
ajst-27534	23	62	average	average	ADJ
ajst-27534	23	63	accuracy	accuracy	NOUN
ajst-27534	23	64	of	of	ADP
ajst-27534	23	65	10	10	NUM
ajst-27534	23	66	experiments	experiment	NOUN
ajst-27534	23	67	with	with	ADP
ajst-27534	23	68	the	the	DET
ajst-27534	23	69	introduction	introduction	NOUN
ajst-27534	23	70	of	of	ADP
ajst-27534	23	71	pre	pre	ADJ
ajst-27534	23	72	activated	activate	VERB
ajst-27534	23	73	residual	residual	ADJ
ajst-27534	23	74	structure	structure	NOUN
ajst-27534	23	75	was	be	AUX
ajst-27534	23	76	only	only	ADV
ajst-27534	23	77	0.90979	0.90979	NUM
ajst-27534	23	78	.	.	PUNCT
ajst-27534	24	1	the	the	DET
ajst-27534	24	2	average	average	ADJ
ajst-27534	24	3	accuracy	accuracy	NOUN
ajst-27534	24	4	of	of	ADP
ajst-27534	24	5	the	the	DET
ajst-27534	24	6	original	original	ADJ
ajst-27534	24	7	residual	residual	ADJ
ajst-27534	24	8	structure	structure	NOUN
ajst-27534	24	9	was	be	AUX
ajst-27534	24	10	0.91956	0.91956	NUM
ajst-27534	24	11	.	.	PUNCT
ajst-27534	25	1	the	the	DET
ajst-27534	25	2	residual	residual	ADJ
ajst-27534	25	3	re	re	NUM
ajst-27534	25	4	improved	improved	ADJ
ajst-27534	25	5	structure	structure	NOUN
ajst-27534	25	6	with	with	ADP
ajst-27534	25	7	pre	pre	VERB
ajst-27534	25	8	activated	activate	VERB
ajst-27534	25	9	structure	structure	NOUN
ajst-27534	25	10	has	have	AUX
ajst-27534	25	11	increased	increase	VERB
ajst-27534	25	12	the	the	DET
ajst-27534	25	13	accuracy	accuracy	NOUN
ajst-27534	25	14	by	by	ADP
ajst-27534	25	15	1.8	1.8	NUM
ajst-27534	25	16	percentage	percentage	NOUN
ajst-27534	25	17	points	point	NOUN
ajst-27534	25	18	compared	compare	VERB
ajst-27534	25	19	to	to	ADP
ajst-27534	25	20	the	the	DET
ajst-27534	25	21	original	original	ADJ
ajst-27534	25	22	structure	structure	NOUN
ajst-27534	25	23	,	,	PUNCT
ajst-27534	25	24	and	and	CCONJ
ajst-27534	25	25	by	by	ADP
ajst-27534	25	26	2.785	2.785	NUM
ajst-27534	25	27	percentage	percentage	NOUN
ajst-27534	25	28	points	point	NOUN
ajst-27534	25	29	compared	compare	VERB
ajst-27534	25	30	to	to	ADP
ajst-27534	25	31	the	the	DET
ajst-27534	25	32	residual	residual	ADJ
ajst-27534	25	33	structure	structure	NOUN
ajst-27534	25	34	with	with	ADP
ajst-27534	25	35	pre	pre	ADJ
ajst-27534	25	36	activated	activated	ADJ
ajst-27534	25	37	structure	structure	NOUN
ajst-27534	25	38	.	.	PUNCT
ajst-27534	26	1	the	the	DET
ajst-27534	26	2	highest	high	ADJ
ajst-27534	26	3	accuracy	accuracy	NOUN
ajst-27534	26	4	of	of	ADP
ajst-27534	26	5	the	the	DET
ajst-27534	26	6	residual	residual	ADJ
ajst-27534	26	7	re	re	ADJ
ajst-27534	26	8	improved	improved	ADJ
ajst-27534	26	9	structure	structure	NOUN
ajst-27534	26	10	with	with	ADP
ajst-27534	26	11	pre	pre	VERB
ajst-27534	26	12	activated	activated	ADJ
ajst-27534	26	13	structure	structure	NOUN
ajst-27534	26	14	can	can	AUX
ajst-27534	26	15	reach	reach	VERB
ajst-27534	26	16	0.96	0.96	NUM
ajst-27534	26	17	.	.	PUNCT
ajst-27534	27	1	the	the	DET
ajst-27534	27	2	overall	overall	ADJ
ajst-27534	27	3	performance	performance	NOUN
ajst-27534	27	4	is	be	AUX
ajst-27534	27	5	significantly	significantly	ADV
ajst-27534	27	6	improved	improve	VERB
ajst-27534	27	7	due	due	ADP
ajst-27534	27	8	to	to	ADP
ajst-27534	27	9	two	two	NUM
ajst-27534	27	10	other	other	ADJ
ajst-27534	27	11	residual	residual	ADJ
ajst-27534	27	12	structures	structure	NOUN
ajst-27534	27	13	.	.	PUNCT
ajst-27534	28	1	input	input	NOUN
ajst-27534	28	2	graph	graph	NOUN
ajst-27534	28	3	bn	bn	PROPN
ajst-27534	28	4	relu	relu	NOUN
ajst-27534	28	5	conv	conv	PROPN
ajst-27534	28	6	bn	bn	PROPN
ajst-27534	28	7	relu	relu	PROPN
ajst-27534	28	8	conv	conv	PROPN
ajst-27534	28	9	output	output	NOUN
ajst-27534	28	10	graph	graph	NOUN
ajst-27534	28	11	input	input	NOUN
ajst-27534	28	12	graph	graph	NOUN
ajst-27534	28	13	bn	bn	NUM
ajst-27534	28	14	relu	relu	NOUN
ajst-27534	28	15	conv	conv	PROPN
ajst-27534	28	16	bn	bn	PROPN
ajst-27534	28	17	conv	conv	PROPN
ajst-27534	28	18	bn	bn	PROPN
ajst-27534	28	19	relu	relu	NOUN
ajst-27534	28	20	output	output	NOUN
ajst-27534	28	21	graph	graph	NOUN
ajst-27534	28	22	203	203	NUM
ajst-27534	28	23	in	in	ADP
ajst-27534	28	24	terms	term	NOUN
ajst-27534	28	25	of	of	ADP
ajst-27534	28	26	running	run	VERB
ajst-27534	28	27	time	time	NOUN
ajst-27534	28	28	,	,	PUNCT
ajst-27534	28	29	these	these	DET
ajst-27534	28	30	three	three	NUM
ajst-27534	28	31	experiments	experiment	NOUN
ajst-27534	28	32	have	have	VERB
ajst-27534	28	33	similar	similar	ADJ
ajst-27534	28	34	running	running	NOUN
ajst-27534	28	35	times	time	NOUN
ajst-27534	28	36	,	,	PUNCT
ajst-27534	28	37	with	with	ADP
ajst-27534	28	38	an	an	DET
ajst-27534	28	39	average	average	ADJ
ajst-27534	28	40	running	running	NOUN
ajst-27534	28	41	time	time	NOUN
ajst-27534	28	42	of	of	ADP
ajst-27534	28	43	about	about	ADV
ajst-27534	28	44	14	14	NUM
ajst-27534	28	45	minutes	minute	NOUN
ajst-27534	28	46	and	and	CCONJ
ajst-27534	28	47	40	40	NUM
ajst-27534	28	48	seconds	second	NOUN
ajst-27534	28	49	for	for	ADP
ajst-27534	28	50	10	10	NUM
ajst-27534	28	51	experiments	experiment	NOUN
ajst-27534	28	52	,	,	PUNCT
ajst-27534	28	53	and	and	CCONJ
ajst-27534	28	54	the	the	DET
ajst-27534	28	55	maximum	maximum	ADJ
ajst-27534	28	56	difference	difference	NOUN
ajst-27534	28	57	in	in	ADP
ajst-27534	28	58	running	running	NOUN
ajst-27534	28	59	time	time	NOUN
ajst-27534	28	60	is	be	AUX
ajst-27534	28	61	less	less	ADJ
ajst-27534	28	62	than	than	ADP
ajst-27534	28	63	30	30	NUM
ajst-27534	28	64	seconds	second	NOUN
ajst-27534	28	65	.	.	PUNCT
ajst-27534	29	1	so	so	ADV
ajst-27534	29	2	,	,	PUNCT
ajst-27534	29	3	from	from	ADP
ajst-27534	29	4	an	an	DET
ajst-27534	29	5	overall	overall	ADJ
ajst-27534	29	6	perspective	perspective	NOUN
ajst-27534	29	7	,	,	PUNCT
ajst-27534	29	8	the	the	DET
ajst-27534	29	9	introduction	introduction	NOUN
ajst-27534	29	10	of	of	ADP
ajst-27534	29	11	a	a	DET
ajst-27534	29	12	residual	residual	ADJ
ajst-27534	29	13	improvement	improvement	NOUN
ajst-27534	29	14	structure	structure	NOUN
ajst-27534	29	15	with	with	ADP
ajst-27534	29	16	pre	pre	NOUN
ajst-27534	29	17	activation	activation	NOUN
ajst-27534	29	18	structure	structure	NOUN
ajst-27534	29	19	significantly	significantly	ADV
ajst-27534	29	20	improves	improve	VERB
ajst-27534	29	21	the	the	DET
ajst-27534	29	22	accuracy	accuracy	NOUN
ajst-27534	29	23	of	of	ADP
ajst-27534	29	24	citrus	citrus	NOUN
ajst-27534	29	25	classification	classification	NOUN
ajst-27534	29	26	while	while	SCONJ
ajst-27534	29	27	keeping	keep	VERB
ajst-27534	29	28	the	the	DET
ajst-27534	29	29	number	number	NOUN
ajst-27534	29	30	of	of	ADP
ajst-27534	29	31	network	network	NOUN
ajst-27534	29	32	parameters	parameter	NOUN
ajst-27534	29	33	basically	basically	ADV
ajst-27534	29	34	unchanged	unchanged	ADJ
ajst-27534	29	35	and	and	CCONJ
ajst-27534	29	36	the	the	DET
ajst-27534	29	37	training	training	NOUN
ajst-27534	29	38	time	time	NOUN
ajst-27534	29	39	basically	basically	ADV
ajst-27534	29	40	the	the	DET
ajst-27534	29	41	same	same	ADJ
ajst-27534	29	42	.	.	PUNCT
ajst-27534	30	1	3	3	X
ajst-27534	30	2	.	.	X
ajst-27534	30	3	experiment	experiment	NOUN
ajst-27534	30	4	3.1	3.1	NUM
ajst-27534	30	5	.	.	PUNCT
ajst-27534	30	6	data	datum	NOUN
ajst-27534	30	7	collection	collection	NOUN
ajst-27534	30	8	and	and	CCONJ
ajst-27534	30	9	classification	classification	NOUN
ajst-27534	30	10	three	three	NUM
ajst-27534	30	11	fifths	fifth	NOUN
ajst-27534	30	12	of	of	ADP
ajst-27534	30	13	the	the	DET
ajst-27534	30	14	dataset	dataset	NOUN
ajst-27534	30	15	used	use	VERB
ajst-27534	30	16	comes	come	VERB
ajst-27534	30	17	from	from	ADP
ajst-27534	30	18	kaggle	kaggle	PROPN
ajst-27534	30	19	,	,	PUNCT
ajst-27534	30	20	and	and	CCONJ
ajst-27534	30	21	two	two	NUM
ajst-27534	30	22	fifths	fifth	NOUN
ajst-27534	30	23	is	be	AUX
ajst-27534	30	24	collected	collect	VERB
ajst-27534	30	25	by	by	ADP
ajst-27534	30	26	oneself	oneself	PRON
ajst-27534	30	27	,	,	PUNCT
ajst-27534	30	28	including	include	VERB
ajst-27534	30	29	1000	1000	NUM
ajst-27534	30	30	high	high	ADJ
ajst-27534	30	31	-	-	PUNCT
ajst-27534	30	32	quality	quality	NOUN
ajst-27534	30	33	products	product	NOUN
ajst-27534	30	34	,	,	PUNCT
ajst-27534	30	35	1000	1000	NUM
ajst-27534	30	36	first	first	ADJ
ajst-27534	30	37	-	-	PUNCT
ajst-27534	30	38	class	class	NOUN
ajst-27534	30	39	products	product	NOUN
ajst-27534	30	40	,	,	PUNCT
ajst-27534	30	41	1000	1000	NUM
ajst-27534	30	42	second	second	ADJ
ajst-27534	30	43	-	-	PUNCT
ajst-27534	30	44	class	class	NOUN
ajst-27534	30	45	products	product	NOUN
ajst-27534	30	46	,	,	PUNCT
ajst-27534	30	47	and	and	CCONJ
ajst-27534	30	48	1000	1000	NUM
ajst-27534	30	49	inferior	inferior	ADJ
ajst-27534	30	50	products	product	NOUN
ajst-27534	30	51	.	.	PUNCT
ajst-27534	31	1	each	each	DET
ajst-27534	31	2	level	level	NOUN
ajst-27534	31	3	is	be	AUX
ajst-27534	31	4	divided	divide	VERB
ajst-27534	31	5	into	into	ADP
ajst-27534	31	6	a	a	DET
ajst-27534	31	7	test	test	NOUN
ajst-27534	31	8	set	set	VERB
ajst-27534	31	9	and	and	CCONJ
ajst-27534	31	10	a	a	DET
ajst-27534	31	11	training	training	NOUN
ajst-27534	31	12	set	set	NOUN
ajst-27534	31	13	according	accord	VERB
ajst-27534	31	14	to	to	ADP
ajst-27534	31	15	a	a	DET
ajst-27534	31	16	9:1	9:1	NUM
ajst-27534	31	17	ratio	ratio	NOUN
ajst-27534	31	18	.	.	PUNCT
ajst-27534	32	1	the	the	DET
ajst-27534	32	2	training	training	NOUN
