id	sid	tid	token	lemma	pos
brj-22592	1	1	peer	peer	NOUN
brj-22592	1	2	-	-	PUNCT
brj-22592	1	3	review	review	NOUN
brj-22592	1	4	article	article	NOUN
brj-22592	1	5	peer	peer	NOUN
brj-22592	1	6	-	-	PUNCT
brj-22592	1	7	reviewed	review	VERB
brj-22592	1	8	article	article	NOUN
brj-22592	1	9	bioresources.com	bioresources.com	X
brj-22592	1	10	xie	xie	PROPN
brj-22592	1	11	&	&	CCONJ
brj-22592	1	12	ling	ling	PROPN
brj-22592	1	13	(	(	PUNCT
brj-22592	1	14	2023	2023	NUM
brj-22592	1	15	)	)	PUNCT
brj-22592	1	16	.	.	PUNCT
brj-22592	2	1	“	"	PUNCT
brj-22592	2	2	wood	wood	NOUN
brj-22592	2	3	defect	defect	NOUN
brj-22592	2	4	classification	classification	NOUN
brj-22592	2	5	,	,	PUNCT
brj-22592	2	6	”	"	PUNCT
brj-22592	2	7	bioresources	bioresource	NOUN
brj-22592	2	8	18(4	18(4	NUM
brj-22592	2	9	)	)	PUNCT
brj-22592	2	10	,	,	PUNCT
brj-22592	2	11	7663	7663	NUM
brj-22592	2	12	-	-	SYM
brj-22592	2	13	7680	7680	NUM
brj-22592	2	14	.	.	PUNCT
brj-22592	3	1	7663	7663	NUM
brj-22592	3	2	wood	wood	NOUN
brj-22592	3	3	defect	defect	NOUN
brj-22592	3	4	classification	classification	NOUN
brj-22592	3	5	based	base	VERB
brj-22592	3	6	on	on	ADP
brj-22592	3	7	lightweight	lightweight	ADJ
brj-22592	3	8	convolutional	convolutional	ADJ
brj-22592	3	9	neural	neural	ADJ
brj-22592	3	10	networks	network	NOUN
brj-22592	3	11	yonghua	yonghua	PROPN
brj-22592	3	12	xie	xie	PROPN
brj-22592	4	1	*	*	PUNCT
brj-22592	4	2	and	and	CCONJ
brj-22592	4	3	jiaxin	jiaxin	PROPN
brj-22592	4	4	ling	le	VERB
brj-22592	4	5	different	different	ADJ
brj-22592	4	6	types	type	NOUN
brj-22592	4	7	of	of	ADP
brj-22592	4	8	wood	wood	NOUN
brj-22592	4	9	defects	defect	NOUN
brj-22592	4	10	correspond	correspond	VERB
brj-22592	4	11	to	to	ADP
brj-22592	4	12	different	different	ADJ
brj-22592	4	13	processing	processing	NOUN
brj-22592	4	14	methods	method	NOUN
brj-22592	4	15	.	.	PUNCT
brj-22592	5	1	good	good	ADJ
brj-22592	5	2	classification	classification	NOUN
brj-22592	5	3	means	mean	VERB
brj-22592	5	4	can	can	AUX
brj-22592	5	5	transform	transform	VERB
brj-22592	5	6	defective	defective	ADJ
brj-22592	5	7	boards	board	NOUN
brj-22592	5	8	into	into	ADP
brj-22592	5	9	practical	practical	ADJ
brj-22592	5	10	boards	board	NOUN
brj-22592	5	11	after	after	ADP
brj-22592	5	12	appropriate	appropriate	ADJ
brj-22592	5	13	processing	processing	NOUN
brj-22592	5	14	.	.	PUNCT
brj-22592	6	1	the	the	DET
brj-22592	6	2	detection	detection	NOUN
brj-22592	6	3	accuracy	accuracy	NOUN
brj-22592	6	4	of	of	ADP
brj-22592	6	5	the	the	DET
brj-22592	6	6	wood	wood	NOUN
brj-22592	6	7	surface	surface	NOUN
brj-22592	6	8	defects	defect	NOUN
brj-22592	6	9	is	be	AUX
brj-22592	6	10	particularly	particularly	ADV
brj-22592	6	11	important	important	ADJ
brj-22592	6	12	for	for	ADP
brj-22592	6	13	improving	improve	VERB
brj-22592	6	14	the	the	DET
brj-22592	6	15	utilization	utilization	NOUN
brj-22592	6	16	rate	rate	NOUN
brj-22592	6	17	and	and	CCONJ
brj-22592	6	18	speed	speed	NOUN
brj-22592	6	19	of	of	ADP
brj-22592	6	20	processing	process	VERB
brj-22592	6	21	the	the	DET
brj-22592	6	22	boards	board	NOUN
brj-22592	6	23	.	.	PUNCT
brj-22592	7	1	the	the	DET
brj-22592	7	2	regnet	regnet	NOUN
brj-22592	7	3	stands	stand	VERB
brj-22592	7	4	out	out	ADP
brj-22592	7	5	in	in	ADP
brj-22592	7	6	the	the	DET
brj-22592	7	7	field	field	NOUN
brj-22592	7	8	of	of	ADP
brj-22592	7	9	computer	computer	NOUN
brj-22592	7	10	vision	vision	NOUN
brj-22592	7	11	.	.	PUNCT
brj-22592	8	1	it	it	PRON
brj-22592	8	2	automatically	automatically	ADV
brj-22592	8	3	designs	design	VERB
brj-22592	8	4	the	the	DET
brj-22592	8	5	network	network	NOUN
brj-22592	8	6	model	model	NOUN
brj-22592	8	7	based	base	VERB
brj-22592	8	8	on	on	ADP
brj-22592	8	9	the	the	DET
brj-22592	8	10	design	design	NOUN
brj-22592	8	11	space	space	NOUN
brj-22592	8	12	and	and	CCONJ
brj-22592	8	13	applies	apply	VERB
brj-22592	8	14	it	it	PRON
brj-22592	8	15	to	to	ADP
brj-22592	8	16	wood	wood	NOUN
brj-22592	8	17	defect	defect	NOUN
brj-22592	8	18	detection	detection	NOUN
brj-22592	8	19	,	,	PUNCT
brj-22592	8	20	which	which	PRON
brj-22592	8	21	can	can	AUX
brj-22592	8	22	improve	improve	VERB
brj-22592	8	23	the	the	DET
brj-22592	8	24	classification	classification	NOUN
brj-22592	8	25	accuracy	accuracy	NOUN
brj-22592	8	26	.	.	PUNCT
brj-22592	9	1	when	when	SCONJ
brj-22592	9	2	the	the	DET
brj-22592	9	3	convolutional	convolutional	ADJ
brj-22592	9	4	structure	structure	NOUN
brj-22592	9	5	of	of	ADP
brj-22592	9	6	the	the	DET
brj-22592	9	7	regnet	regnet	NOUN
brj-22592	9	8	network	network	NOUN
brj-22592	9	9	is	be	AUX
brj-22592	9	10	applied	apply	VERB
brj-22592	9	11	to	to	ADP
brj-22592	9	12	industrial	industrial	ADJ
brj-22592	9	13	detection	detection	NOUN
brj-22592	9	14	and	and	CCONJ
brj-22592	9	15	classification	classification	NOUN
brj-22592	9	16	,	,	PUNCT
brj-22592	9	17	the	the	DET
brj-22592	9	18	problems	problem	NOUN
brj-22592	9	19	of	of	ADP
brj-22592	9	20	long	long	ADJ
brj-22592	9	21	real	real	ADJ
brj-22592	9	22	-	-	PUNCT
brj-22592	9	23	time	time	NOUN
brj-22592	9	24	detection	detection	NOUN
brj-22592	9	25	time	time	NOUN
brj-22592	9	26	and	and	CCONJ
brj-22592	9	27	large	large	ADJ