ajst-27534	32	3	set	set	NOUN
ajst-27534	32	4	is	be	AUX
ajst-27534	32	5	further	far	ADV
ajst-27534	32	6	divided	divide	VERB
ajst-27534	32	7	into	into	ADP
ajst-27534	32	8	a	a	DET
ajst-27534	32	9	validation	validation	NOUN
ajst-27534	32	10	set	set	NOUN
ajst-27534	32	11	and	and	CCONJ
ajst-27534	32	12	a	a	DET
ajst-27534	32	13	training	training	NOUN
ajst-27534	32	14	set	set	VERB
ajst-27534	32	15	in	in	ADP
ajst-27534	32	16	a	a	DET
ajst-27534	32	17	ratio	ratio	NOUN
ajst-27534	32	18	of	of	ADP
ajst-27534	32	19	2:8	2:8	NUM
ajst-27534	32	20	.	.	PUNCT
ajst-27534	33	1	the	the	DET
ajst-27534	33	2	validation	validation	NOUN
ajst-27534	33	3	set	set	NOUN
ajst-27534	33	4	does	do	AUX
ajst-27534	33	5	not	not	PART
ajst-27534	33	6	participate	participate	VERB
ajst-27534	33	7	in	in	ADP
ajst-27534	33	8	backpropagation	backpropagation	NOUN
ajst-27534	33	9	,	,	PUNCT
ajst-27534	33	10	and	and	CCONJ
ajst-27534	33	11	the	the	DET
ajst-27534	33	12	test	test	NOUN
ajst-27534	33	13	set	set	NOUN
ajst-27534	33	14	does	do	AUX
ajst-27534	33	15	not	not	PART
ajst-27534	33	16	participate	participate	VERB
ajst-27534	33	17	in	in	ADP
ajst-27534	33	18	training	training	NOUN
ajst-27534	33	19	.	.	PUNCT
ajst-27534	34	1	citrus	citrus	PROPN
ajst-27534	34	2	classification	classification	NOUN
ajst-27534	34	3	:	:	PUNCT
ajst-27534	34	4	based	base	VERB
ajst-27534	34	5	on	on	ADP
ajst-27534	34	6	market	market	NOUN
ajst-27534	34	7	research	research	NOUN
ajst-27534	34	8	and	and	CCONJ
ajst-27534	34	9	personal	personal	ADJ
ajst-27534	34	10	life	life	NOUN
ajst-27534	34	11	experience	experience	NOUN
ajst-27534	34	12	,	,	PUNCT
ajst-27534	34	13	citrus	citrus	NOUN
ajst-27534	34	14	fruits	fruit	NOUN
ajst-27534	34	15	are	be	AUX
ajst-27534	34	16	mainly	mainly	ADV
ajst-27534	34	17	classified	classify	VERB
ajst-27534	34	18	into	into	ADP
ajst-27534	34	19	superior	superior	ADJ
ajst-27534	34	20	fruits	fruit	NOUN
ajst-27534	34	21	,	,	PUNCT
ajst-27534	34	22	first	first	ADJ
ajst-27534	34	23	-	-	PUNCT
ajst-27534	34	24	class	class	NOUN
ajst-27534	34	25	fruits	fruit	NOUN
ajst-27534	34	26	,	,	PUNCT
ajst-27534	34	27	second	second	ADJ
ajst-27534	34	28	-	-	PUNCT
ajst-27534	34	29	class	class	NOUN
ajst-27534	34	30	fruits	fruit	NOUN
ajst-27534	34	31	,	,	PUNCT
ajst-27534	34	32	and	and	CCONJ
ajst-27534	34	33	inferior	inferior	ADJ
ajst-27534	34	34	fruits	fruit	NOUN
ajst-27534	34	35	based	base	VERB
ajst-27534	34	36	on	on	ADP
ajst-27534	34	37	color	color	NOUN
ajst-27534	34	38	,	,	PUNCT
ajst-27534	34	39	decay	decay	NOUN
ajst-27534	34	40	,	,	PUNCT
ajst-27534	34	41	and	and	CCONJ
ajst-27534	34	42	defects	defect	NOUN
ajst-27534	34	43	(	(	PUNCT
ajst-27534	34	44	referring	refer	VERB
ajst-27534	34	45	to	to	ADP
ajst-27534	34	46	spots	spot	NOUN
ajst-27534	34	47	,	,	PUNCT
ajst-27534	34	48	damage	damage	NOUN
ajst-27534	34	49	,	,	PUNCT
ajst-27534	34	50	and	and	CCONJ
ajst-27534	34	51	dents	dent	NOUN
ajst-27534	34	52	)	)	PUNCT
ajst-27534	34	53	.	.	PUNCT
ajst-27534	35	1	citrus	citrus	NOUN
ajst-27534	35	2	fruits	fruit	NOUN
ajst-27534	35	3	with	with	ADP
ajst-27534	35	4	bright	bright	ADJ
ajst-27534	35	5	and	and	CCONJ
ajst-27534	35	6	uniform	uniform	ADJ
ajst-27534	35	7	color	color	NOUN
ajst-27534	35	8	,	,	PUNCT
ajst-27534	35	9	deep	deep	ADJ
ajst-27534	35	10	color	color	NOUN
ajst-27534	35	11	,	,	PUNCT
ajst-27534	35	12	overall	overall	ADJ
ajst-27534	35	13	orange	orange	ADJ
ajst-27534	35	14	or	or	CCONJ
ajst-27534	35	15	red	red	ADJ
ajst-27534	35	16	color	color	NOUN
ajst-27534	35	17	,	,	PUNCT
ajst-27534	35	18	and	and	CCONJ
ajst-27534	35	19	0.86	0.86	NUM
ajst-27534	35	20	0.88	0.88	NUM
ajst-27534	35	21	0.9	0.9	NUM
ajst-27534	35	22	0.92	0.92	NUM
ajst-27534	35	23	0.94	0.94	NUM
ajst-27534	35	24	0.96	0.96	NUM
ajst-27534	35	25	0.98	0.98	NUM
ajst-27534	35	26	1	1	NUM
ajst-27534	35	27	0	0	NUM
ajst-27534	35	28	2	2	NUM
ajst-27534	35	29	4	4	NUM
ajst-27534	35	30	6	6	NUM
ajst-27534	35	31	8	8	NUM
ajst-27534	35	32	10	10	NUM
ajst-27534	35	33	12	12	NUM
ajst-27534	35	34	a	a	PRON
ajst-27534	35	35	cc	cc	ADP
ajst-27534	35	36	ur	ur	INTJ
ajst-27534	35	37	ac	ac	PROPN
ajst-27534	35	38	y	y	PROPN
ajst-27534	35	39	number	number	NOUN
ajst-27534	35	40	of	of	ADP
ajst-27534	35	41	experiments	experiment	NOUN
ajst-27534	35	42	comparison	comparison	NOUN
ajst-27534	35	43	of	of	ADP
ajst-27534	35	44	accuracy	accuracy	NOUN
ajst-27534	35	45	in	in	ADP
ajst-27534	35	46	testing	test	VERB
ajst-27534	35	47	incomplete	incomplete	ADJ
ajst-27534	35	48	structures	structure	NOUN
ajst-27534	35	49	partial	partial	ADJ
ajst-27534	35	50	improvement	improvement	NOUN
ajst-27534	35	51	based	base	VERB
ajst-27534	35	52	on	on	ADP
ajst-27534	35	53	pre	pre	NOUN
ajst-27534	35	54	activation	activation	NOUN
ajst-27534	35	55	introducing	introduce	VERB
ajst-27534	35	56	pre	pre	VERB
ajst-27534	35	57	activated	activate	VERB
ajst-27534	35	58	incomplete	incomplete	ADJ
ajst-27534	35	59	improvements	improvement	NOUN
ajst-27534	35	60	original	original	ADJ
ajst-27534	35	61	incomplete	incomplete	ADJ
ajst-27534	35	62	structure	structure	NOUN
ajst-27534	35	63	800	800	NUM
ajst-27534	35	64	820	820	NUM
ajst-27534	35	65	840	840	NUM
ajst-27534	35	66	860	860	NUM
ajst-27534	35	67	880	880	NUM
ajst-27534	35	68	900	900	NUM
ajst-27534	35	69	920	920	NUM
ajst-27534	35	70	940	940	NUM
ajst-27534	35	71	1	1	NUM
ajst-27534	35	72	2	2	NUM
ajst-27534	35	73	3	3	NUM
ajst-27534	35	74	4	4	NUM
ajst-27534	35	75	5	5	NUM
ajst-27534	35	76	6	6	NUM
ajst-27534	35	77	7	7	NUM
ajst-27534	35	78	8	8	NUM
ajst-27534	35	79	9	9	NUM
ajst-27534	35	80	10	10	NUM
ajst-27534	35	81	r	r	NOUN
ajst-27534	35	82	un	un	PROPN
ajst-27534	35	83	ni	ni	PROPN
ajst-27534	35	84	ng	ng	PROPN
ajst-27534	35	85	ti	ti	PROPN
ajst-27534	35	86	m	m	VERB
ajst-27534	35	87	e/	e/	NOUN
ajst-27534	35	88	s	s	NOUN
ajst-27534	35	89	number	number	NOUN
ajst-27534	35	90	of	of	ADP
ajst-27534	35	91	experiments	experiment	NOUN
ajst-27534	35	92	comparison	comparison	NOUN
ajst-27534	35	93	of	of	ADP
ajst-27534	35	94	residual	residual	ADJ
ajst-27534	35	95	structure	structure	NOUN
ajst-27534	35	96	running	running	NOUN
ajst-27534	35	97	time	time	NOUN
ajst-27534	35	98	partial	partial	ADJ
ajst-27534	35	99	improvement	improvement	NOUN
ajst-27534	35	100	based	base	VERB
ajst-27534	35	101	on	on	ADP
ajst-27534	35	102	pre	pre	NOUN
ajst-27534	35	103	activation	activation	NOUN
ajst-27534	35	104	introducing	introduce	VERB
ajst-27534	35	105	pre	pre	VERB
ajst-27534	35	106	activated	activate	VERB
ajst-27534	35	107	incomplete	incomplete	ADJ
ajst-27534	35	108	improvements	improvement	NOUN
ajst-27534	35	109	original	original	ADJ
ajst-27534	35	110	incomplete	incomplete	ADJ
ajst-27534	35	111	structure	structure	NOUN
ajst-27534	35	112	204	204	NUM
ajst-27534	35	113	slight	slight	ADJ
ajst-27534	35	114	defects	defect	NOUN
ajst-27534	35	115	are	be	AUX
ajst-27534	35	116	considered	consider	VERB
ajst-27534	35	117	superior	superior	ADJ
ajst-27534	35	118	fruits	fruit	NOUN
ajst-27534	35	119	,	,	PUNCT
ajst-27534	35	120	as	as	SCONJ
ajst-27534	35	121	shown	show	VERB
ajst-27534	35	122	in	in	ADP
ajst-27534	35	123	figure	figure	NOUN
ajst-27534	35	124	1	1	NUM
ajst-27534	35	125	.	.	PUNCT
ajst-27534	36	1	citrus	citrus	NOUN
ajst-27534	36	2	fruits	fruit	NOUN
ajst-27534	36	3	with	with	ADP
ajst-27534	36	4	bright	bright	ADJ
ajst-27534	36	5	color	color	NOUN
ajst-27534	36	6	,	,	PUNCT
ajst-27534	36	7	overall	overall	ADJ
ajst-27534	36	8	yellow	yellow	ADJ
ajst-27534	36	9	color	color	NOUN
ajst-27534	36	10	,	,	PUNCT
ajst-27534	36	11	and	and	CCONJ
ajst-27534	36	12	deep	deep	ADJ
ajst-27534	36	13	color	color	NOUN
ajst-27534	36	14	but	but	CCONJ
ajst-27534	36	15	obvious	obvious	ADJ
ajst-27534	36	16	defects	defect	NOUN
ajst-27534	36	17	are	be	AUX
ajst-27534	36	18	classified	classify	VERB
ajst-27534	36	19	as	as	ADP
ajst-27534	36	20	first	first	ADJ
ajst-27534	36	21	-	-	PUNCT
ajst-27534	36	22	class	class	NOUN
ajst-27534	36	23	fruits	fruit	NOUN
ajst-27534	36	24	,	,	PUNCT
ajst-27534	36	25	as	as	SCONJ
ajst-27534	36	26	shown	show	VERB
ajst-27534	36	27	in	in	ADP
ajst-27534	36	28	figure	figure	NOUN
ajst-27534	36	29	2	2	NUM
ajst-27534	36	30	.	.	PUNCT
ajst-27534	37	1	citrus	citrus	NOUN
ajst-27534	37	2	fruits	fruit	NOUN
ajst-27534	37	3	with	with	ADP
ajst-27534	37	4	obvious	obvious	ADJ
ajst-27534	37	5	green	green	ADJ
ajst-27534	37	6	color	color	NOUN
ajst-27534	37	7	are	be	AUX
ajst-27534	37	8	classified	classify	VERB
ajst-27534	37	9	as	as	ADP
ajst-27534	37	10	second	second	ADJ
ajst-27534	37	11	-	-	PUNCT
ajst-27534	37	12	class	class	NOUN
ajst-27534	37	13	fruits	fruit	NOUN
ajst-27534	37	14	,	,	PUNCT
ajst-27534	37	15	as	as	SCONJ
ajst-27534	37	16	shown	show	VERB
ajst-27534	37	17	in	in	ADP
ajst-27534	37	18	figure	figure	NOUN
ajst-27534	37	19	3	3	NUM
ajst-27534	37	20	.	.	PUNCT
ajst-27534	38	1	citrus	citrus	NOUN
ajst-27534	38	2	fruits	fruit	NOUN
ajst-27534	38	3	with	with	ADP
ajst-27534	38	4	rotten	rotten	ADJ
ajst-27534	38	5	fruit	fruit	NOUN
ajst-27534	38	6	and	and	CCONJ
ajst-27534	38	7	green	green	ADJ
ajst-27534	38	8	main	main	ADJ
ajst-27534	38	9	body	body	NOUN
ajst-27534	38	10	but	but	CCONJ
ajst-27534	38	11	severe	severe	ADJ
ajst-27534	38	12	damage	damage	NOUN
ajst-27534	38	13	are	be	AUX
ajst-27534	38	14	classified	classify	VERB
ajst-27534	38	15	as	as	ADP
ajst-27534	38	16	inferior	inferior	ADJ