brj-22592	9	28	algorithm	algorithm	NOUN
brj-22592	9	29	parameters	parameter	NOUN
brj-22592	9	30	persist	persist	VERB
brj-22592	9	31	.	.	PUNCT
brj-22592	10	1	this	this	DET
brj-22592	10	2	study	study	NOUN
brj-22592	10	3	focuses	focus	VERB
brj-22592	10	4	on	on	ADP
brj-22592	10	5	collecting	collect	VERB
brj-22592	10	6	wood	wood	NOUN
brj-22592	10	7	material	material	NOUN
brj-22592	10	8	images	image	NOUN
brj-22592	10	9	of	of	ADP
brj-22592	10	10	common	common	ADJ
brj-22592	10	11	coniferous	coniferous	ADJ
brj-22592	10	12	and	and	CCONJ
brj-22592	10	13	broad	broad	ADV
brj-22592	10	14	-	-	PUNCT
brj-22592	10	15	leaved	leave	VERB
brj-22592	10	16	trees	tree	NOUN
brj-22592	10	17	in	in	ADP
brj-22592	10	18	northeast	northeast	ADJ
brj-22592	10	19	china	china	PROPN
brj-22592	10	20	with	with	ADP
brj-22592	10	21	three	three	NUM
brj-22592	10	22	types	type	NOUN
brj-22592	10	23	of	of	ADP
brj-22592	10	24	defects	defect	NOUN
brj-22592	10	25	:	:	PUNCT
brj-22592	10	26	wormholes	wormhole	NOUN
brj-22592	10	27	,	,	PUNCT
brj-22592	10	28	slip	slip	NOUN
brj-22592	10	29	knots	knot	NOUN
brj-22592	10	30	,	,	PUNCT
brj-22592	10	31	and	and	CCONJ
brj-22592	10	32	dead	dead	ADJ
brj-22592	10	33	knots	knot	NOUN
brj-22592	10	34	.	.	PUNCT
brj-22592	11	1	to	to	PART
brj-22592	11	2	improve	improve	VERB
brj-22592	11	3	the	the	DET
brj-22592	11	4	allocation	allocation	NOUN
brj-22592	11	5	of	of	ADP
brj-22592	11	6	computing	compute	VERB
brj-22592	11	7	resources	resource	NOUN
brj-22592	11	8	,	,	PUNCT
brj-22592	11	9	based	base	VERB
brj-22592	11	10	on	on	ADP
brj-22592	11	11	the	the	DET
brj-22592	11	12	regnet	regnet	NOUN
brj-22592	11	13	network	network	NOUN
brj-22592	11	14	model	model	NOUN
brj-22592	11	15	,	,	PUNCT
brj-22592	11	16	an	an	DET
brj-22592	11	17	attention	attention	NOUN
brj-22592	11	18	mechanism	mechanism	NOUN
brj-22592	11	19	module	module	NOUN
brj-22592	11	20	was	be	AUX
brj-22592	11	21	added	add	VERB
brj-22592	11	22	,	,	PUNCT
brj-22592	11	23	and	and	CCONJ
brj-22592	11	24	the	the	DET
brj-22592	11	25	ghostconv	ghostconv	NOUN
brj-22592	11	26	structure	structure	NOUN
brj-22592	11	27	was	be	AUX
brj-22592	11	28	introduced	introduce	VERB
brj-22592	11	29	.	.	PUNCT
brj-22592	12	1	the	the	DET
brj-22592	12	2	structure	structure	NOUN
brj-22592	12	3	quickly	quickly	ADV
brj-22592	12	4	and	and	CCONJ
brj-22592	12	5	accurately	accurately	ADV
brj-22592	12	6	highlighted	highlight	VERB
brj-22592	12	7	the	the	DET
brj-22592	12	8	types	type	NOUN
brj-22592	12	9	of	of	ADP
brj-22592	12	10	wood	wood	NOUN
brj-22592	12	11	defects	defect	NOUN
brj-22592	12	12	,	,	PUNCT
brj-22592	12	13	improved	improve	VERB
brj-22592	12	14	the	the	DET
brj-22592	12	15	classification	classification	NOUN
brj-22592	12	16	accuracy	accuracy	NOUN
brj-22592	12	17	,	,	PUNCT
brj-22592	12	18	reduced	reduce	VERB
brj-22592	12	19	the	the	DET
brj-22592	12	20	parameters	parameter	NOUN
brj-22592	12	21	of	of	ADP
brj-22592	12	22	the	the	DET
brj-22592	12	23	network	network	NOUN
brj-22592	12	24	,	,	PUNCT
brj-22592	12	25	and	and	CCONJ
brj-22592	12	26	exhibited	exhibit	VERB
brj-22592	12	27	generalization	generalization	NOUN
brj-22592	12	28	ability	ability	NOUN
brj-22592	12	29	.	.	PUNCT
brj-22592	13	1	to	to	PART
brj-22592	13	2	verify	verify	VERB
brj-22592	13	3	the	the	DET
brj-22592	13	4	performance	performance	NOUN
brj-22592	13	5	of	of	ADP
brj-22592	13	6	the	the	DET
brj-22592	13	7	improved	improved	ADJ
brj-22592	13	8	network	network	NOUN
brj-22592	13	9	,	,	PUNCT
brj-22592	13	10	mobilenet	mobilenet	NOUN
brj-22592	13	11	-	-	PUNCT
brj-22592	13	12	v2	v2	NOUN
brj-22592	13	13	,	,	PUNCT
brj-22592	13	14	efficientnet	efficientnet	NOUN
brj-22592	13	15	,	,	PUNCT
brj-22592	13	16	and	and	CCONJ
brj-22592	13	17	vision	vision	NOUN
brj-22592	13	18	-	-	PUNCT
brj-22592	13	19	transformer	transformer	NOUN
brj-22592	13	20	networks	network	NOUN
brj-22592	13	21	were	be	AUX
brj-22592	13	22	introduced	introduce	VERB
brj-22592	13	23	for	for	ADP
brj-22592	13	24	comparative	comparative	ADJ
brj-22592	13	25	analysis	analysis	NOUN
brj-22592	13	26	.	.	PUNCT
brj-22592	14	1	the	the	DET
brj-22592	14	2	improved	improved	ADJ
brj-22592	14	3	regnet	regnet	NOUN
brj-22592	14	4	network	network	NOUN
brj-22592	14	5	had	have	VERB
brj-22592	14	6	smaller	small	ADJ
brj-22592	14	7	weight	weight	NOUN
brj-22592	14	8	and	and	CCONJ
brj-22592	14	9	higher	high	ADJ
brj-22592	14	10	accuracy	accuracy	NOUN
brj-22592	14	11	,	,	PUNCT
brj-22592	14	12	with	with	ADP
brj-22592	14	13	a	a	DET
brj-22592	14	14	classification	classification	NOUN
brj-22592	14	15	accuracy	accuracy	NOUN
brj-22592	14	16	of	of	ADP
brj-22592	14	17	96.58	96.58	NUM
brj-22592	14	18	%	%	NOUN
brj-22592	14	19	.	.	PUNCT
brj-22592	15	1	doi	doi	NOUN
brj-22592	15	2	:	:	PUNCT
brj-22592	15	3	10.15376	10.15376	NUM
brj-22592	15	4	/	/	SYM
brj-22592	15	5	biores.18.4.7663	biores.18.4.7663	PROPN
brj-22592	15	6	-	-	PUNCT
brj-22592	15	7	7680	7680	NUM
brj-22592	15	8	keywords	keyword	NOUN
brj-22592	15	9	:	:	PUNCT
brj-22592	15	10	board	board	NOUN
brj-22592	15	11	defect	defect	NOUN
brj-22592	15	12	;	;	PUNCT
brj-22592	15	13	neural	neural	ADJ