ajst-27534	38	17	fruits	fruit	NOUN
ajst-27534	38	18	,	,	PUNCT
ajst-27534	38	19	as	as	SCONJ
ajst-27534	38	20	shown	show	VERB
ajst-27534	38	21	in	in	ADP
ajst-27534	38	22	figure	figure	NOUN
ajst-27534	38	23	4	4	NUM
ajst-27534	39	1	[	[	SYM
ajst-27534	39	2	8	8	NUM
ajst-27534	39	3	-	-	SYM
ajst-27534	39	4	9	9	NUM
ajst-27534	39	5	]	]	PUNCT
ajst-27534	39	6	.	.	PUNCT
ajst-27534	40	1	figure	figure	NOUN
ajst-27534	40	2	1	1	NUM
ajst-27534	40	3	.	.	PUNCT
ajst-27534	41	1	superior	superior	ADJ
ajst-27534	41	2	fruits	fruit	NOUN
ajst-27534	41	3	figure	figure	VERB
ajst-27534	41	4	2	2	NUM
ajst-27534	41	5	.	.	PUNCT
ajst-27534	41	6	first	first	ADJ
ajst-27534	41	7	-	-	PUNCT
ajst-27534	41	8	class	class	NOUN
ajst-27534	41	9	fruit	fruit	NOUN
ajst-27534	41	10	figure	figure	NOUN
ajst-27534	41	11	3	3	NUM
ajst-27534	41	12	.	.	PUNCT
ajst-27534	41	13	second	second	ADJ
ajst-27534	41	14	-	-	PUNCT
ajst-27534	41	15	class	class	NOUN
ajst-27534	41	16	fruit	fruit	NOUN
ajst-27534	41	17	figure	figure	NOUN
ajst-27534	41	18	4	4	NUM
ajst-27534	41	19	.	.	PUNCT
ajst-27534	41	20	inferior	inferior	ADJ
ajst-27534	41	21	fruits	fruit	NOUN
ajst-27534	41	22	3.2	3.2	NUM
ajst-27534	41	23	.	.	PUNCT
ajst-27534	42	1	experimental	experimental	ADJ
ajst-27534	42	2	considerations	consideration	NOUN
ajst-27534	42	3	the	the	DET
ajst-27534	42	4	images	image	NOUN
ajst-27534	42	5	in	in	ADP
ajst-27534	42	6	the	the	DET
ajst-27534	42	7	dataset	dataset	NOUN
ajst-27534	42	8	are	be	AUX
ajst-27534	42	9	divided	divide	VERB
ajst-27534	42	10	into	into	ADP
ajst-27534	42	11	a	a	DET
ajst-27534	42	12	training	training	NOUN
ajst-27534	42	13	set	set	NOUN
ajst-27534	42	14	and	and	CCONJ
ajst-27534	42	15	a	a	DET
ajst-27534	42	16	testing	testing	NOUN
ajst-27534	42	17	set	set	VERB
ajst-27534	42	18	according	accord	VERB
ajst-27534	42	19	to	to	ADP
ajst-27534	42	20	a	a	DET
ajst-27534	42	21	certain	certain	ADJ
ajst-27534	42	22	proportion	proportion	NOUN
ajst-27534	42	23	,	,	PUNCT
ajst-27534	42	24	and	and	CCONJ
ajst-27534	42	25	then	then	ADV
ajst-27534	42	26	the	the	DET
ajst-27534	42	27	images	image	NOUN
ajst-27534	42	28	in	in	ADP
ajst-27534	42	29	the	the	DET
ajst-27534	42	30	training	training	NOUN
ajst-27534	42	31	set	set	NOUN
ajst-27534	42	32	are	be	AUX
ajst-27534	42	33	normalized	normalize	VERB
ajst-27534	42	34	to	to	PART
ajst-27534	42	35	meet	meet	VERB
ajst-27534	42	36	a	a	DET
ajst-27534	42	37	normal	normal	ADJ
ajst-27534	42	38	distribution	distribution	NOUN
ajst-27534	42	39	.	.	PUNCT
ajst-27534	43	1	the	the	DET
ajst-27534	43	2	image	image	NOUN
ajst-27534	43	3	size	size	NOUN
ajst-27534	43	4	of	of	ADP
ajst-27534	43	5	all	all	DET
ajst-27534	43	6	training	training	NOUN
ajst-27534	43	7	sets	set	NOUN
ajst-27534	43	8	is	be	AUX
ajst-27534	43	9	adjusted	adjust	VERB
ajst-27534	43	10	to	to	ADP
ajst-27534	43	11	a	a	DET
ajst-27534	43	12	uniform	uniform	ADJ
ajst-27534	43	13	size	size	NOUN
ajst-27534	43	14	,	,	PUNCT
ajst-27534	43	15	and	and	CCONJ
ajst-27534	43	16	then	then	ADV
ajst-27534	43	17	the	the	DET
ajst-27534	43	18	training	training	NOUN
ajst-27534	43	19	set	set	NOUN
ajst-27534	43	20	images	image	NOUN
ajst-27534	43	21	are	be	AUX
ajst-27534	43	22	further	far	ADV
ajst-27534	43	23	divided	divide	VERB
ajst-27534	43	24	into	into	ADP
ajst-27534	43	25	a	a	DET
ajst-27534	43	26	validation	validation	NOUN
ajst-27534	43	27	set	set	NOUN
ajst-27534	43	28	and	and	CCONJ
ajst-27534	43	29	a	a	DET
ajst-27534	43	30	training	training	NOUN
ajst-27534	43	31	set	set	NOUN
ajst-27534	43	32	according	accord	VERB
ajst-27534	43	33	to	to	ADP
ajst-27534	43	34	a	a	DET
ajst-27534	43	35	certain	certain	ADJ
ajst-27534	43	36	proportion	proportion	NOUN
ajst-27534	43	37	.	.	PUNCT
ajst-27534	44	1	the	the	DET
ajst-27534	44	2	validation	validation	NOUN
ajst-27534	44	3	set	set	NOUN
ajst-27534	44	4	does	do	AUX
ajst-27534	44	5	not	not	PART
ajst-27534	44	6	participate	participate	VERB
ajst-27534	44	7	in	in	ADP
ajst-27534	44	8	backpropagation	backpropagation	NOUN
ajst-27534	44	9	.	.	PUNCT
ajst-27534	45	1	the	the	DET
ajst-27534	45	2	total	total	ADJ
ajst-27534	45	3	training	training	NOUN
ajst-27534	45	4	set	set	NOUN
ajst-27534	45	5	images	image	NOUN
ajst-27534	45	6	are	be	AUX
ajst-27534	45	7	input	input	VERB
ajst-27534	45	8	into	into	ADP
ajst-27534	45	9	205	205	NUM
ajst-27534	45	10	the	the	DET
ajst-27534	45	11	resnet	resnet	NOUN
ajst-27534	45	12	network	network	NOUN
ajst-27534	45	13	model	model	NOUN
ajst-27534	45	14	for	for	ADP
ajst-27534	45	15	training	training	NOUN
ajst-27534	45	16	,	,	PUNCT
ajst-27534	45	17	and	and	CCONJ
ajst-27534	45	18	then	then	ADV
ajst-27534	45	19	backpropagation	backpropagation	NOUN
ajst-27534	45	20	is	be	AUX
ajst-27534	45	21	performed	perform	VERB
ajst-27534	45	22	according	accord	VERB
ajst-27534	45	23	to	to	ADP
ajst-27534	45	24	the	the	DET
ajst-27534	45	25	loss	loss	NOUN
ajst-27534	45	26	function	function	NOUN
ajst-27534	45	27	.	.	PUNCT
ajst-27534	46	1	the	the	DET
ajst-27534	46	2	parameters	parameter	NOUN
ajst-27534	46	3	are	be	AUX
ajst-27534	46	4	changed	change	VERB
ajst-27534	46	5	to	to	PART
ajst-27534	46	6	save	save	VERB
ajst-27534	46	7	the	the	DET
ajst-27534	46	8	best	good	ADJ
ajst-27534	46	9	training	training	NOUN
ajst-27534	46	10	parameter	parameter	NOUN
ajst-27534	46	11	results	result	NOUN
ajst-27534	46	12	,	,	PUNCT
ajst-27534	46	13	and	and	CCONJ
ajst-27534	46	14	the	the	DET
ajst-27534	46	15	best	good	ADJ
ajst-27534	46	16	parameter	parameter	NOUN
ajst-27534	46	17	results	result	NOUN
ajst-27534	46	18	are	be	AUX
ajst-27534	46	19	put	put	VERB
ajst-27534	46	20	into	into	ADP
ajst-27534	46	21	the	the	DET
ajst-27534	46	22	testing	testing	NOUN
ajst-27534	46	23	set	set	VERB
ajst-27534	46	24	for	for	ADP
ajst-27534	46	25	testing	testing	NOUN
ajst-27534	46	26	.	.	PUNCT
ajst-27534	47	1	during	during	ADP
ajst-27534	47	2	testing	testing	NOUN
ajst-27534	47	3	,	,	PUNCT
ajst-27534	47	4	the	the	DET
ajst-27534	47	5	testing	testing	NOUN
ajst-27534	47	6	data	datum	NOUN
ajst-27534	47	7	also	also	ADV
ajst-27534	47	8	needs	need	VERB
ajst-27534	47	9	to	to	PART
ajst-27534	47	10	be	be	AUX
ajst-27534	47	11	normalized	normalize	VERB
ajst-27534	47	12	according	accord	VERB
ajst-27534	47	13	to	to	ADP
ajst-27534	47	14	the	the	DET
ajst-27534	47	15	mean	mean	NOUN
ajst-27534	47	16	and	and	CCONJ
ajst-27534	47	17	variance	variance	NOUN
ajst-27534	47	18	of	of	ADP
ajst-27534	47	19	the	the	DET
ajst-27534	47	20	training	training	NOUN
ajst-27534	47	21	set	set	NOUN
ajst-27534	47	22	.	.	PUNCT
ajst-27534	48	1	3.3	3.3	NUM
ajst-27534	48	2	.	.	PUNCT
ajst-27534	49	1	loss	loss	NOUN
ajst-27534	49	2	function	function	NOUN
ajst-27534	49	3	design	design	NOUN
ajst-27534	49	4	using	use	VERB
ajst-27534	49	5	the	the	DET
ajst-27534	49	6	cross	cross	NOUN
ajst-27534	49	7	entropy	entropy	PROPN
ajst-27534	49	8	loss	loss	NOUN
ajst-27534	49	9	function	function	NOUN
ajst-27534	49	10	,	,	PUNCT
ajst-27534	49	11	measure	measure	VERB
ajst-27534	49	12	the	the	DET
ajst-27534	49	13	similarity	similarity	NOUN
ajst-27534	49	14	between	between	ADP
ajst-27534	49	15	the	the	DET
ajst-27534	49	16	predicted	predict	VERB
ajst-27534	49	17	probability	probability	NOUN
ajst-27534	49	18	distribution	distribution	NOUN
ajst-27534	49	19	and	and	CCONJ
ajst-27534	49	20	the	the	DET
ajst-27534	49	21	true	true	ADJ
ajst-27534	49	22	probability	probability	NOUN
ajst-27534	49	23	distribution	distribution	NOUN
ajst-27534	49	24	by	by	ADP
ajst-27534	49	25	calculating	calculate	VERB
ajst-27534	49	26	the	the	DET
ajst-27534	49	27	cross	cross	NOUN
ajst-27534	49	28	entropy	entropy	NOUN
ajst-27534	49	29	between	between	ADP
ajst-27534	49	30	the	the	DET
ajst-27534	49	31	two	two	NUM
ajst-27534	49	32	.	.	PUNCT
ajst-27534	50	1	the	the	DET
ajst-27534	50	2	function	function	NOUN
ajst-27534	50	3	model	model	NOUN
ajst-27534	50	4	is	be	AUX
ajst-27534	50	5	as	as	SCONJ
ajst-27534	50	6	follows	follow	VERB
ajst-27534	50	7	:	:	PUNCT
ajst-27534	50	8	l=∑	l=∑	VERB
ajst-27534	50	9	∑	∑	PROPN
ajst-27534	50	10	𝑦	𝑦	PRON
ajst-27534	50	11	log	log	VERB
ajst-27534	50	12	𝑝	𝑝	NOUN
ajst-27534	50	13	n	n	PRON
ajst-27534	50	14	represents	represent	VERB
ajst-27534	50	15	the	the	DET
ajst-27534	50	16	number	number	NOUN
ajst-27534	50	17	of	of	ADP
ajst-27534	50	18	samples	sample	NOUN
ajst-27534	50	19	,	,	PUNCT
ajst-27534	50	20	k	k	PROPN
ajst-27534	50	21	represents	represent	VERB
ajst-27534	50	22	the	the	DET
ajst-27534	50	23	number	number	NOUN
ajst-27534	50	24	of	of	ADP
ajst-27534	50	25	categories	category	NOUN
ajst-27534	50	26	,	,	PUNCT
ajst-27534	50	27	pic	pic	NOUN
ajst-27534	50	28	represents	represent	VERB
ajst-27534	50	29	the	the	DET
ajst-27534	50	30	probability	probability	NOUN
ajst-27534	50	31	that	that	SCONJ
ajst-27534	50	32	the	the	DET
ajst-27534	50	33	ith	ith	PROPN
ajst-27534	50	34	sample	sample	NOUN
ajst-27534	50	35	belongs	belong	VERB
ajst-27534	50	36	to	to	ADP
ajst-27534	50	37	category	category	PROPN
ajst-27534	50	38	c	c	NOUN
ajst-27534	50	39	,	,	PUNCT
ajst-27534	50	40	yic={0,1	yic={0,1	NOUN
ajst-27534	50	41	}	}	PUNCT
ajst-27534	50	42	.	.	PUNCT
ajst-27534	51	1	if	if	SCONJ
ajst-27534	51	2	the	the	DET
ajst-27534	51	3	i	i	PROPN
ajst-27534	51	4	-	-	PUNCT
ajst-27534	51	5	th	th	VERB
ajst-27534	51	6	sample	sample	NOUN
ajst-27534	51	7	belongs	belong	VERB
ajst-27534	51	8	to	to	ADP
ajst-27534	51	9	category	category	PROPN
ajst-27534	51	10	c	c	NOUN
ajst-27534	51	11	,	,	PUNCT
ajst-27534	51	12	then	then	ADV
ajst-27534	51	13	the	the	DET
ajst-27534	51	14	corresponding	correspond	VERB
ajst-27534	51	15	position	position	NOUN
ajst-27534	51	16	's	's	PART
ajst-27534	51	17	yic	yic	NOUN
ajst-27534	51	18	is	be	AUX
ajst-27534	51	19	taken	take	VERB
ajst-27534	51	20	as	as	ADP
ajst-27534	51	21	1	1	NUM
ajst-27534	51	22	,	,	PUNCT
ajst-27534	51	23	otherwise	otherwise	ADV
ajst-27534	51	24	it	it	PRON
ajst-27534	51	25	is	be	AUX
ajst-27534	51	26	taken	take	VERB
ajst-27534	51	27	as	as	ADP
ajst-27534	51	28	0	0	NUM
ajst-27534	51	29	.	.	NOUN
ajst-27534	51	30	3.4	3.4	NUM
ajst-27534	51	31	.	.	PUNCT
ajst-27534	52	1	network	network	NOUN
ajst-27534	52	2	architecture	architecture	NOUN
ajst-27534	52	3	construction	construction	NOUN
ajst-27534	52	4	the	the	DET
ajst-27534	52	5	network	network	NOUN
ajst-27534	52	6	structure	structure	NOUN
ajst-27534	52	7	is	be	AUX
ajst-27534	52	8	basically	basically	ADV
ajst-27534	52	9	built	build	VERB
ajst-27534	52	10	according	accord	VERB
ajst-27534	52	11	to	to	ADP
ajst-27534	52	12	the	the	DET
ajst-27534	52	13	resnet18	resnet18	NOUN
ajst-27534	52	14	network	network	NOUN
ajst-27534	52	15	,	,	PUNCT
ajst-27534	52	16	and	and	CCONJ
ajst-27534	52	17	the	the	DET
ajst-27534	52	18	residual	residual	ADJ
ajst-27534	52	19	blocks	block	NOUN
ajst-27534	52	20	are	be	AUX
ajst-27534	52	21	replaced	replace	VERB
ajst-27534	52	22	according	accord	VERB
ajst-27534	52	23	to	to	ADP
ajst-27534	52	24	different	different	ADJ
ajst-27534	52	25	schemes	scheme	NOUN
ajst-27534	52	26	.	.	PUNCT
ajst-27534	53	1	when	when	SCONJ
ajst-27534	53	2	training	train	VERB
ajst-27534	53	3	the	the	DET
ajst-27534	53	4	network	network	NOUN
ajst-27534	53	5	,	,	PUNCT
ajst-27534	53	6	the	the	DET
ajst-27534	53	7	batch	batch	NOUN
ajst-27534	53	8	size	size	NOUN
ajst-27534	53	9	is	be	AUX
ajst-27534	53	10	32	32	NUM
ajst-27534	53	11	.	.	PUNCT
ajst-27534	54	1	at	at	ADP
ajst-27534	54	2	the	the	DET
ajst-27534	54	3	same	same	ADJ
ajst-27534	54	4	time	time	NOUN
ajst-27534	54	5	,	,	PUNCT
ajst-27534	54	6	the	the	DET
ajst-27534	54	7	softmax	softmax	NOUN
ajst-27534	54	8	structure	structure	NOUN
ajst-27534	54	9	of	of	ADP
ajst-27534	54	10	resnet18	resnet18	NOUN
ajst-27534	54	11	,	,	PUNCT
ajst-27534	54	12	which	which	PRON
ajst-27534	54	13	was	be	AUX
ajst-27534	54	14	originally	originally	ADV
ajst-27534	54	15	fully	fully	ADV
ajst-27534	54	16	connected	connect	VERB
ajst-27534	54	17	to	to	ADP
ajst-27534	54	18	the	the	DET
ajst-27534	54	19	network	network	NOUN
ajst-27534	54	20	,	,	PUNCT
ajst-27534	54	21	was	be	AUX
ajst-27534	54	22	cancelled	cancel	VERB
ajst-27534	54	23	.	.	PUNCT
ajst-27534	55	1	using	use	VERB
ajst-27534	55	2	the	the	DET
ajst-27534	55	3	resnet18	resnet18	NOUN
ajst-27534	55	4	network	network	NOUN
ajst-27534	55	5	structure	structure	NOUN
ajst-27534	55	6	,	,	PUNCT
ajst-27534	55	7	after	after	ADP
ajst-27534	55	8	training	train	VERB
ajst-27534	55	9	2	2	NUM
ajst-27534	55	10	-	-	SYM
ajst-27534	55	11	3	3	NUM
ajst-27534	55	12	times	time	NOUN
ajst-27534	55	13	,	,	PUNCT
ajst-27534	55	14	the	the	DET
ajst-27534	55	15	accuracy	accuracy	NOUN
ajst-27534	55	16	of	of	ADP
ajst-27534	55	17	the	the	DET
ajst-27534	55	18	validation	validation	NOUN
ajst-27534	55	19	set	set	NOUN
ajst-27534	55	20	can	can	AUX
ajst-27534	55	21	achieve	achieve	VERB
ajst-27534	55	22	a	a	DET
ajst-27534	55	23	relatively	relatively	ADV
ajst-27534	55	24	good	good	ADJ
ajst-27534	55	25	result	result	NOUN
ajst-27534	55	26	.	.	PUNCT
ajst-27534	56	1	the	the	DET
ajst-27534	56	2	use	use	NOUN
ajst-27534	56	3	of	of	ADP
ajst-27534	56	4	the	the	DET
ajst-27534	56	5	softmax	softmax	NOUN
ajst-27534	56	6	structure	structure	NOUN
ajst-27534	56	7	can	can	AUX
ajst-27534	56	8	significantly	significantly	ADV
ajst-27534	56	9	reduce	reduce	VERB
ajst-27534	56	10	the	the	DET
ajst-27534	56	11	probability	probability	NOUN
ajst-27534	56	12	of	of	ADP
ajst-27534	56	13	other	other	ADJ
ajst-27534	56	14	categories	category	NOUN
ajst-27534	56	15	when	when	SCONJ
ajst-27534	56	16	a	a	DET
ajst-27534	56	17	large	large	ADJ
ajst-27534	56	18	positive	positive	ADJ
ajst-27534	56	19	input	input	NOUN
ajst-27534	56	20	value	value	NOUN
ajst-27534	56	21	appears	appear	VERB
ajst-27534	56	22	,	,	PUNCT
ajst-27534	56	23	making	make	VERB
ajst-27534	56	24	the	the	DET
ajst-27534	56	25	network	network	NOUN
ajst-27534	56	26	prone	prone	ADJ
ajst-27534	56	27	to	to	ADP
ajst-27534	56	28	falling	fall	VERB
ajst-27534	56	29	into	into	ADP
ajst-27534	56	30	a	a	DET
ajst-27534	56	31	local	local	ADJ
ajst-27534	56	32	optimum	optimum	NOUN
ajst-27534	56	33	.	.	PUNCT
ajst-27534	57	1	therefore	therefore	ADV
ajst-27534	57	2	,	,	PUNCT
ajst-27534	57	3	considering	consider	VERB
ajst-27534	57	4	the	the	DET
ajst-27534	57	5	overall	overall	ADJ
ajst-27534	57	6	situation	situation	NOUN
ajst-27534	57	7	,	,	PUNCT
ajst-27534	57	8	the	the	DET
ajst-27534	57	9	softmax	softmax	NOUN
ajst-27534	57	10	structure	structure	NOUN
ajst-27534	57	11	was	be	AUX
ajst-27534	57	12	eliminated	eliminate	VERB
ajst-27534	57	13	[	[	PUNCT
ajst-27534	57	14	9	9	NUM
ajst-27534	57	15	]	]	PUNCT
ajst-27534	57	16	.	.	PUNCT
ajst-27534	58	1	referring	refer	VERB
ajst-27534	58	2	to	to	ADP
ajst-27534	58	3	the	the	DET
ajst-27534	58	4	paper	paper	NOUN
ajst-27534	58	5	"	"	PUNCT
ajst-27534	58	6	research	research	NOUN
ajst-27534	58	7	on	on	ADP
ajst-27534	58	8	clothing	clothing	NOUN
ajst-27534	58	9	image	image	NOUN
ajst-27534	58	10	style	style	NOUN
ajst-27534	58	11	recognition	recognition	NOUN
ajst-27534	58	12	based	base	VERB
ajst-27534	58	13	on	on	ADP
ajst-27534	58	14	improved	improve	VERB
ajst-27534	58	15	resnet	resnet	NOUN
ajst-27534	58	16	network	network	NOUN
ajst-27534	58	17	and	and	CCONJ
ajst-27534	58	18	transfer	transfer	NOUN
ajst-27534	58	19	learning	learning	NOUN
ajst-27534	58	20	"	"	PUNCT
ajst-27534	58	21	,	,	PUNCT
ajst-27534	58	22	modifications	modification	NOUN
ajst-27534	58	23	were	be	AUX
ajst-27534	58	24	made	make	VERB
ajst-27534	58	25	to	to	ADP
ajst-27534	58	26	the	the	DET
ajst-27534	58	27	resnet18	resnet18	NOUN
ajst-27534	58	28	network	network	NOUN
ajst-27534	58	29	by	by	ADP
ajst-27534	58	30	changing	change	VERB
ajst-27534	58	31	the	the	DET
ajst-27534	58	32	first	first	ADJ
ajst-27534	58	33	layer	layer	NOUN
ajst-27534	58	34	structure	structure	NOUN
ajst-27534	58	35	as	as	SCONJ
ajst-27534	58	36	shown	show	VERB
ajst-27534	58	37	in	in	ADP
ajst-27534	58	38	figure	figure	NOUN
ajst-27534	58	39	5	5	NUM
ajst-27534	58	40	:	:	PUNCT
ajst-27534	58	41	figure	figure	NOUN
ajst-27534	58	42	5	5	NUM
ajst-27534	58	43	.	.	PUNCT
ajst-27534	58	44	resnet18	resnet18	NOUN
ajst-27534	58	45	network	network	NOUN
ajst-27534	58	46	first	first	ADJ
ajst-27534	58	47	layer	layer	NOUN
ajst-27534	58	48	structure	structure	NOUN
ajst-27534	58	49	improvement	improvement	NOUN
ajst-27534	58	50	using	use	VERB
ajst-27534	58	51	a	a	DET
ajst-27534	58	52	combination	combination	NOUN
ajst-27534	58	53	of	of	ADP
ajst-27534	58	54	three	three	NUM
ajst-27534	58	55	3x3	3x3	NUM
ajst-27534	58	56	convolutional	convolutional	ADJ
ajst-27534	58	57	kernels	kernel	NOUN
ajst-27534	58	58	instead	instead	ADV
ajst-27534	58	59	of	of	ADP
ajst-27534	58	60	one	one	NUM
ajst-27534	58	61	7x7	7x7	NUM
ajst-27534	58	62	convolutional	convolutional	ADJ
ajst-27534	58	63	kernel	kernel	NOUN
ajst-27534	58	64	can	can	AUX
ajst-27534	58	65	maintain	maintain	VERB
ajst-27534	58	66	the	the	DET
ajst-27534	58	67	receptive	receptive	ADJ
ajst-27534	58	68	field	field	NOUN
ajst-27534	58	69	and	and	CCONJ
ajst-27534	58	70	output	output	NOUN
ajst-27534	58	71	feature	feature	NOUN
ajst-27534	58	72	map	map	NOUN
ajst-27534	58	73	size	size	NOUN
ajst-27534	58	74	unchanged	unchanged	ADJ
ajst-27534	58	75	in	in	ADP
ajst-27534	58	76	the	the	DET
ajst-27534	58	77	network	network	NOUN
ajst-27534	58	78	[	[	X
ajst-27534	58	79	11	11	NUM
ajst-27534	58	80	]	]	PUNCT
ajst-27534	58	81	.	.	PUNCT
ajst-27534	59	1	the	the	DET
ajst-27534	59	2	receptive	receptive	ADJ
ajst-27534	59	3	field	field	NOUN
ajst-27534	59	4	is	be	AUX
ajst-27534	59	5	the	the	DET
ajst-27534	59	6	region	region	NOUN
ajst-27534	59	7	of	of	ADP
ajst-27534	59	8	influence	influence	NOUN
ajst-27534	59	9	of	of	ADP
ajst-27534	59	10	a	a	DET
ajst-27534	59	11	unit	unit	NOUN
ajst-27534	59	12	on	on	ADP
ajst-27534	59	13	the	the	DET
ajst-27534	59	14	input	input	NOUN
ajst-27534	59	15	of	of	ADP
ajst-27534	59	16	a	a	DET
ajst-27534	59	17	feature	feature	NOUN
ajst-27534	59	18	map	map	NOUN
ajst-27534	59	19	output	output	NOUN
ajst-27534	59	20	by	by	ADP
ajst-27534	59	21	a	a	DET
ajst-27534	59	22	certain	certain	ADJ
ajst-27534	59	23	layer	layer	NOUN
ajst-27534	59	24	in	in	ADP
ajst-27534	59	25	a	a	DET
ajst-27534	59	26	neural	neural	ADJ
ajst-27534	59	27	network	network	NOUN
ajst-27534	59	28	[	[	X
ajst-27534	59	29	9	9	NUM
ajst-27534	59	30	]	]	PUNCT
ajst-27534	59	31	.	.	PUNCT
ajst-27534	60	1	the	the	DET
ajst-27534	60	2	calculation	calculation	NOUN
ajst-27534	60	3	formula	formula	NOUN
ajst-27534	60	4	is	be	AUX
ajst-27534	60	5	as	as	SCONJ
ajst-27534	60	6	follows	follow	VERB
ajst-27534	60	7	:	:	PUNCT
ajst-27534	61	1	f	f	PROPN
ajst-27534	62	1	i	i	PRON
ajst-27534	62	2	f	f	PROPN