brj-22592	15	14	architecture	architecture	NOUN
brj-22592	15	15	search	search	NOUN
brj-22592	15	16	;	;	PUNCT
brj-22592	15	17	regnet	regnet	NOUN
brj-22592	15	18	;	;	PUNCT
brj-22592	15	19	attention	attention	NOUN
brj-22592	15	20	mechanism	mechanism	NOUN
brj-22592	15	21	contact	contact	NOUN
brj-22592	15	22	information	information	NOUN
brj-22592	15	23	:	:	PUNCT
brj-22592	15	24	college	college	NOUN
brj-22592	15	25	of	of	ADP
brj-22592	15	26	computer	computer	NOUN
brj-22592	15	27	and	and	CCONJ
brj-22592	15	28	control	control	PROPN
brj-22592	15	29	engineering	engineering	PROPN
brj-22592	15	30	,	,	PUNCT
brj-22592	15	31	northeast	northeast	ADJ
brj-22592	15	32	forestry	forestry	PROPN
brj-22592	15	33	university	university	PROPN
brj-22592	15	34	,	,	PUNCT
brj-22592	15	35	harbin	harbin	PROPN
brj-22592	15	36	,	,	PUNCT
brj-22592	15	37	150040	150040	NUM
brj-22592	15	38	,	,	PUNCT
brj-22592	15	39	people	people	NOUN
brj-22592	15	40	’s	’s	PART
brj-22592	15	41	republic	republic	NOUN
brj-22592	15	42	of	of	ADP
brj-22592	15	43	china	china	PROPN
brj-22592	15	44	;	;	PUNCT
brj-22592	15	45	*	*	PUNCT
brj-22592	15	46	corresponding	correspond	VERB
brj-22592	15	47	author	author	NOUN
brj-22592	15	48	:	:	PUNCT
brj-22592	15	49	zdhxyh@163.com	zdhxyh@163.com	PROPN
brj-22592	15	50	introduction	introduction	NOUN
brj-22592	15	51	wood	wood	NOUN
brj-22592	15	52	defects	defect	NOUN
brj-22592	15	53	refer	refer	VERB
brj-22592	15	54	to	to	ADP
brj-22592	15	55	the	the	DET
brj-22592	15	56	general	general	ADJ
brj-22592	15	57	names	name	NOUN
brj-22592	15	58	of	of	ADP
brj-22592	15	59	various	various	ADJ
brj-22592	15	60	characteristics	characteristic	NOUN
brj-22592	15	61	that	that	PRON
brj-22592	15	62	reduce	reduce	VERB
brj-22592	15	63	the	the	DET
brj-22592	15	64	commodity	commodity	NOUN
brj-22592	15	65	and	and	CCONJ
brj-22592	15	66	use	use	VERB
brj-22592	15	67	value	value	NOUN
brj-22592	15	68	of	of	ADP
brj-22592	15	69	wood	wood	NOUN
brj-22592	15	70	.	.	PUNCT
brj-22592	16	1	according	accord	VERB
brj-22592	16	2	to	to	ADP
brj-22592	16	3	the	the	DET
brj-22592	16	4	formation	formation	NOUN
brj-22592	16	5	mode	mode	NOUN
brj-22592	16	6	,	,	PUNCT
brj-22592	16	7	wood	wood	NOUN
brj-22592	16	8	defects	defect	NOUN
brj-22592	16	9	can	can	AUX
brj-22592	16	10	be	be	AUX
brj-22592	16	11	divided	divide	VERB
brj-22592	16	12	into	into	ADP
brj-22592	16	13	growth	growth	NOUN
brj-22592	16	14	,	,	PUNCT
brj-22592	16	15	biological	biological	ADJ
brj-22592	16	16	hazard	hazard	NOUN
brj-22592	16	17	,	,	PUNCT
brj-22592	16	18	and	and	CCONJ
brj-22592	16	19	processing	processing	NOUN
brj-22592	16	20	defects	defect	NOUN
brj-22592	16	21	(	(	PUNCT
brj-22592	16	22	liu	liu	PROPN
brj-22592	16	23	2008	2008	NUM
brj-22592	16	24	;	;	PUNCT
brj-22592	16	25	diao	diao	VERB
brj-22592	16	26	et	et	PROPN
brj-22592	16	27	al	al	PROPN
brj-22592	16	28	.	.	PROPN
brj-22592	16	29	2012	2012	NUM
brj-22592	16	30	;	;	PUNCT
brj-22592	16	31	luo	luo	PROPN
brj-22592	16	32	and	and	CCONJ
brj-22592	16	33	sun	sun	PROPN
brj-22592	16	34	2019	2019	NUM
brj-22592	16	35	;	;	PUNCT
brj-22592	16	36	yan	yan	PROPN
brj-22592	16	37	et	et	PROPN
brj-22592	16	38	al	al	PROPN
brj-22592	16	39	.	.	PROPN
brj-22592	16	40	2022	2022	NUM
brj-22592	16	41	)	)	PUNCT
brj-22592	16	42	.	.	PUNCT
brj-22592	17	1	different	different	ADJ
brj-22592	17	2	types	type	NOUN
brj-22592	17	3	of	of	ADP
brj-22592	17	4	board	board	NOUN
brj-22592	17	5	defects	defect	NOUN
brj-22592	17	6	have	have	VERB
brj-22592	17	7	different	different	ADJ
brj-22592	17	8	wood	wood	NOUN
brj-22592	17	9	processing	processing	NOUN
brj-22592	17	10	methods	method	NOUN
brj-22592	17	11	,	,	PUNCT
brj-22592	17	12	which	which	PRON
brj-22592	17	13	are	be	AUX
brj-22592	17	14	convenient	convenient	ADJ
brj-22592	17	15	for	for	ADP
brj-22592	17	16	the	the	DET
brj-22592	17	17	subsequent	subsequent	ADJ
brj-22592	17	18	processing	processing	NOUN
brj-22592	17	19	,	,	PUNCT
brj-22592	17	20	production	production	NOUN
brj-22592	17	21	,	,	PUNCT
brj-22592	17	22	and	and	CCONJ
brj-22592	17	23	daily	daily	ADJ
brj-22592	17	24	use	use	NOUN
brj-22592	17	25	of	of	ADP
brj-22592	17	26	wood	wood	NOUN
brj-22592	17	27	.	.	PUNCT
brj-22592	18	1	the	the	DET
brj-22592	18	2	traditional	traditional	ADJ
brj-22592	18	3	method	method	NOUN
brj-22592	18	4	of	of	ADP
brj-22592	18	5	manually	manually	ADV
brj-22592	18	6	identifying	identify	VERB
brj-22592	18	7	wood	wood	NOUN
brj-22592	18	8	defects	defect	NOUN
brj-22592	18	9	involves	involve	VERB
brj-22592	18	10	a	a	DET
brj-22592	18	11	large	large	ADJ
brj-22592	18	12	workload	workload	NOUN
brj-22592	18	13	and	and	CCONJ
brj-22592	18	14	is	be	AUX
brj-22592	18	15	time	time	NOUN
brj-22592	18	16	-	-	PUNCT
brj-22592	18	17	consuming	consume	VERB
brj-22592	18	18	,	,	PUNCT
brj-22592	18	19	which	which	PRON
brj-22592	18	20	is	be	AUX
brj-22592	18	21	a	a	DET
brj-22592	18	22	huge	huge	ADJ
brj-22592	18	23	demand	demand	NOUN
brj-22592	18	24	on	on	ADP
brj-22592	18	25	human	human	ADJ
brj-22592	18	26	and	and	CCONJ
brj-22592	18	27	material	material	ADJ
brj-22592	18	28	resources	resource	NOUN
brj-22592	18	29	.	.	PUNCT
brj-22592	19	1	with	with	ADP