ajst-27534	63	1	i	i	NOUN
ajst-27534	63	2	1	1	NUM
ajst-27534	63	3	1	1	NUM
ajst-27534	63	4	stride	stride	NOUN
ajst-27534	63	5	ksize	ksize	NOUN
ajst-27534	63	6	according	accord	VERB
ajst-27534	63	7	to	to	ADP
ajst-27534	63	8	the	the	DET
ajst-27534	63	9	formula	formula	NOUN
ajst-27534	63	10	,	,	PUNCT
ajst-27534	63	11	it	it	PRON
ajst-27534	63	12	can	can	AUX
ajst-27534	63	13	be	be	AUX
ajst-27534	63	14	calculated	calculate	VERB
ajst-27534	63	15	that	that	SCONJ
ajst-27534	63	16	using	use	VERB
ajst-27534	63	17	a	a	DET
ajst-27534	63	18	combination	combination	NOUN
ajst-27534	63	19	of	of	ADP
ajst-27534	63	20	three	three	NUM
ajst-27534	63	21	3	3	NUM
ajst-27534	63	22	×	×	NOUN
ajst-27534	63	23	3	3	NUM
ajst-27534	63	24	convolutional	convolutional	ADJ
ajst-27534	63	25	kernels	kernel	NOUN
ajst-27534	63	26	to	to	PART
ajst-27534	63	27	replace	replace	VERB
ajst-27534	63	28	one	one	NUM
ajst-27534	63	29	7	7	NUM
ajst-27534	63	30	×	×	NOUN
ajst-27534	63	31	7	7	NUM
ajst-27534	63	32	convolutional	convolutional	ADJ
ajst-27534	63	33	kernel	kernel	NOUN
ajst-27534	63	34	can	can	AUX
ajst-27534	63	35	maintain	maintain	VERB
ajst-27534	63	36	the	the	DET
ajst-27534	63	37	same	same	ADJ
ajst-27534	63	38	receptive	receptive	ADJ
ajst-27534	63	39	field	field	NOUN
ajst-27534	63	40	size	size	NOUN
ajst-27534	63	41	,	,	PUNCT
ajst-27534	63	42	and	and	CCONJ
ajst-27534	63	43	the	the	DET
ajst-27534	63	44	hierarchical	hierarchical	ADJ
ajst-27534	63	45	structure	structure	NOUN
ajst-27534	63	46	of	of	ADP
ajst-27534	63	47	three	three	NUM
ajst-27534	63	48	3x3	3x3	NUM
ajst-27534	63	49	convolution	convolution	NOUN
ajst-27534	63	50	kernels	kernel	NOUN
ajst-27534	63	51	can	can	AUX
ajst-27534	63	52	increase	increase	VERB
ajst-27534	63	53	the	the	DET
ajst-27534	63	54	depth	depth	NOUN
ajst-27534	63	55	of	of	ADP
ajst-27534	63	56	the	the	DET
ajst-27534	63	57	network	network	NOUN
ajst-27534	63	58	,	,	PUNCT
ajst-27534	63	59	capture	capture	VERB
ajst-27534	63	60	style	style	NOUN
ajst-27534	63	61	features	feature	NOUN
ajst-27534	63	62	at	at	ADP
ajst-27534	63	63	multiple	multiple	ADJ
ajst-27534	63	64	scales	scale	NOUN
ajst-27534	63	65	,	,	PUNCT
ajst-27534	63	66	introduce	introduce	VERB
ajst-27534	63	67	more	more	ADJ
ajst-27534	63	68	nonlinearity	nonlinearity	NOUN
ajst-27534	63	69	,	,	PUNCT
ajst-27534	63	70	reduce	reduce	VERB
ajst-27534	63	71	overfitting	overfitting	NOUN
ajst-27534	63	72	,	,	PUNCT
ajst-27534	63	73	and	and	CCONJ
ajst-27534	63	74	improve	improve	VERB
ajst-27534	63	75	the	the	DET
ajst-27534	63	76	performance	performance	NOUN
ajst-27534	63	77	of	of	ADP
ajst-27534	63	78	the	the	DET
ajst-27534	63	79	network	network	NOUN
ajst-27534	63	80	[	[	X
ajst-27534	63	81	9	9	NUM
ajst-27534	63	82	]	]	PUNCT
ajst-27534	63	83	.	.	PUNCT
ajst-27534	64	1	we	we	PRON
ajst-27534	64	2	randomly	randomly	ADV
ajst-27534	64	3	divided	divide	VERB
ajst-27534	64	4	the	the	DET
ajst-27534	64	5	dataset	dataset	NOUN
ajst-27534	64	6	into	into	ADP
ajst-27534	64	7	a	a	DET
ajst-27534	64	8	test	test	NOUN
ajst-27534	64	9	set	set	VERB
ajst-27534	64	10	and	and	CCONJ
ajst-27534	64	11	a	a	DET
ajst-27534	64	12	training	training	NOUN
ajst-27534	64	13	set	set	NOUN
ajst-27534	64	14	,	,	PUNCT
ajst-27534	64	15	and	and	CCONJ
ajst-27534	64	16	then	then	ADV
ajst-27534	64	17	trained	train	VERB
ajst-27534	64	18	it	it	PRON
ajst-27534	64	19	20	20	NUM
ajst-27534	64	20	times	time	NOUN
ajst-27534	64	21	using	use	VERB
ajst-27534	64	22	the	the	DET
ajst-27534	64	23	resnet	resnet	ADJ
ajst-27534	64	24	network	network	NOUN
ajst-27534	64	25	.	.	PUNCT
ajst-27534	65	1	finally	finally	ADV
ajst-27534	65	2	,	,	PUNCT
ajst-27534	65	3	we	we	PRON
ajst-27534	65	4	conducted	conduct	VERB
ajst-27534	65	5	testing	testing	NOUN
ajst-27534	65	6	on	on	ADP
ajst-27534	65	7	the	the	DET
ajst-27534	65	8	test	test	NOUN
ajst-27534	65	9	set	set	NOUN
ajst-27534	65	10	,	,	PUNCT
ajst-27534	65	11	which	which	PRON
ajst-27534	65	12	was	be	AUX
ajst-27534	65	13	a	a	DET
ajst-27534	65	14	round	round	NOUN
ajst-27534	65	15	of	of	ADP
ajst-27534	65	16	experiments	experiment	NOUN
ajst-27534	65	17	.	.	PUNCT
ajst-27534	66	1	then	then	ADV
ajst-27534	66	2	,	,	PUNCT
ajst-27534	66	3	we	we	PRON
ajst-27534	66	4	conducted	conduct	VERB
ajst-27534	66	5	10	10	NUM
ajst-27534	66	6	rounds	round	NOUN
ajst-27534	66	7	of	of	ADP
ajst-27534	66	8	comparative	comparative	ADJ
ajst-27534	66	9	experiments	experiment	NOUN
ajst-27534	66	10	on	on	ADP
ajst-27534	66	11	the	the	DET
ajst-27534	66	12	two	two	NUM
ajst-27534	66	13	resnet	resnet	NOUN
ajst-27534	66	14	network	network	NOUN
ajst-27534	66	15	structures	structure	NOUN
ajst-27534	66	16	.	.	PUNCT
ajst-27534	67	1	in	in	ADP
ajst-27534	67	2	each	each	DET
ajst-27534	67	3	round	round	NOUN
ajst-27534	67	4	of	of	ADP
ajst-27534	67	5	comparative	comparative	ADJ
ajst-27534	67	6	experiments	experiment	NOUN
ajst-27534	67	7	,	,	PUNCT
ajst-27534	67	8	the	the	DET
ajst-27534	67	9	training	training	NOUN
ajst-27534	67	10	and	and	CCONJ
ajst-27534	67	11	testing	testing	NOUN
ajst-27534	67	12	sets	set	NOUN
ajst-27534	67	13	of	of	ADP
ajst-27534	67	14	the	the	DET
ajst-27534	67	15	two	two	NUM
ajst-27534	67	16	resnet	resnet	NOUN
ajst-27534	67	17	network	network	NOUN
ajst-27534	67	18	structures	structure	NOUN
ajst-27534	67	19	were	be	AUX
ajst-27534	67	20	consistent	consistent	ADJ
ajst-27534	67	21	.	.	PUNCT
ajst-27534	68	1	experimental	experimental	ADJ
ajst-27534	68	2	analysis	analysis	NOUN
ajst-27534	68	3	shows	show	VERB
ajst-27534	68	4	that	that	SCONJ
ajst-27534	68	5	after	after	ADP
ajst-27534	68	6	using	use	VERB
ajst-27534	68	7	the	the	DET
ajst-27534	68	8	first	first	ADJ
ajst-27534	68	9	layer	layer	NOUN
ajst-27534	68	10	improvement	improvement	NOUN
ajst-27534	68	11	of	of	ADP
ajst-27534	68	12	the	the	DET
ajst-27534	68	13	network	network	NOUN
ajst-27534	68	14	,	,	PUNCT
ajst-27534	68	15	both	both	CCONJ
ajst-27534	68	16	the	the	DET
ajst-27534	68	17	accuracy	accuracy	NOUN
ajst-27534	68	18	of	of	ADP
ajst-27534	68	19	the	the	DET
ajst-27534	68	20	test	test	NOUN
ajst-27534	68	21	set	set	VERB
ajst-27534	68	22	and	and	CCONJ
ajst-27534	68	23	the	the	DET
ajst-27534	68	24	training	training	NOUN
ajst-27534	68	25	time	time	NOUN
ajst-27534	68	26	have	have	AUX
ajst-27534	68	27	been	be	AUX
ajst-27534	68	28	significantly	significantly	ADV
ajst-27534	68	29	optimized	optimize	VERB
ajst-27534	68	30	in	in	ADP
ajst-27534	68	31	each	each	DET
ajst-27534	68	32	experiment	experiment	NOUN
ajst-27534	68	33	.	.	PUNCT
ajst-27534	69	1	in	in	ADP
ajst-27534	69	2	each	each	DET
ajst-27534	69	3	comparative	comparative	ADJ
ajst-27534	69	4	experiment	experiment	NOUN
ajst-27534	69	5	,	,	PUNCT
ajst-27534	69	6	experiments	experiment	NOUN
ajst-27534	69	7	using	use	VERB
ajst-27534	69	8	roll	roll	NOUN
ajst-27534	69	9	in	in	ADP
ajst-27534	69	10	improvement	improvement	NOUN
ajst-27534	69	11	have	have	AUX
ajst-27534	69	12	shown	show	VERB
ajst-27534	69	13	significant	significant	ADJ
ajst-27534	69	14	improvements	improvement	NOUN
ajst-27534	69	15	in	in	ADP
ajst-27534	69	16	both	both	CCONJ
ajst-27534	69	17	the	the	DET
ajst-27534	69	18	training	training	NOUN
ajst-27534	69	19	time	time	NOUN
ajst-27534	69	20	and	and	CCONJ
ajst-27534	69	21	the	the	DET
ajst-27534	69	22	accuracy	accuracy	NOUN
ajst-27534	69	23	of	of	ADP
ajst-27534	69	24	the	the	DET
ajst-27534	69	25	test	test	NOUN
ajst-27534	69	26	set	set	NOUN
ajst-27534	69	27	compared	compare	VERB
ajst-27534	69	28	to	to	ADP
ajst-27534	69	29	the	the	DET
ajst-27534	69	30	original	original	ADJ
ajst-27534	69	31	experimental	experimental	ADJ
ajst-27534	69	32	scheme	scheme	NOUN
ajst-27534	69	33	without	without	ADP
ajst-27534	69	34	improvement	improvement	NOUN
ajst-27534	69	35	.	.	PUNCT
ajst-27534	70	1	using	use	VERB
ajst-27534	70	2	convolutional	convolutional	ADJ
ajst-27534	70	3	blocks	block	NOUN
ajst-27534	70	4	for	for	ADP
ajst-27534	70	5	improvement	improvement	NOUN
ajst-27534	70	6	,	,	PUNCT
ajst-27534	70	7	the	the	DET
ajst-27534	70	8	average	average	ADJ
ajst-27534	70	9	accuracy	accuracy	NOUN
ajst-27534	70	10	of	of	ADP
ajst-27534	70	11	the	the	DET
ajst-27534	70	12	test	test	NOUN
ajst-27534	70	13	set	set	VERB
ajst-27534	70	14	is	be	AUX
ajst-27534	70	15	0.95245	0.95245	NUM
ajst-27534	70	16	with	with	ADP
ajst-27534	70	17	a	a	DET
ajst-27534	70	18	variance	variance	NOUN
ajst-27534	70	19	of	of	ADP
ajst-27534	70	20	1.43125	1.43125	NUM
ajst-27534	70	21	×	×	NOUN
ajst-27534	70	22	10	10	NUM
ajst-27534	70	23	-4	-4	ADP
ajst-27534	70	24	,	,	PUNCT
ajst-27534	70	25	while	while	SCONJ
ajst-27534	70	26	the	the	DET
ajst-27534	70	27	original	original	ADJ
ajst-27534	70	28	plan	plan	NOUN
ajst-27534	70	29	without	without	ADP
ajst-27534	70	30	improvement	improvement	NOUN
ajst-27534	70	31	has	have	VERB
ajst-27534	70	32	an	an	DET
ajst-27534	70	33	average	average	ADJ
ajst-27534	70	34	test	test	NOUN
ajst-27534	70	35	set	set	NOUN
ajst-27534	70	36	of	of	ADP
ajst-27534	70	37	0.91	0.91	NUM
ajst-27534	70	38	with	with	ADP
ajst-27534	70	39	a	a	DET
ajst-27534	70	40	variance	variance	NOUN
ajst-27534	70	41	of	of	ADP
ajst-27534	70	42	1.958	1.958	NUM
ajst-27534	70	43	×	×	NOUN
ajst-27534	70	44	10	10	NUM
ajst-27534	70	45	-4	-4	PUNCT
ajst-27534	70	46	after	after	ADP
ajst-27534	70	47	10	10	NUM
ajst-27534	70	48	experiments	experiment	NOUN