brj-22592	19	2	improvements	improvement	NOUN
brj-22592	19	3	in	in	ADP
brj-22592	19	4	computing	compute	VERB
brj-22592	19	5	technology	technology	NOUN
brj-22592	19	6	,	,	PUNCT
brj-22592	19	7	convolutional	convolutional	ADJ
brj-22592	19	8	neural	neural	ADJ
brj-22592	19	9	network	network	NOUN
brj-22592	19	10	provides	provide	VERB
brj-22592	19	11	convenient	convenient	ADJ
brj-22592	19	12	conditions	condition	NOUN
brj-22592	19	13	for	for	ADP
brj-22592	19	14	image	image	NOUN
brj-22592	19	15	recognition	recognition	NOUN
brj-22592	19	16	and	and	CCONJ
brj-22592	19	17	classification	classification	NOUN
brj-22592	19	18	.	.	PUNCT
brj-22592	20	1	many	many	ADJ
brj-22592	20	2	excellent	excellent	ADJ
brj-22592	20	3	algorithm	algorithm	NOUN
brj-22592	20	4	models	model	NOUN
brj-22592	20	5	have	have	AUX
brj-22592	20	6	significantly	significantly	ADV
brj-22592	20	7	progressed	progress	VERB
brj-22592	20	8	in	in	ADP
brj-22592	20	9	visual	visual	ADJ
brj-22592	20	10	recognition	recognition	NOUN
brj-22592	20	11	.	.	PUNCT
brj-22592	21	1	zhang	zhang	PROPN
brj-22592	21	2	et	et	PROPN
brj-22592	21	3	al	al	PROPN
brj-22592	21	4	.	.	PROPN
brj-22592	21	5	(	(	PUNCT
brj-22592	21	6	2021	2021	NUM
brj-22592	21	7	)	)	PUNCT
brj-22592	21	8	designed	design	VERB
brj-22592	21	9	an	an	DET
brj-22592	21	10	improved	improved	ADJ
brj-22592	21	11	lenet-5	lenet-5	NUM
brj-22592	21	12	model	model	NOUN
brj-22592	21	13	that	that	PRON
brj-22592	21	14	can	can	AUX
brj-22592	21	15	accurately	accurately	ADV
brj-22592	21	16	classify	classify	VERB
brj-22592	21	17	common	common	ADJ
brj-22592	21	18	defects	defect	NOUN
brj-22592	21	19	and	and	CCONJ
brj-22592	21	20	defect	defect	NOUN
brj-22592	21	21	-	-	PUNCT
brj-22592	21	22	free	free	ADJ
brj-22592	21	23	samples	sample	NOUN
brj-22592	21	24	of	of	ADP
brj-22592	21	25	wood	wood	NOUN
brj-22592	21	26	by	by	ADP
brj-22592	21	27	increasing	increase	VERB
brj-22592	21	28	the	the	DET
brj-22592	21	29	network	network	NOUN
brj-22592	21	30	depth	depth	NOUN
brj-22592	21	31	and	and	CCONJ
brj-22592	21	32	batch	batch	NOUN
brj-22592	21	33	normalization	normalization	NOUN
brj-22592	21	34	.	.	PUNCT
brj-22592	22	1	shi	shi	PROPN
brj-22592	22	2	peer	peer	NOUN
brj-22592	22	3	-	-	PUNCT
brj-22592	22	4	reviewed	review	VERB
brj-22592	22	5	article	article	NOUN
brj-22592	22	6	bioresources.com	bioresources.com	X
brj-22592	22	7	xie	xie	PROPN
brj-22592	22	8	&	&	CCONJ
brj-22592	22	9	ling	ling	PROPN
brj-22592	22	10	(	(	PUNCT
brj-22592	22	11	2023	2023	NUM
brj-22592	22	12	)	)	PUNCT
brj-22592	22	13	.	.	PUNCT
brj-22592	23	1	“	"	PUNCT
brj-22592	23	2	wood	wood	NOUN
brj-22592	23	3	defect	defect	NOUN
brj-22592	23	4	classification	classification	NOUN
brj-22592	23	5	,	,	PUNCT
brj-22592	23	6	”	"	PUNCT
brj-22592	23	7	bioresources	bioresource	NOUN
brj-22592	23	8	18(4	18(4	NUM
brj-22592	23	9	)	)	PUNCT
brj-22592	23	10	,	,	PUNCT
brj-22592	23	11	7663	7663	NUM
brj-22592	23	12	-	-	SYM
brj-22592	23	13	7680	7680	NUM
brj-22592	23	14	.	.	PUNCT
brj-22592	24	1	7664	7664	NUM
brj-22592	24	2	et	et	PROPN
brj-22592	24	3	al	al	PROPN
brj-22592	24	4	.	.	PROPN
brj-22592	25	1	(	(	PUNCT
brj-22592	25	2	2020	2020	NUM
brj-22592	25	3	)	)	PUNCT
brj-22592	25	4	proposed	propose	VERB
brj-22592	25	5	a	a	DET
brj-22592	25	6	method	method	NOUN
brj-22592	25	7	combining	combine	VERB
brj-22592	25	8	adaptive	adaptive	ADJ
brj-22592	25	9	component	component	NOUN
brj-22592	25	10	analysis	analysis	NOUN
brj-22592	25	11	and	and	CCONJ
brj-22592	25	12	deep	deep	ADJ
brj-22592	25	13	migration	migration	NOUN
brj-22592	25	14	feed	feed	VERB
brj-22592	25	15	forward	forward	ADV
brj-22592	25	16	neural	neural	ADJ
brj-22592	25	17	network	network	NOUN
brj-22592	25	18	to	to	PART
brj-22592	25	19	effectively	effectively	ADV
brj-22592	25	20	migrate	migrate	VERB
brj-22592	25	21	the	the	DET
brj-22592	25	22	spectrum	spectrum	NOUN
brj-22592	25	23	and	and	CCONJ
brj-22592	25	24	defect	defect	VERB
brj-22592	25	25	corresponding	corresponding	ADJ
brj-22592	25	26	knowledge	knowledge	NOUN
brj-22592	25	27	of	of	ADP
brj-22592	25	28	other	other	ADJ
brj-22592	25	29	tree	tree	NOUN
brj-22592	25	30	species	specie	NOUN
brj-22592	25	31	to	to	ADP
brj-22592	25	32	the	the	DET
brj-22592	25	33	target	target	NOUN
brj-22592	25	34	classifier	classifier	NOUN
brj-22592	25	35	and	and	CCONJ
brj-22592	25	36	improve	improve	VERB
brj-22592	25	37	the	the	DET
brj-22592	25	38	performance	performance	NOUN
brj-22592	25	39	of	of	ADP
brj-22592	25	40	the	the	DET
brj-22592	25	41	classifier	classifier	NOUN
brj-22592	25	42	.	.	PUNCT
brj-22592	26	1	riana	riana	PROPN
brj-22592	26	2	et	et	PROPN
brj-22592	26	3	al	al	PROPN
brj-22592	26	4	.	.	PROPN
brj-22592	27	1	(	(	PUNCT
brj-22592	27	2	2021	2021	NUM
brj-22592	27	3	)	)	PUNCT
brj-22592	27	4	used	use	VERB
brj-22592	27	5	k	k	NOUN
brj-22592	27	6	-	-	PUNCT
brj-22592	27	7	means	means	NOUN
brj-22592	27	8	to	to	PART
brj-22592	27	9	segment	segment	VERB
brj-22592	27	10	wood	wood	NOUN
brj-22592	27	11	defect	defect	NOUN
brj-22592	27	12	images	image	NOUN
brj-22592	27	13	,	,	PUNCT
brj-22592	27	14	extracted	extract	VERB
brj-22592	27	15	six	six	NUM
brj-22592	27	16	texture	texture	NOUN
brj-22592	27	17	and	and	CCONJ
brj-22592	27	18	shape	shape	NOUN
brj-22592	27	19	features	feature	NOUN
brj-22592	27	20	using	use	VERB
brj-22592	27	21	the	the	DET
brj-22592	27	22	gray	gray	ADJ
brj-22592	27	23	level	level	NOUN
brj-22592	27	24	co	co	NOUN
brj-22592	27	25	-	-	NOUN
brj-22592	27	26	occurrence	occurrence	ADJ
brj-22592	27	27	matrix	matrix	NOUN
brj-22592	27	28	(	(	PUNCT
brj-22592	27	29	glcm	glcm	PROPN
brj-22592	27	30	)	)	PUNCT
brj-22592	27	31	method	method	NOUN
brj-22592	27	32	,	,	PUNCT
brj-22592	27	33	and	and	CCONJ
brj-22592	27	34	calculated	calculate	VERB
brj-22592	27	35	the	the	DET
brj-22592	27	36	shortest	short	ADJ
brj-22592	27	37	distance	distance	NOUN
brj-22592	27	38	between	between	ADP
brj-22592	27	39	the	the	DET
brj-22592	27	40	feature	feature	NOUN
brj-22592	27	41	values	value	NOUN
brj-22592	27	42	of	of	ADP
brj-22592	27	43	the	the	DET
brj-22592	27	44	test	test	NOUN
brj-22592	27	45	set	set	VERB
brj-22592	27	46	and	and	CCONJ
brj-22592	27	47	the	the	DET
brj-22592	27	48	training	training	NOUN
brj-22592	27	49	set	set	NOUN
brj-22592	27	50	using	use	VERB
brj-22592	27	51	the	the	DET
brj-22592	27	52	euclidean	euclidean	ADJ
brj-22592	27	53	distance	distance	NOUN
brj-22592	27	54	method	method	NOUN
brj-22592	27	55	to	to	PART
brj-22592	27	56	achieve	achieve	VERB
brj-22592	27	57	the	the	DET
brj-22592	27	58	purpose	purpose	NOUN
brj-22592	27	59	of	of	ADP
brj-22592	27	60	identifying	identify	VERB
brj-22592	27	61	wood	wood	NOUN
brj-22592	27	62	defects	defect	NOUN
brj-22592	27	63	(	(	PUNCT
brj-22592	27	64	riana	riana	PROPN
brj-22592	27	65	et	et	PROPN
brj-22592	27	66	al	al	PROPN
brj-22592	27	67	.	.	PROPN
brj-22592	27	68	2021	2021	NUM
brj-22592	27	69	)	)	PUNCT
brj-22592	27	70	.	.	PUNCT
brj-22592	28	1	at	at	ADP
brj-22592	28	2	the	the	DET
brj-22592	28	3	2017	2017	NUM
brj-22592	28	4	iclr	iclr	NOUN
brj-22592	28	5	conference	conference	NOUN
brj-22592	28	6	,	,	PUNCT
brj-22592	28	7	barret	barret	NOUN
brj-22592	28	8	zoph	zoph	NOUN
brj-22592	28	9	and	and	CCONJ
brj-22592	28	10	his	his	PRON
brj-22592	28	11	research	research	NOUN
brj-22592	28	12	team	team	NOUN
brj-22592	28	13	presented	present	VERB
brj-22592	28	14	the	the	DET
brj-22592	28	15	nas	nas	NOUN
brj-22592	28	16	algorithm	algorithm	NOUN
brj-22592	28	17	for	for	ADP
brj-22592	28	18	the	the	DET
brj-22592	28	19	first	first	ADJ
brj-22592	28	20	time	time	NOUN
brj-22592	28	21	,	,	PUNCT
brj-22592	28	22	using	use	VERB
brj-22592	28	23	the	the	DET
brj-22592	28	24	controller	controller	NOUN
brj-22592	28	25	rnn	rnn	VERB
brj-22592	28	26	model	model	NOUN
brj-22592	28	27	to	to	PART
brj-22592	28	28	create	create	VERB
brj-22592	28	29	a	a	DET
brj-22592	28	30	string	string	NOUN
brj-22592	28	31	describing	describe	VERB
brj-22592	28	32	the	the	DET
brj-22592	28	33	network	network	NOUN
brj-22592	28	34	structure	structure	NOUN
brj-22592	28	35	,	,	PUNCT
brj-22592	28	36	and	and	CCONJ
brj-22592	28	37	then	then	ADV
brj-22592	28	38	using	use	VERB
brj-22592	28	39	the	the	DET
brj-22592	28	40	policy	policy	NOUN
brj-22592	28	41	gradient	gradient	NOUN
brj-22592	28	42	algorithm	algorithm	NOUN
brj-22592	28	43	to	to	PART
brj-22592	28	44	update	update	VERB
brj-22592	28	45	the	the	DET
brj-22592	28	46	control	control	NOUN
brj-22592	28	47	variables	variable	NOUN
brj-22592	28	48	to	to	PART
brj-22592	28	49	maximize	maximize	VERB
brj-22592	28	50	the	the	DET
brj-22592	28	51	accuracy	accuracy	NOUN
brj-22592	28	52	of	of	ADP
brj-22592	28	53	the	the	DET
brj-22592	28	54	created	create	VERB
brj-22592	28	55	network	network	NOUN
brj-22592	28	56	(	(	PUNCT
brj-22592	28	57	zoph	zoph	NOUN
brj-22592	28	58	and	and	CCONJ
brj-22592	28	59	le	le	ADP
brj-22592	28	60	2016	2016	NUM
brj-22592	28	61	)	)	PUNCT
brj-22592	28	62	.	.	PUNCT
brj-22592	29	1	after	after	ADP
brj-22592	29	2	experimental	experimental	ADJ
brj-22592	29	3	verification	verification	NOUN
brj-22592	29	4	,	,	PUNCT
brj-22592	29	5	in	in	ADP
brj-22592	29	6	some	some	DET
brj-22592	29	7	fields	field	NOUN
brj-22592	29	8	of	of	ADP
brj-22592	29	9	artificial	artificial	ADJ
brj-22592	29	10	intelligence	intelligence	NOUN
brj-22592	29	11	,	,	PUNCT
brj-22592	29	12	such	such	ADJ
brj-22592	29	13	as	as	ADP
brj-22592	29	14	image	image	NOUN
brj-22592	29	15	classification	classification	NOUN
brj-22592	29	16	(	(	PUNCT
brj-22592	29	17	zhang	zhang	PROPN
brj-22592	29	18	et	et	PROPN
brj-22592	29	19	al	al	PROPN
brj-22592	29	20	.	.	PROPN
brj-22592	29	21	2021	2021	NUM
brj-22592	29	22	)	)	PUNCT
brj-22592	29	23	target	target	NOUN
brj-22592	29	24	detection	detection	NOUN
brj-22592	29	25	(	(	PUNCT
brj-22592	29	26	wang	wang	PROPN
brj-22592	29	27	et	et	PROPN
brj-22592	29	28	al	al	PROPN
brj-22592	29	29	.	.	PROPN
brj-22592	29	30	2020	2020	NUM
brj-22592	29	31	)	)	PUNCT
brj-22592	29	32	,	,	PUNCT
brj-22592	29	33	semantic	semantic	ADJ
brj-22592	29	34	segmentation	segmentation	NOUN
brj-22592	29	35	(	(	PUNCT
brj-22592	29	36	hu	hu	PROPN
brj-22592	29	37	et	et	PROPN
brj-22592	29	38	al	al	PROPN
brj-22592	29	39	.	.	PROPN
brj-22592	29	40	2022	2022	NUM
brj-22592	29	41	)	)	PUNCT
brj-22592	29	42	,	,	PUNCT