ajst-27534	70	49	.	.	PUNCT
ajst-27534	71	1	compared	compare	VERB
ajst-27534	71	2	to	to	ADP
ajst-27534	71	3	the	the	DET
ajst-27534	71	4	original	original	ADJ
ajst-27534	71	5	scheme	scheme	NOUN
ajst-27534	71	6	,	,	PUNCT
ajst-27534	71	7	the	the	DET
ajst-27534	71	8	convolutional	convolutional	ADJ
ajst-27534	71	9	improvement	improvement	NOUN
ajst-27534	71	10	scheme	scheme	NOUN
ajst-27534	71	11	has	have	AUX
ajst-27534	71	12	increased	increase	VERB
ajst-27534	71	13	accuracy	accuracy	NOUN
ajst-27534	71	14	by	by	ADP
ajst-27534	71	15	4.245	4.245	NUM
ajst-27534	71	16	percentage	percentage	NOUN
ajst-27534	71	17	points	point	NOUN
ajst-27534	71	18	and	and	CCONJ
ajst-27534	71	19	is	be	AUX
ajst-27534	71	20	also	also	ADV
ajst-27534	71	21	more	more	ADV
ajst-27534	71	22	stable	stable	ADJ
ajst-27534	71	23	.	.	PUNCT
ajst-27534	72	1	in	in	ADP
ajst-27534	72	2	terms	term	NOUN
ajst-27534	72	3	of	of	ADP
ajst-27534	72	4	training	training	NOUN
ajst-27534	72	5	time	time	NOUN
ajst-27534	72	6	,	,	PUNCT
ajst-27534	72	7	the	the	DET
ajst-27534	72	8	original	original	ADJ
ajst-27534	72	9	scheme	scheme	NOUN
ajst-27534	72	10	had	have	VERB
ajst-27534	72	11	an	an	DET
ajst-27534	72	12	average	average	ADJ
ajst-27534	72	13	training	training	NOUN
ajst-27534	72	14	time	time	NOUN
ajst-27534	72	15	of	of	ADP
ajst-27534	72	16	2527.7s	2527.7s	NUM
ajst-27534	72	17	and	and	CCONJ
ajst-27534	72	18	a	a	DET
ajst-27534	72	19	variance	variance	NOUN
ajst-27534	72	20	of	of	ADP
ajst-27534	72	21	2689.79s	2689.79s	NUM
ajst-27534	72	22	2	2	NUM
ajst-27534	72	23	.	.	PUNCT
ajst-27534	73	1	the	the	DET
ajst-27534	73	2	improved	improved	ADJ
ajst-27534	73	3	convolution	convolution	NOUN
ajst-27534	73	4	scheme	scheme	NOUN
ajst-27534	73	5	had	have	VERB
ajst-27534	73	6	an	an	DET
ajst-27534	73	7	average	average	ADJ
ajst-27534	73	8	training	training	NOUN
ajst-27534	73	9	time	time	NOUN
ajst-27534	73	10	of	of	ADP
ajst-27534	73	11	1479.4s	1479.4s	NUM
ajst-27534	73	12	and	and	CCONJ
ajst-27534	73	13	a	a	DET
ajst-27534	73	14	variance	variance	NOUN
ajst-27534	73	15	of	of	ADP
ajst-27534	73	16	331.6s	331.6s	NUM
ajst-27534	73	17	2	2	NUM
ajst-27534	73	18	.	.	PUNCT
ajst-27534	74	1	the	the	DET
ajst-27534	74	2	improved	improved	ADJ
ajst-27534	74	3	convolution	convolution	NOUN
ajst-27534	74	4	scheme	scheme	NOUN
ajst-27534	74	5	reduced	reduce	VERB
ajst-27534	74	6	the	the	DET
ajst-27534	74	7	training	training	NOUN
ajst-27534	74	8	time	time	NOUN
ajst-27534	74	9	by	by	ADP
ajst-27534	74	10	an	an	DET
ajst-27534	74	11	average	average	NOUN
ajst-27534	74	12	of	of	ADP
ajst-27534	74	13	17.17min	17.17min	NUM
ajst-27534	74	14	and	and	CCONJ
ajst-27534	74	15	was	be	AUX
ajst-27534	74	16	also	also	ADV
ajst-27534	74	17	more	more	ADV
ajst-27534	74	18	stable	stable	ADJ
ajst-27534	74	19	in	in	ADP
ajst-27534	74	20	terms	term	NOUN
ajst-27534	74	21	of	of	ADP
ajst-27534	74	22	training	training	NOUN
ajst-27534	74	23	time	time	NOUN
ajst-27534	74	24	.	.	PUNCT
ajst-27534	75	1	so	so	ADV
ajst-27534	75	2	overall	overall	ADV
ajst-27534	75	3	,	,	PUNCT
ajst-27534	75	4	the	the	DET
ajst-27534	75	5	convolution	convolution	NOUN
ajst-27534	75	6	improvement	improvement	NOUN
ajst-27534	75	7	scheme	scheme	NOUN
ajst-27534	75	8	is	be	AUX
ajst-27534	75	9	significantly	significantly	ADV
ajst-27534	75	10	optimized	optimize	VERB
ajst-27534	75	11	.	.	PUNCT
ajst-27534	76	1	comparison	comparison	NOUN
ajst-27534	76	2	with	with	ADP
ajst-27534	76	3	other	other	ADJ
ajst-27534	76	4	network	network	NOUN
ajst-27534	76	5	structures	structure	NOUN
ajst-27534	76	6	:	:	PUNCT
ajst-27534	76	7	test	test	NOUN
ajst-27534	76	8	set	set	VERB
ajst-27534	76	9	accuracy	accuracy	NOUN
ajst-27534	76	10	introducing	introduce	VERB
ajst-27534	76	11	pre	pre	VERB
ajst-27534	76	12	activated	activate	VERB
ajst-27534	76	13	residuals	residual	NOUN
ajst-27534	76	14	to	to	PART
ajst-27534	76	15	improve	improve	VERB
ajst-27534	76	16	resnet	resnet	ADJ
ajst-27534	76	17	network	network	NOUN
ajst-27534	76	18	0.9376	0.9376	NUM
ajst-27534	76	19	l	l	NOUN
ajst-27534	76	20	-	-	ADJ
ajst-27534	76	21	se	se	ADJ
ajst-27534	76	22	-	-	ADJ
ajst-27534	76	23	vgg	vgg	ADJ
ajst-27534	76	24	network	network	NOUN
ajst-27534	76	25	0.8875	0.8875	NUM
ajst-27534	76	26	input	input	NOUN
ajst-27534	76	27	graph	graph	NOUN
ajst-27534	76	28	conv(3ⅹ3	conv(3ⅹ3	PROPN
ajst-27534	76	29	)	)	PUNCT
ajst-27534	76	30	conv(3ⅹ3	conv(3ⅹ3	PROPN
ajst-27534	76	31	)	)	PUNCT
ajst-27534	76	32	conv(3ⅹ3	conv(3ⅹ3	PROPN
ajst-27534	76	33	)	)	PUNCT
ajst-27534	76	34	input	input	NOUN
ajst-27534	76	35	graph	graph	NOUN
ajst-27534	76	36	conv(7ⅹ7	conv(7ⅹ7	PROPN
ajst-27534	76	37	)	)	PUNCT
ajst-27534	76	38	206	206	NUM
ajst-27534	76	39	4	4	NUM
ajst-27534	76	40	.	.	PUNCT
ajst-27534	76	41	summarize	summarize	VERB
ajst-27534	76	42	at	at	ADP
ajst-27534	76	43	present	present	ADJ
ajst-27534	76	44	,	,	PUNCT
ajst-27534	76	45	the	the	DET
ajst-27534	76	46	research	research	NOUN
ajst-27534	76	47	and	and	CCONJ
ajst-27534	76	48	development	development	NOUN
ajst-27534	76	49	achievements	achievement	NOUN
ajst-27534	76	50	in	in	ADP
ajst-27534	76	51	key	key	ADJ
ajst-27534	76	52	components	component	NOUN
ajst-27534	76	53	of	of	ADP
ajst-27534	76	54	fruit	fruit	NOUN
ajst-27534	76	55	picking	picking	NOUN
ajst-27534	76	56	devices	device	NOUN
ajst-27534	76	57	in	in	ADP
ajst-27534	76	58	china	china	PROPN
ajst-27534	76	59	are	be	AUX
ajst-27534	76	60	still	still	ADV
ajst-27534	76	61	insufficient	insufficient	ADJ
ajst-27534	76	62	,	,	PUNCT
ajst-27534	76	63	especially	especially	ADV
ajst-27534	76	64	in	in	ADP
ajst-27534	76	65	terms	term	NOUN
ajst-27534	76	66	of	of	ADP
ajst-27534	76	67	intelligent	intelligent	ADJ
ajst-27534	76	68	and	and	CCONJ
ajst-27534	76	69	efficient	efficient	ADJ
ajst-27534	76	70	,	,	PUNCT
ajst-27534	76	71	environmental	environmental	ADJ
ajst-27534	76	72	adaptability	adaptability	NOUN
ajst-27534	76	73	,	,	PUNCT
ajst-27534	76	74	and	and	CCONJ
ajst-27534	76	75	coordination	coordination	NOUN
ajst-27534	76	76	and	and	CCONJ
ajst-27534	76	77	cooperation	cooperation	NOUN
ajst-27534	76	78	ability	ability	NOUN
ajst-27534	76	79	during	during	ADP
ajst-27534	76	80	the	the	DET
ajst-27534	76	81	picking	picking	NOUN
ajst-27534	76	82	process	process	NOUN
ajst-27534	76	83	.	.	PUNCT
ajst-27534	77	1	there	there	PRON
ajst-27534	77	2	is	be	VERB
ajst-27534	77	3	still	still	ADV
ajst-27534	77	4	significant	significant	ADJ
ajst-27534	77	5	room	room	NOUN
ajst-27534	77	6	for	for	ADP
ajst-27534	77	7	improvement	improvement	NOUN
ajst-27534	77	8	.	.	PUNCT
ajst-27534	78	1	given	give	VERB
ajst-27534	78	2	the	the	DET
ajst-27534	78	3	urgent	urgent	ADJ
ajst-27534	78	4	demand	demand	NOUN
ajst-27534	78	5	for	for	ADP
ajst-27534	78	6	citrus	citrus	NOUN
ajst-27534	78	7	picking	pick	VERB
ajst-27534	78	8	in	in	ADP
ajst-27534	78	9	the	the	DET
ajst-27534	78	10	industry	industry	NOUN
ajst-27534	78	11	,	,	PUNCT
ajst-27534	78	12	this	this	DET
ajst-27534	78	13	article	article	NOUN
ajst-27534	78	14	aims	aim	VERB
ajst-27534	78	15	to	to	PART
ajst-27534	78	16	explore	explore	VERB
ajst-27534	78	17	and	and	CCONJ
ajst-27534	78	18	design	design	VERB
ajst-27534	78	19	an	an	DET
ajst-27534	78	20	innovative	innovative	ADJ
ajst-27534	78	21	citrus	citrus	NOUN
ajst-27534	78	22	picker	picker	NOUN
ajst-27534	78	23	with	with	ADP
ajst-27534	78	24	automatic	automatic	ADJ
ajst-27534	78	25	detection	detection	NOUN
ajst-27534	78	26	,	,	PUNCT
ajst-27534	78	27	classification	classification	NOUN
ajst-27534	78	28	and	and	CCONJ
ajst-27534	78	29	recognition	recognition	NOUN
ajst-27534	78	30	functions	function	NOUN
ajst-27534	78	31	,	,	PUNCT
ajst-27534	78	32	which	which	PRON
ajst-27534	78	33	can	can	AUX
ajst-27534	78	34	reduce	reduce	VERB
ajst-27534	78	35	labor	labor	NOUN
ajst-27534	78	36	and	and	CCONJ
ajst-27534	78	37	economic	economic	ADJ
ajst-27534	78	38	costs	cost	NOUN
ajst-27534	78	39	and	and	CCONJ
ajst-27534	78	40	improve	improve	VERB
ajst-27534	78	41	picking	pick	VERB
ajst-27534	78	42	efficiency	efficiency	NOUN
ajst-27534	78	43	.	.	PUNCT
ajst-27534	79	1	based	base	VERB
ajst-27534	79	2	on	on	ADP
ajst-27534	79	3	the	the	DET
ajst-27534	79	4	external	external	ADJ
ajst-27534	79	5	characteristics	characteristic	NOUN
ajst-27534	79	6	of	of	ADP
ajst-27534	79	7	citrus	citrus	NOUN
ajst-27534	79	8	fruits	fruit	NOUN
ajst-27534	79	9	and	and	CCONJ
ajst-27534	79	10	their	their	PRON
ajst-27534	79	11	growth	growth	NOUN
ajst-27534	79	12	environment	environment	NOUN
ajst-27534	79	13	,	,	PUNCT
ajst-27534	79	14	a	a	DET
ajst-27534	79	15	flexible	flexible	ADJ
ajst-27534	79	16	end	end	NOUN
ajst-27534	79	17	effector	effector	NOUN
ajst-27534	79	18	device	device	NOUN
ajst-27534	79	19	for	for	ADP
ajst-27534	79	20	cutting	cut	VERB
ajst-27534	79	21	and	and	CCONJ
ajst-27534	79	22	clamping	clamping	NOUN
ajst-27534	79	23	was	be	AUX
ajst-27534	79	24	designed	design	VERB
ajst-27534	79	25	to	to	PART
ajst-27534	79	26	mimic	mimic	VERB
ajst-27534	79	27	the	the	DET
ajst-27534	79	28	manual	manual	NOUN
ajst-27534	79	29	picking	picking	NOUN
ajst-27534	79	30	of	of	ADP
ajst-27534	79	31	citrus	citrus	NOUN
ajst-27534	79	32	fruits	fruit	NOUN
ajst-27534	79	33	.	.	PUNCT
ajst-27534	80	1	this	this	DET
ajst-27534	80	2	device	device	NOUN
ajst-27534	80	3	ensures	ensure	VERB
ajst-27534	80	4	the	the	DET