brj-22592	29	43	and	and	CCONJ
brj-22592	29	44	other	other	ADJ
brj-22592	29	45	fields	field	NOUN
brj-22592	29	46	,	,	PUNCT
brj-22592	29	47	guo	guo	PROPN
brj-22592	29	48	et	et	PROPN
brj-22592	29	49	al	al	PROPN
brj-22592	29	50	.	.	PROPN
brj-22592	29	51	(	(	PUNCT
brj-22592	29	52	2020	2020	NUM
brj-22592	29	53	)	)	PUNCT
brj-22592	29	54	proposed	propose	VERB
brj-22592	29	55	a	a	DET
brj-22592	29	56	multichannel	multichannel	ADJ
brj-22592	29	57	image	image	NOUN
brj-22592	29	58	segmentation	segmentation	NOUN
brj-22592	29	59	method	method	NOUN
brj-22592	29	60	for	for	ADP
brj-22592	29	61	wood	wood	NOUN
brj-22592	29	62	knot	knot	NOUN
brj-22592	29	63	defects	defect	NOUN
brj-22592	29	64	based	base	VERB
brj-22592	29	65	on	on	ADP
brj-22592	29	66	the	the	DET
brj-22592	29	67	tvcv	tvcv	NOUN
brj-22592	29	68	model	model	NOUN
brj-22592	29	69	.	.	PUNCT
brj-22592	30	1	this	this	DET
brj-22592	30	2	method	method	NOUN
brj-22592	30	3	controls	control	VERB
brj-22592	30	4	the	the	DET
brj-22592	30	5	range	range	NOUN
brj-22592	30	6	of	of	ADP
brj-22592	30	7	the	the	DET
brj-22592	30	8	active	active	ADJ
brj-22592	30	9	contour	contour	NOUN
brj-22592	30	10	via	via	ADP
brj-22592	30	11	regularization	regularization	NOUN
brj-22592	30	12	parameters	parameter	NOUN
brj-22592	30	13	to	to	PART
brj-22592	30	14	capture	capture	VERB
brj-22592	30	15	the	the	DET
brj-22592	30	16	object	object	NOUN
brj-22592	30	17	.	.	PUNCT
brj-22592	31	1	furthermore	furthermore	ADV
brj-22592	31	2	,	,	PUNCT
brj-22592	31	3	it	it	PRON
brj-22592	31	4	uses	use	VERB
brj-22592	31	5	the	the	DET
brj-22592	31	6	multi	multi	ADJ
brj-22592	31	7	-	-	ADJ
brj-22592	31	8	channel	channel	ADJ
brj-22592	31	9	weighted	weight	VERB
brj-22592	31	10	level	level	NOUN
brj-22592	31	11	set	set	VERB
brj-22592	31	12	as	as	ADP
brj-22592	31	13	the	the	DET
brj-22592	31	14	initial	initial	ADJ
brj-22592	31	15	value	value	NOUN
brj-22592	31	16	,	,	PUNCT
brj-22592	31	17	iterates	iterate	VERB
brj-22592	31	18	with	with	ADP
brj-22592	31	19	the	the	DET
brj-22592	31	20	tvcv	tvcv	NOUN
brj-22592	31	21	model	model	NOUN
brj-22592	31	22	,	,	PUNCT
brj-22592	31	23	and	and	CCONJ
brj-22592	31	24	finally	finally	ADV
brj-22592	31	25	considers	consider	VERB
brj-22592	31	26	the	the	DET
brj-22592	31	27	calculated	calculated	ADJ
brj-22592	31	28	value	value	NOUN
brj-22592	31	29	as	as	SCONJ
brj-22592	31	30	the	the	DET
brj-22592	31	31	segmentation	segmentation	NOUN
brj-22592	31	32	result	result	VERB
brj-22592	31	33	.	.	PUNCT
brj-22592	32	1	the	the	DET
brj-22592	32	2	segmentation	segmentation	NOUN
brj-22592	32	3	effect	effect	NOUN
brj-22592	32	4	of	of	ADP
brj-22592	32	5	the	the	DET
brj-22592	32	6	model	model	NOUN
brj-22592	32	7	is	be	AUX
brj-22592	32	8	better	well	ADJ
brj-22592	32	9	.	.	PUNCT
brj-22592	33	1	the	the	DET
brj-22592	33	2	aforementioned	aforementioned	ADJ
brj-22592	33	3	model	model	NOUN
brj-22592	33	4	is	be	AUX
brj-22592	33	5	designed	design	VERB
brj-22592	33	6	manually	manually	ADV
brj-22592	33	7	,	,	PUNCT
brj-22592	33	8	and	and	CCONJ
brj-22592	33	9	the	the	DET
brj-22592	33	10	design	design	NOUN
brj-22592	33	11	of	of	ADP
brj-22592	33	12	the	the	DET
brj-22592	33	13	network	network	NOUN
brj-22592	33	14	model	model	NOUN
brj-22592	33	15	is	be	AUX
brj-22592	33	16	mainly	mainly	ADV
brj-22592	33	17	reflected	reflect	VERB
brj-22592	33	18	in	in	ADP
brj-22592	33	19	the	the	DET
brj-22592	33	20	application	application	NOUN
brj-22592	33	21	of	of	ADP
brj-22592	33	22	network	network	NOUN
brj-22592	33	23	layer	layer	NOUN
brj-22592	33	24	operators	operator	NOUN
brj-22592	33	25	.	.	PUNCT
brj-22592	34	1	furthermore	furthermore	ADV
brj-22592	34	2	,	,	PUNCT
brj-22592	34	3	the	the	DET
brj-22592	34	4	performance	performance	NOUN
brj-22592	34	5	optimization	optimization	NOUN
brj-22592	34	6	process	process	NOUN
brj-22592	34	7	of	of	ADP
brj-22592	34	8	the	the	DET
brj-22592	34	9	model	model	NOUN
brj-22592	34	10	is	be	AUX
brj-22592	34	11	accidental	accidental	ADJ
brj-22592	34	12	and	and	CCONJ
brj-22592	34	13	uncertain	uncertain	ADJ
brj-22592	34	14	.	.	PUNCT
brj-22592	35	1	network	network	NOUN
brj-22592	35	2	design	design	NOUN
brj-22592	35	3	has	have	AUX
brj-22592	35	4	changed	change	VERB
brj-22592	35	5	from	from	ADP
brj-22592	35	6	manual	manual	ADJ
brj-22592	35	7	exploration	exploration	NOUN
brj-22592	35	8	process	process	NOUN
brj-22592	35	9	to	to	ADP
brj-22592	35	10	automatic	automatic	ADJ
brj-22592	35	11	network	network	NOUN
brj-22592	35	12	design	design	NOUN
brj-22592	35	13	,	,	PUNCT
brj-22592	35	14	and	and	CCONJ
brj-22592	35	15	several	several	ADJ
brj-22592	35	16	automatic	automatic	ADJ
brj-22592	35	17	search	search	NOUN
brj-22592	35	18	network	network	NOUN
brj-22592	35	19	structure	structure	NOUN
brj-22592	35	20	models	model	NOUN
brj-22592	35	21	have	have	AUX
brj-22592	35	22	been	be	AUX
brj-22592	35	23	developed	develop	VERB
brj-22592	35	24	.	.	PUNCT
brj-22592	36	1	for	for	ADP
brj-22592	36	2	example	example	NOUN
brj-22592	36	3	,	,	PUNCT
brj-22592	36	4	gong	gong	PROPN
brj-22592	36	5	et	et	PROPN
brj-22592	36	6	al	al	PROPN
brj-22592	36	7	.	.	PROPN