ajst-27534	80	5	avoidance	avoidance	NOUN
ajst-27534	80	6	of	of	ADP
ajst-27534	80	7	obstacles	obstacle	NOUN
ajst-27534	80	8	,	,	PUNCT
ajst-27534	80	9	achieves	achieve	VERB
ajst-27534	80	10	precise	precise	ADJ
ajst-27534	80	11	and	and	CCONJ
ajst-27534	80	12	non	non	ADJ
ajst-27534	80	13	-	-	ADJ
ajst-27534	80	14	destructive	destructive	ADJ
ajst-27534	80	15	grasping	grasping	NOUN
ajst-27534	80	16	,	,	PUNCT
ajst-27534	80	17	and	and	CCONJ
ajst-27534	80	18	successfully	successfully	ADV
ajst-27534	80	19	completes	complete	VERB
ajst-27534	80	20	the	the	DET
ajst-27534	80	21	citrus	citrus	NOUN
ajst-27534	80	22	fruit	fruit	NOUN
ajst-27534	80	23	picking	picking	NOUN
ajst-27534	80	24	process	process	NOUN
ajst-27534	80	25	.	.	PUNCT
ajst-27534	81	1	then	then	ADV
ajst-27534	81	2	,	,	PUNCT
ajst-27534	81	3	according	accord	VERB
ajst-27534	81	4	to	to	ADP
ajst-27534	81	5	the	the	DET
ajst-27534	81	6	built	build	VERB
ajst-27534	81	7	-	-	PUNCT
ajst-27534	81	8	in	in	ADP
ajst-27534	81	9	classification	classification	NOUN
ajst-27534	81	10	detection	detection	NOUN
ajst-27534	81	11	algorithm	algorithm	NOUN
ajst-27534	81	12	model	model	NOUN
ajst-27534	81	13	,	,	PUNCT
ajst-27534	81	14	citrus	citrus	NOUN
ajst-27534	81	15	fruits	fruit	NOUN
ajst-27534	81	16	can	can	AUX
ajst-27534	81	17	be	be	AUX
ajst-27534	81	18	classified	classify	VERB
ajst-27534	81	19	and	and	CCONJ
ajst-27534	81	20	processed	process	VERB
ajst-27534	81	21	.	.	PUNCT
ajst-27534	82	1	due	due	ADP
ajst-27534	82	2	to	to	ADP
ajst-27534	82	3	the	the	DET
ajst-27534	82	4	improvement	improvement	NOUN
ajst-27534	82	5	of	of	ADP
ajst-27534	82	6	the	the	DET
ajst-27534	82	7	resnet18	resnet18	NOUN
ajst-27534	82	8	first	first	ADJ
ajst-27534	82	9	layer	layer	NOUN
ajst-27534	82	10	network	network	NOUN
ajst-27534	82	11	in	in	ADP
ajst-27534	82	12	programming	programming	NOUN
ajst-27534	82	13	,	,	PUNCT
ajst-27534	82	14	this	this	DET
ajst-27534	82	15	design	design	NOUN
ajst-27534	82	16	has	have	AUX
ajst-27534	82	17	shortened	shorten	VERB
ajst-27534	82	18	the	the	DET
ajst-27534	82	19	time	time	NOUN
ajst-27534	82	20	and	and	CCONJ
ajst-27534	82	21	improved	improve	VERB
ajst-27534	82	22	the	the	DET
ajst-27534	82	23	accuracy	accuracy	NOUN
ajst-27534	82	24	of	of	ADP
ajst-27534	82	25	classification	classification	NOUN
ajst-27534	82	26	and	and	CCONJ
ajst-27534	82	27	recognition	recognition	NOUN
ajst-27534	82	28	.	.	PUNCT
ajst-27534	83	1	therefore	therefore	ADV
ajst-27534	83	2	,	,	PUNCT
ajst-27534	83	3	updating	update	VERB
ajst-27534	83	4	the	the	DET
ajst-27534	83	5	classification	classification	NOUN
ajst-27534	83	6	recognition	recognition	NOUN
ajst-27534	83	7	detection	detection	NOUN
ajst-27534	83	8	network	network	NOUN
ajst-27534	83	9	model	model	NOUN
ajst-27534	83	10	can	can	AUX
ajst-27534	83	11	promote	promote	VERB
ajst-27534	83	12	the	the	DET
ajst-27534	83	13	development	development	NOUN
ajst-27534	83	14	of	of	ADP
ajst-27534	83	15	citrus	citrus	NOUN
ajst-27534	83	16	industrialization	industrialization	NOUN
ajst-27534	83	17	and	and	CCONJ
ajst-27534	83	18	scientific	scientific	ADJ
ajst-27534	83	19	management	management	NOUN
ajst-27534	83	20	of	of	ADP
ajst-27534	83	21	orchards	orchard	NOUN
ajst-27534	83	22	.	.	PUNCT
ajst-27534	84	1	the	the	DET
ajst-27534	84	2	scientific	scientific	ADJ
ajst-27534	84	3	and	and	CCONJ
ajst-27534	84	4	rational	rational	ADJ
ajst-27534	84	5	design	design	NOUN
ajst-27534	84	6	of	of	ADP
ajst-27534	84	7	fruit	fruit	NOUN
ajst-27534	84	8	picker	picker	NOUN
ajst-27534	84	9	structure	structure	NOUN
ajst-27534	84	10	is	be	AUX
ajst-27534	84	11	of	of	ADP
ajst-27534	84	12	great	great	ADJ
ajst-27534	84	13	significance	significance	NOUN
ajst-27534	84	14	for	for	ADP
ajst-27534	84	15	improving	improve	VERB
ajst-27534	84	16	the	the	DET
ajst-27534	84	17	intelligence	intelligence	NOUN
ajst-27534	84	18	level	level	NOUN
ajst-27534	84	19	of	of	ADP
ajst-27534	84	20	citrus	citrus	NOUN
ajst-27534	84	21	picking	picking	NOUN
ajst-27534	84	22	and	and	CCONJ
ajst-27534	84	23	citrus	citrus	VERB
ajst-27534	84	24	agricultural	agricultural	ADJ
ajst-27534	84	25	production	production	NOUN
ajst-27534	84	26	.	.	PUNCT
ajst-27534	85	1	5	5	X
ajst-27534	85	2	.	.	X
ajst-27534	85	3	funding	funding	NOUN
ajst-27534	85	4	statement	statement	NOUN
ajst-27534	85	5	this	this	DET
ajst-27534	85	6	work	work	NOUN
ajst-27534	85	7	was	be	AUX
ajst-27534	85	8	supported	support	VERB
ajst-27534	85	9	by	by	ADP
ajst-27534	85	10	the	the	DET
ajst-27534	85	11	chongqing	chongqing	PROPN
ajst-27534	85	12	natural	natural	PROPN
ajst-27534	85	13	science	science	PROPN
ajst-27534	85	14	foundation	foundation	PROPN
ajst-27534	85	15	upper	upper	ADJ
ajst-27534	85	16	-	-	PUNCT
ajst-27534	85	17	level	level	NOUN
ajst-27534	85	18	project	project	NOUN
ajst-27534	85	19	(	(	PUNCT
ajst-27534	85	20	cstb2023nscqmsx1069	cstb2023nscqmsx1069	PROPN
ajst-27534	85	21	)	)	PUNCT
ajst-27534	85	22	;	;	PUNCT
ajst-27534	85	23	science	science	NOUN
ajst-27534	85	24	and	and	CCONJ
ajst-27534	85	25	technology	technology	NOUN
ajst-27534	85	26	research	research	NOUN
ajst-27534	85	27	project	project	NOUN
ajst-27534	85	28	of	of	ADP
ajst-27534	85	29	chongqing	chongqing	PROPN
ajst-27534	85	30	municipal	municipal	ADJ
ajst-27534	85	31	education	education	PROPN
ajst-27534	85	32	commission	commission	PROPN
ajst-27534	85	33	(	(	PUNCT
ajst-27534	85	34	kjzdk202401801	kjzdk202401801	PROPN
ajst-27534	85	35	)	)	PUNCT
ajst-27534	85	36	chongqing	chongqing	PROPN
ajst-27534	85	37	college	college	NOUN
ajst-27534	85	38	of	of	ADP
ajst-27534	85	39	humanities	humanity	NOUN
ajst-27534	85	40	and	and	CCONJ
ajst-27534	85	41	technology	technology	NOUN
ajst-27534	85	42	school	school	NOUN
ajst-27534	85	43	level	level	NOUN
ajst-27534	85	44	project	project	NOUN
ajst-27534	85	45	(	(	PUNCT
ajst-27534	85	46	crkzk2024002	crkzk2024002	PROPN
ajst-27534	85	47	)	)	PUNCT
ajst-27534	85	48	;	;	PUNCT
ajst-27534	85	49	chongqing	chongqe	VERB
ajst-27534	85	50	college	college	NOUN
ajst-27534	85	51	of	of	ADP
ajst-27534	85	52	humanities	humanity	NOUN
ajst-27534	85	53	and	and	CCONJ
ajst-27534	85	54	technology	technology	NOUN
ajst-27534	85	55	school	school	NOUN
ajst-27534	85	56	level	level	NOUN
ajst-27534	85	57	project	project	NOUN
ajst-27534	85	58	(	(	PUNCT
ajst-27534	85	59	crkzk2021003	crkzk2021003	PROPN
ajst-27534	85	60	)	)	PUNCT
ajst-27534	85	61	;	;	PUNCT
ajst-27534	85	62	science	science	NOUN
ajst-27534	85	63	and	and	CCONJ
ajst-27534	85	64	technology	technology	NOUN
ajst-27534	85	65	project	project	NOUN
ajst-27534	85	66	of	of	ADP
ajst-27534	85	67	chongqing	chongqing	PROPN
ajst-27534	85	68	municipal	municipal	ADJ
ajst-27534	85	69	education	education	NOUN
ajst-27534	85	70	commission	commission	NOUN
ajst-27534	85	71	(	(	PUNCT
ajst-27534	85	72	kjqn202101801	kjqn202101801	PROPN
ajst-27534	85	73	)	)	PUNCT
ajst-27534	85	74	;	;	PUNCT
ajst-27534	85	75	chongqing	chongqe	VERB
ajst-27534	85	76	municipal	municipal	ADJ
ajst-27534	85	77	education	education	PROPN
ajst-27534	85	78	commission	commission	PROPN
ajst-27534	85	79	scientific	scientific	ADJ
ajst-27534	85	80	research	research	NOUN
ajst-27534	85	81	technology	technology	NOUN
ajst-27534	85	82	project	project	NOUN
ajst-27534	85	83	youth	youth	NOUN
ajst-27534	85	84	project	project	NOUN
ajst-27534	85	85	(	(	PUNCT
ajst-27534	85	86	kjqn202001802	kjqn202001802	PROPN
ajst-27534	85	87	)	)	PUNCT
ajst-27534	85	88	;	;	PUNCT
ajst-27534	85	89	disclosures	disclosure	NOUN
ajst-27534	85	90	.	.	PUNCT
ajst-27534	86	1	the	the	DET
ajst-27534	86	2	authors	author	NOUN
ajst-27534	86	3	declare	declare	VERB
ajst-27534	86	4	no	no	DET
ajst-27534	86	5	conflicts	conflict	NOUN
ajst-27534	86	6	of	of	ADP
ajst-27534	86	7	interest	interest	NOUN
ajst-27534	86	8	.	.	PUNCT
ajst-27534	87	1	data	datum	NOUN
ajst-27534	87	2	availability	availability	NOUN
ajst-27534	87	3	.	.	PUNCT
ajst-27534	88	1	no	no	DET
ajst-27534	88	2	data	datum	NOUN
ajst-27534	88	3	were	be	AUX
ajst-27534	88	4	generated	generate	VERB
ajst-27534	88	5	or	or	CCONJ
ajst-27534	88	6	analyzed	analyze	VERB
ajst-27534	88	7	in	in	ADP
ajst-27534	88	8	the	the	DET
ajst-27534	88	9	presented	present	VERB
ajst-27534	88	10	research	research	NOUN
ajst-27534	88	11	.	.	PUNCT
ajst-27534	89	1	references	reference	NOUN
ajst-27534	89	2	[	[	X
ajst-27534	89	3	1	1	NUM
ajst-27534	89	4	]	]	X
ajst-27534	89	5	zhang	zhang	PROPN
ajst-27534	89	6	chenyu	chenyu	PROPN
ajst-27534	89	7	,	,	PUNCT
ajst-27534	89	8	nie	nie	PROPN
ajst-27534	89	9	shuping	shuping	PROPN
ajst-27534	89	10	,	,	PUNCT
ajst-27534	89	11	zhang	zhang	PROPN
ajst-27534	89	12	hongzhen	hongzhen	PROPN
ajst-27534	89	13	,	,	PUNCT
ajst-27534	89	14	etc	etc	X
ajst-27534	89	15	design	design	NOUN
ajst-27534	89	16	of	of	ADP
ajst-27534	89	17	an	an	DET
ajst-27534	89	18	automated	automate	VERB
ajst-27534	89	19	citrus	citrus	NOUN
ajst-27534	89	20	picking	picking	NOUN
ajst-27534	89	21	robot	robot	NOUN
ajst-27534	89	22	[	[	X
ajst-27534	89	23	j	j	X
ajst-27534	89	24	]	]	X
ajst-27534	89	25	agricultural	agricultural	ADJ
ajst-27534	89	26	machinery	machinery	NOUN
ajst-27534	89	27	use	use	NOUN
ajst-27534	89	28	and	and	CCONJ
ajst-27534	89	29	maintenance	maintenance	NOUN
ajst-27534	89	30	,	,	PUNCT
ajst-27534	89	31	2023	2023	NUM
ajst-27534	89	32	,	,	PUNCT