brj-22592	37	1	(	(	PUNCT
brj-22592	37	2	2019	2019	NUM
brj-22592	37	3	)	)	PUNCT
brj-22592	37	4	combined	combine	VERB
brj-22592	37	5	a	a	DET
brj-22592	37	6	neural	neural	ADJ
brj-22592	37	7	architecture	architecture	NOUN
brj-22592	37	8	search	search	NOUN
brj-22592	37	9	(	(	PUNCT
brj-22592	37	10	nas	nas	PROPN
brj-22592	37	11	)	)	PUNCT
brj-22592	37	12	with	with	ADP
brj-22592	37	13	a	a	DET
brj-22592	37	14	generative	generative	ADJ
brj-22592	37	15	adversarial	adversarial	ADJ
brj-22592	37	16	network	network	NOUN
brj-22592	37	17	(	(	PUNCT
brj-22592	37	18	gan	gan	PROPN
brj-22592	37	19	)	)	PUNCT
brj-22592	37	20	,	,	PUNCT
brj-22592	37	21	defined	define	VERB
brj-22592	37	22	the	the	DET
brj-22592	37	23	search	search	NOUN
brj-22592	37	24	space	space	NOUN
brj-22592	37	25	of	of	ADP
brj-22592	37	26	gan	gan	ADJ
brj-22592	37	27	network	network	NOUN
brj-22592	37	28	structure	structure	NOUN
brj-22592	37	29	change	change	NOUN
brj-22592	37	30	,	,	PUNCT
brj-22592	37	31	used	use	VERB
brj-22592	37	32	an	an	DET
brj-22592	37	33	rnn	rnn	NOUN
brj-22592	37	34	controller	controller	NOUN
brj-22592	37	35	to	to	PART
brj-22592	37	36	guide	guide	VERB
brj-22592	37	37	structure	structure	NOUN
brj-22592	37	38	search	search	NOUN
brj-22592	37	39	,	,	PUNCT
brj-22592	37	40	and	and	CCONJ
brj-22592	37	41	combined	combine	VERB
brj-22592	37	42	parameter	parameter	NOUN
brj-22592	37	43	sharing	sharing	NOUN
brj-22592	37	44	and	and	CCONJ
brj-22592	37	45	dynamic	dynamic	ADJ
brj-22592	37	46	reset	reset	NOUN
brj-22592	37	47	strategy	strategy	NOUN
brj-22592	37	48	to	to	PART
brj-22592	37	49	improve	improve	VERB
brj-22592	37	50	training	training	NOUN
brj-22592	37	51	speed	speed	NOUN
brj-22592	37	52	.	.	PUNCT
brj-22592	38	1	sun	sun	PROPN
brj-22592	38	2	et	et	PROPN
brj-22592	38	3	al	al	PROPN
brj-22592	38	4	.	.	PROPN
brj-22592	38	5	(	(	PUNCT
brj-22592	38	6	2020	2020	NUM
brj-22592	38	7	)	)	PUNCT
brj-22592	38	8	proposed	propose	VERB
brj-22592	38	9	a	a	DET
brj-22592	38	10	method	method	NOUN
brj-22592	38	11	to	to	PART
brj-22592	38	12	solve	solve	VERB
brj-22592	38	13	the	the	DET
brj-22592	38	14	image	image	NOUN
brj-22592	38	15	classification	classification	NOUN
brj-22592	38	16	problem	problem	NOUN
brj-22592	38	17	by	by	ADP
brj-22592	38	18	using	use	VERB
brj-22592	38	19	a	a	DET
brj-22592	38	20	genetic	genetic	ADJ
brj-22592	38	21	algorithm	algorithm	NOUN
brj-22592	38	22	to	to	PART
brj-22592	38	23	evolve	evolve	VERB
brj-22592	38	24	the	the	DET
brj-22592	38	25	structure	structure	NOUN
brj-22592	38	26	and	and	CCONJ
brj-22592	38	27	connection	connection	NOUN
brj-22592	38	28	weight	weight	NOUN
brj-22592	38	29	initialization	initialization	NOUN
brj-22592	38	30	value	value	NOUN
brj-22592	38	31	of	of	ADP
brj-22592	38	32	a	a	DET
brj-22592	38	33	deep	deep	ADJ
brj-22592	38	34	convolution	convolution	NOUN
brj-22592	38	35	neural	neural	ADJ
brj-22592	38	36	network	network	NOUN
brj-22592	38	37	.	.	PUNCT
brj-22592	39	1	in	in	ADP
brj-22592	39	2	this	this	DET
brj-22592	39	3	algorithm	algorithm	NOUN
brj-22592	39	4	,	,	PUNCT
brj-22592	39	5	an	an	DET
brj-22592	39	6	effective	effective	ADJ
brj-22592	39	7	variable	variable	ADJ
brj-22592	39	8	length	length	NOUN
brj-22592	39	9	gene	gene	NOUN
brj-22592	39	10	coding	code	VERB
brj-22592	39	11	strategy	strategy	NOUN
brj-22592	39	12	is	be	AUX
brj-22592	39	13	designed	design	VERB
brj-22592	39	14	to	to	PART
brj-22592	39	15	represent	represent	VERB
brj-22592	39	16	the	the	DET
brj-22592	39	17	different	different	ADJ
brj-22592	39	18	building	building	NOUN
brj-22592	39	19	blocks	block	NOUN
brj-22592	39	20	and	and	CCONJ
brj-22592	39	21	potential	potential	ADJ
brj-22592	39	22	optimal	optimal	ADJ
brj-22592	39	23	depth	depth	NOUN
brj-22592	39	24	of	of	ADP
brj-22592	39	25	convolution	convolution	NOUN
brj-22592	39	26	neural	neural	ADJ
brj-22592	39	27	network	network	NOUN
brj-22592	39	28	.	.	PUNCT
brj-22592	40	1	li	li	PROPN
brj-22592	40	2	et	et	PROPN
brj-22592	40	3	al	al	PROPN
brj-22592	40	4	.	.	PROPN
brj-22592	40	5	(	(	PUNCT
brj-22592	40	6	2020	2020	NUM
brj-22592	40	7	)	)	PUNCT
brj-22592	40	8	proposed	propose	VERB
brj-22592	40	9	a	a	DET
brj-22592	40	10	sequential	sequential	ADJ
brj-22592	40	11	greedy	greedy	ADJ
brj-22592	40	12	architecture	architecture	NOUN
brj-22592	40	13	search	search	NOUN
brj-22592	40	14	(	(	PUNCT
brj-22592	40	15	sgas	sga	NOUN
brj-22592	40	16	)	)	PUNCT
brj-22592	40	17	algorithm	algorithm	NOUN
brj-22592	40	18	to	to	PART
brj-22592	40	19	automatically	automatically	ADV
brj-22592	40	20	design	design	VERB
brj-22592	40	21	the	the	DET
brj-22592	40	22	architecture	architecture	NOUN
brj-22592	40	23	of	of	ADP
brj-22592	40	24	cnn	cnn	PROPN
brj-22592	40	25	and	and	CCONJ
brj-22592	40	26	gcn	gcn	NOUN
brj-22592	40	27	.	.	PUNCT
brj-22592	41	1	by	by	ADP
brj-22592	41	2	considering	consider	VERB
brj-22592	41	3	the	the	DET
brj-22592	41	4	heuristic	heuristic	ADJ
brj-22592	41	5	criteria	criterion	NOUN
brj-22592	41	6	of	of	ADP
brj-22592	41	7	edge	edge	NOUN
brj-22592	41	8	importance	importance	NOUN
brj-22592	41	9	,	,	PUNCT
brj-22592	41	10	selection	selection	NOUN
brj-22592	41	11	certainty	certainty	NOUN