ajst-27534	89	33	(	(	PUNCT
ajst-27534	89	34	09	09	NUM
ajst-27534	89	35	):	):	PUNCT
ajst-27534	89	36	10	10	NUM
ajst-27534	89	37	-	-	SYM
ajst-27534	89	38	15	15	NUM
ajst-27534	90	1	[	[	X
ajst-27534	90	2	2	2	NUM
ajst-27534	90	3	]	]	PUNCT
ajst-27534	90	4	xiong	xiong	PROPN
ajst-27534	90	5	longye	longye	PROPN
ajst-27534	90	6	research	research	NOUN
ajst-27534	90	7	on	on	ADP
ajst-27534	90	8	classification	classification	NOUN
ajst-27534	90	9	,	,	PUNCT
ajst-27534	90	10	recognition	recognition	NOUN
ajst-27534	90	11	,	,	PUNCT
ajst-27534	90	12	and	and	CCONJ
ajst-27534	90	13	localization	localization	NOUN
ajst-27534	90	14	methods	method	NOUN
ajst-27534	90	15	of	of	ADP
ajst-27534	90	16	mature	mature	ADJ
ajst-27534	90	17	citrus	citrus	NOUN
ajst-27534	90	18	fruits	fruit	NOUN
ajst-27534	90	19	in	in	ADP
ajst-27534	90	20	natural	natural	ADJ
ajst-27534	90	21	scenes	scene	NOUN
ajst-27534	90	22	[	[	X
ajst-27534	90	23	d	d	X
ajst-27534	90	24	]	]	X
ajst-27534	90	25	chongqing	chongqing	PROPN
ajst-27534	90	26	university	university	PROPN
ajst-27534	90	27	of	of	ADP
ajst-27534	90	28	technology	technology	NOUN
ajst-27534	90	29	,	,	PUNCT
ajst-27534	90	30	2020	2020	NUM
ajst-27534	91	1	[	[	X
ajst-27534	91	2	3	3	X
ajst-27534	91	3	]	]	PUNCT
ajst-27534	91	4	pan	pan	NOUN
ajst-27534	91	5	hong	hong	PROPN
ajst-27534	91	6	research	research	NOUN
ajst-27534	91	7	on	on	ADP
ajst-27534	91	8	multimodal	multimodal	ADJ
ajst-27534	91	9	medical	medical	ADJ
ajst-27534	91	10	image	image	NOUN
ajst-27534	91	11	recognition	recognition	NOUN
ajst-27534	91	12	of	of	ADP
ajst-27534	91	13	thyroid	thyroid	NOUN
ajst-27534	91	14	nodules	nodule	NOUN
ajst-27534	91	15	based	base	VERB
ajst-27534	91	16	on	on	ADP
ajst-27534	91	17	deep	deep	ADJ
ajst-27534	91	18	learning	learning	NOUN
ajst-27534	92	1	[	[	X
ajst-27534	92	2	d	d	X
ajst-27534	92	3	]	]	X
ajst-27534	92	4	zhejiang	zhejiang	PROPN
ajst-27534	92	5	university	university	PROPN
ajst-27534	92	6	of	of	ADP
ajst-27534	92	7	traditional	traditional	ADJ
ajst-27534	92	8	chinese	chinese	ADJ
ajst-27534	92	9	medicine	medicine	NOUN
ajst-27534	92	10	,	,	PUNCT
ajst-27534	92	11	2022	2022	NUM
ajst-27534	92	12	[	[	X
ajst-27534	92	13	4	4	NUM
ajst-27534	92	14	]	]	PUNCT
ajst-27534	92	15	xu	xu	PROPN
ajst-27534	92	16	pingting	pingting	PROPN
ajst-27534	92	17	,	,	PUNCT
ajst-27534	92	18	zheng	zheng	PROPN
ajst-27534	92	19	jiaqi	jiaqi	PROPN
ajst-27534	92	20	a	a	DET
ajst-27534	92	21	strong	strong	ADJ
ajst-27534	92	22	dog	dog	NOUN
ajst-27534	92	23	detection	detection	NOUN
ajst-27534	92	24	and	and	CCONJ
ajst-27534	92	25	reminder	reminder	NOUN
ajst-27534	92	26	system	system	NOUN
ajst-27534	92	27	based	base	VERB
ajst-27534	92	28	on	on	ADP
ajst-27534	92	29	deep	deep	ADJ
ajst-27534	92	30	learning	learning	NOUN
ajst-27534	92	31	algorithms	algorithm	NOUN
ajst-27534	92	32	[	[	X
ajst-27534	92	33	j	j	X
ajst-27534	92	34	]	]	X
ajst-27534	92	35	software	software	NOUN
ajst-27534	92	36	,	,	PUNCT
ajst-27534	92	37	2023	2023	NUM
ajst-27534	92	38	[	[	X
ajst-27534	92	39	5	5	X
ajst-27534	92	40	]	]	PUNCT
ajst-27534	92	41	yang	yang	PROPN
ajst-27534	92	42	hang	hang	PROPN
ajst-27534	92	43	identification	identification	NOUN
ajst-27534	92	44	algorithm	algorithm	NOUN
ajst-27534	92	45	and	and	CCONJ
ajst-27534	92	46	research	research	NOUN
ajst-27534	92	47	of	of	ADP
ajst-27534	92	48	i	i	PROPN
ajst-27534	92	49	d	d	PROPN
ajst-27534	92	50	card	card	NOUN
ajst-27534	92	51	number	number	NOUN
ajst-27534	93	1	[	[	X
ajst-27534	93	2	d	d	X
ajst-27534	93	3	]	]	X
ajst-27534	93	4	chongqing	chongqing	PROPN
ajst-27534	93	5	university	university	PROPN
ajst-27534	93	6	of	of	ADP
ajst-27534	93	7	posts	post	NOUN
ajst-27534	93	8	and	and	CCONJ
ajst-27534	93	9	telecommunications	telecommunication	NOUN
ajst-27534	93	10	,	,	PUNCT
ajst-27534	93	11	2017	2017	NUM
ajst-27534	93	12	[	[	SYM
ajst-27534	93	13	6	6	NUM
ajst-27534	93	14	]	]	PUNCT
ajst-27534	93	15	xiang	xiang	PROPN
ajst-27534	93	16	chao	chao	PROPN
ajst-27534	93	17	research	research	PROPN
ajst-27534	93	18	on	on	ADP
ajst-27534	93	19	human	human	ADJ
ajst-27534	93	20	posture	posture	NOUN
ajst-27534	93	21	analysis	analysis	NOUN
ajst-27534	93	22	of	of	ADP
ajst-27534	93	23	video	video	NOUN
ajst-27534	93	24	images	image	NOUN
ajst-27534	93	25	in	in	ADP
ajst-27534	93	26	unmanned	unmanned	ADJ
ajst-27534	93	27	aerial	aerial	ADJ
ajst-27534	93	28	vehicle	vehicle	NOUN
ajst-27534	93	29	systems	system	NOUN
ajst-27534	94	1	[	[	X
ajst-27534	94	2	d	d	X
ajst-27534	94	3	]	]	X
ajst-27534	94	4	nanjing	nanjing	PROPN
ajst-27534	94	5	university	university	PROPN
ajst-27534	94	6	of	of	ADP
ajst-27534	94	7	aeronautics	aeronautic	NOUN
ajst-27534	94	8	and	and	CCONJ
ajst-27534	94	9	astronautics	astronautic	NOUN
ajst-27534	94	10	,	,	PUNCT
ajst-27534	94	11	2020	2020	NUM
ajst-27534	95	1	[	[	X
ajst-27534	95	2	7	7	X
ajst-27534	95	3	]	]	PUNCT
ajst-27534	95	4	xia	xia	PROPN
ajst-27534	95	5	minggui	minggui	PROPN
ajst-27534	95	6	,	,	PUNCT
ajst-27534	95	7	tian	tian	ADJ
ajst-27534	95	8	ruijun	ruijun	PROPN
ajst-27534	95	9	,	,	PUNCT
ajst-27534	95	10	jiang	jiang	PROPN
ajst-27534	95	11	huiyu	huiyu	PROPN
ajst-27534	95	12	,	,	PUNCT
ajst-27534	95	13	etc	etc	X
ajst-27534	95	14	research	research	NOUN
ajst-27534	95	15	on	on	ADP
ajst-27534	95	16	clothing	clothing	NOUN
ajst-27534	95	17	image	image	NOUN
ajst-27534	95	18	style	style	NOUN
ajst-27534	95	19	recognition	recognition	NOUN
ajst-27534	95	20	based	base	VERB
ajst-27534	95	21	on	on	ADP
ajst-27534	95	22	improved	improve	VERB
ajst-27534	95	23	resnet	resnet	NOUN
ajst-27534	95	24	network	network	NOUN
ajst-27534	95	25	and	and	CCONJ
ajst-27534	95	26	transfer	transfer	NOUN
ajst-27534	95	27	learning	learn	VERB
ajst-27534	95	28	[	[	X
ajst-27534	95	29	j	j	X
ajst-27534	95	30	]	]	X
ajst-27534	95	31	journal	journal	NOUN
ajst-27534	95	32	of	of	ADP
ajst-27534	95	33	textile	textile	NOUN
ajst-27534	95	34	engineering	engineering	NOUN
ajst-27534	95	35	,	,	PUNCT
ajst-27534	95	36	2024	2024	NUM
ajst-27534	95	37	,	,	PUNCT
ajst-27534	95	38	2	2	NUM
ajst-27534	95	39	(	(	PUNCT
ajst-27534	95	40	01	01	NUM
ajst-27534	95	41	):	):	PUNCT
ajst-27534	95	42	12	12	NUM
ajst-27534	95	43	-	-	SYM
ajst-27534	95	44	20	20	NUM
ajst-27534	96	1	[	[	SYM
ajst-27534	96	2	8	8	NUM
ajst-27534	96	3	]	]	X
ajst-27534	96	4	fan	fan	NOUN
ajst-27534	96	5	liming	lime	VERB
ajst-27534	96	6	research	research	NOUN
ajst-27534	96	7	on	on	ADP
ajst-27534	96	8	the	the	DET
ajst-27534	96	9	classification	classification	NOUN
ajst-27534	96	10	method	method	NOUN
ajst-27534	96	11	of	of	ADP
ajst-27534	96	12	cervical	cervical	ADJ
ajst-27534	96	13	cell	cell	NOUN
ajst-27534	96	14	pathology	pathology	NOUN
ajst-27534	96	15	images	image	NOUN
ajst-27534	96	16	based	base	VERB
ajst-27534	96	17	on	on	ADP
ajst-27534	96	18	deep	deep	ADJ
ajst-27534	96	19	learning	learning	NOUN
ajst-27534	96	20	[	[	X
ajst-27534	96	21	d	d	X
ajst-27534	96	22	]	]	X
ajst-27534	96	23	xiangtan	xiangtan	PROPN
ajst-27534	96	24	university	university	PROPN
ajst-27534	96	25	,	,	PUNCT
ajst-27534	96	26	2021	2021	NUM
ajst-27534	97	1	[	[	X
ajst-27534	97	2	9	9	NUM
ajst-27534	97	3	]	]	PUNCT
ajst-27534	97	4	a.	a.	NOUN
ajst-27534	97	5	faraji	faraji	NOUN
ajst-27534	97	6	,	,	PUNCT
ajst-27534	97	7	m.	m.	NOUN
ajst-27534	97	8	noohi	noohi	PROPN
ajst-27534	97	9	,	,	PUNCT
ajst-27534	97	10	s.	s.	PROPN
ajst-27534	97	11	a.	a.	PROPN
ajst-27534	97	12	sadrossadat	sadrossadat	PROPN
ajst-27534	97	13	,	,	PUNCT
ajst-27534	97	14	a.	a.	NOUN
ajst-27534	97	15	mirvakili	mirvakili	NOUN
ajst-27534	97	16	,	,	PUNCT
ajst-27534	97	17	w.	w.	PROPN
ajst-27534	97	18	na	na	PROPN
ajst-27534	97	19	and	and	CCONJ
ajst-27534	97	20	f.	f.	PROPN
ajst-27534	97	21	feng	feng	PROPN
ajst-27534	97	22	,	,	PUNCT
ajst-27534	97	23	"	"	PUNCT
ajst-27534	97	24	batch	batch	NOUN
ajst-27534	97	25	-	-	PUNCT
ajst-27534	97	26	normalized	normalize	VERB
ajst-27534	97	27	deep	deep	ADJ
ajst-27534	97	28	recurrent	recurrent	ADJ
ajst-27534	97	29	neural	neural	ADJ
ajst-27534	97	30	network	network	NOUN
ajst-27534	97	31	for	for	ADP
ajst-27534	97	32	high	high	ADJ
ajst-27534	97	33	-	-	PUNCT
ajst-27534	97	34	speed	speed	NOUN
ajst-27534	97	35	nonlinear	nonlinear	ADJ
ajst-27534	97	36	circuit	circuit	NOUN
ajst-27534	97	37	macromodeling	macromodele	VERB
ajst-27534	97	38	,	,	PUNCT
ajst-27534	97	39	"	"	PUNCT
ajst-27534	97	40	in	in	ADP
ajst-27534	97	41	ieee	ieee	NOUN
ajst-27534	97	42	transactions	transaction	NOUN
ajst-27534	97	43	on	on	ADP
ajst-27534	97	44	microwave	microwave	NOUN
ajst-27534	97	45	theory	theory	NOUN
ajst-27534	97	46	and	and	CCONJ
ajst-27534	97	47	techniques	technique	NOUN
ajst-27534	97	48	,	,	PUNCT
ajst-27534	97	49	vol	vol	NOUN
ajst-27534	97	50	.	.	PROPN
ajst-27534	97	51	70	70	NUM
ajst-27534	97	52	,	,	PUNCT
ajst-27534	97	53	no	no	INTJ
ajst-27534	97	54	.	.	NOUN
ajst-27534	97	55	11	11	NUM
ajst-27534	97	56	,	,	PUNCT
ajst-27534	97	57	pp	pp	ADJ
ajst-27534	97	58	.	.	PUNCT
ajst-27534	97	59	4857	4857	NUM
ajst-27534	97	60	-	-	SYM
ajst-27534	97	61	4868	4868	NUM
ajst-27534	97	62	,	,	PUNCT
ajst-27534	97	63	nov	nov	PROPN
ajst-27534	97	64	.	.	PROPN
ajst-27534	97	65	2022	2022	NUM
ajst-27534	97	66	,	,	PUNCT
ajst-27534	97	67	doi	doi	NOUN
ajst-27534	97	68	:	:	PUNCT
ajst-27534	97	69	10.1109	10.1109	NUM
ajst-27534	97	70	/	/	SYM
ajst-27534	97	71	tmtt.2022.3200071	tmtt.2022.3200071	NOUN