brj-22592	41	12	and	and	CCONJ
brj-22592	41	13	selection	selection	NOUN
brj-22592	41	14	stability	stability	NOUN
brj-22592	41	15	,	,	PUNCT
brj-22592	41	16	they	they	PRON
brj-22592	41	17	solved	solve	VERB
brj-22592	41	18	the	the	DET
brj-22592	41	19	two	two	NUM
brj-22592	41	20	-	-	PUNCT
brj-22592	41	21	level	level	NOUN
brj-22592	41	22	optimization	optimization	NOUN
brj-22592	41	23	problem	problem	NOUN
brj-22592	41	24	in	in	ADP
brj-22592	41	25	nas	nas	NOUN
brj-22592	41	26	in	in	ADP
brj-22592	41	27	a	a	DET
brj-22592	41	28	greedy	greedy	ADJ
brj-22592	41	29	manner	manner	NOUN
brj-22592	41	30	.	.	PUNCT
brj-22592	42	1	although	although	SCONJ
brj-22592	42	2	the	the	DET
brj-22592	42	3	aforementioned	aforementioned	ADJ
brj-22592	42	4	method	method	NOUN
brj-22592	42	5	uses	use	VERB
brj-22592	42	6	an	an	DET
brj-22592	42	7	automatic	automatic	ADJ
brj-22592	42	8	design	design	NOUN
brj-22592	42	9	network	network	NOUN
brj-22592	42	10	algorithm	algorithm	NOUN
brj-22592	42	11	,	,	PUNCT
brj-22592	42	12	it	it	PRON
brj-22592	42	13	still	still	ADV
brj-22592	42	14	has	have	VERB
brj-22592	42	15	the	the	DET
brj-22592	42	16	disadvantage	disadvantage	NOUN
brj-22592	42	17	of	of	ADP
brj-22592	42	18	relying	rely	VERB
brj-22592	42	19	on	on	ADP
brj-22592	42	20	a	a	DET
brj-22592	42	21	manually	manually	ADV
brj-22592	42	22	designed	design	VERB
brj-22592	42	23	network	network	NOUN
brj-22592	42	24	model	model	NOUN
brj-22592	42	25	structure	structure	NOUN
brj-22592	42	26	.	.	PUNCT
brj-22592	43	1	the	the	DET
brj-22592	43	2	regnet	regnet	PROPN
brj-22592	43	3	network	network	NOUN
brj-22592	43	4	is	be	AUX
brj-22592	43	5	an	an	DET
brj-22592	43	6	automatic	automatic	ADJ
brj-22592	43	7	search	search	NOUN
brj-22592	43	8	design	design	NOUN
brj-22592	43	9	model	model	NOUN
brj-22592	43	10	proposed	propose	VERB
brj-22592	43	11	by	by	ADP
brj-22592	43	12	kaiming	kaime	VERB
brj-22592	43	13	he	he	PRON
brj-22592	43	14	(	(	PUNCT
brj-22592	43	15	radosavovic	radosavovic	PROPN
brj-22592	43	16	et	et	PROPN
brj-22592	43	17	al	al	PROPN
brj-22592	43	18	.	.	PROPN
brj-22592	43	19	2020	2020	NUM
brj-22592	43	20	)	)	PUNCT
brj-22592	43	21	.	.	PUNCT
brj-22592	44	1	the	the	DET
brj-22592	44	2	network	network	NOUN
brj-22592	44	3	and	and	CCONJ
brj-22592	44	4	its	its	PRON
brj-22592	44	5	related	related	ADJ
brj-22592	44	6	network	network	NOUN
brj-22592	44	7	models	model	NOUN
brj-22592	44	8	exhibited	exhibit	VERB
brj-22592	44	9	good	good	ADJ
brj-22592	44	10	results	result	NOUN
brj-22592	44	11	in	in	ADP
brj-22592	44	12	the	the	DET
brj-22592	44	13	field	field	NOUN
brj-22592	44	14	of	of	ADP
brj-22592	44	15	computer	computer	NOUN
brj-22592	44	16	vision	vision	NOUN
brj-22592	44	17	,	,	PUNCT
brj-22592	44	18	which	which	PRON
brj-22592	44	19	can	can	AUX
brj-22592	44	20	automatically	automatically	ADV
brj-22592	44	21	design	design	VERB
brj-22592	44	22	the	the	DET
brj-22592	44	23	network	network	NOUN
brj-22592	44	24	framework	framework	NOUN
brj-22592	44	25	in	in	ADP
brj-22592	44	26	a	a	DET
brj-22592	44	27	design	design	NOUN
brj-22592	44	28	space	space	NOUN
brj-22592	44	29	to	to	PART
brj-22592	44	30	simplify	simplify	VERB
brj-22592	44	31	the	the	DET
brj-22592	44	32	task	task	NOUN
brj-22592	44	33	of	of	ADP
brj-22592	44	34	deep	deep	ADJ
brj-22592	44	35	learning	learning	NOUN
brj-22592	44	36	.	.	PUNCT
brj-22592	45	1	however	however	ADV
brj-22592	45	2	,	,	PUNCT
brj-22592	45	3	the	the	DET
brj-22592	45	4	application	application	NOUN
brj-22592	45	5	of	of	ADP
brj-22592	45	6	rapid	rapid	ADJ
brj-22592	45	7	detection	detection	NOUN
brj-22592	45	8	of	of	ADP
brj-22592	45	9	wood	wood	NOUN
brj-22592	45	10	defects	defect	NOUN
brj-22592	45	11	still	still	ADV
brj-22592	45	12	persists	persist	VERB
brj-22592	45	13	with	with	ADP
brj-22592	45	14	the	the	DET
brj-22592	45	15	problem	problem	NOUN
brj-22592	45	16	of	of	ADP
brj-22592	45	17	long	long	ADJ
brj-22592	45	18	real	real	ADJ
brj-22592	45	19	-	-	PUNCT
brj-22592	45	20	time	time	NOUN
brj-22592	45	21	detection	detection	NOUN
brj-22592	45	22	time	time	NOUN
brj-22592	45	23	.	.	PUNCT
brj-22592	46	1	in	in	ADP
brj-22592	46	2	this	this	DET
brj-22592	46	3	study	study	NOUN
brj-22592	46	4	,	,	PUNCT
brj-22592	46	5	the	the	DET
brj-22592	46	6	goal	goal	NOUN
brj-22592	46	7	was	be	AUX
brj-22592	46	8	to	to	PART
brj-22592	46	9	use	use	VERB
brj-22592	46	10	the	the	DET
brj-22592	46	11	regnet	regnet	NOUN
brj-22592	46	12	network	network	NOUN
brj-22592	46	13	with	with	ADP
brj-22592	46	14	an	an	DET
brj-22592	46	15	increased	increase	VERB
brj-22592	46	16	attention	attention	NOUN
brj-22592	46	17	mechanism	mechanism	NOUN
brj-22592	46	18	to	to	PART
brj-22592	46	19	identify	identify	VERB
brj-22592	46	20	the	the	DET
brj-22592	46	21	types	type	NOUN
brj-22592	46	22	of	of	ADP
brj-22592	46	23	wood	wood	NOUN
brj-22592	46	24	defects	defect	NOUN
brj-22592	46	25	,	,	PUNCT
brj-22592	46	26	such	such	ADJ
brj-22592	46	27	as	as	ADP
brj-22592	46	28	wormholes	wormhole	NOUN
brj-22592	46	29	,	,	PUNCT
brj-22592	46	30	slip	slip	VERB
brj-22592	46	31	knots	knot	NOUN
brj-22592	46	32	and	and	CCONJ
brj-22592	46	33	dead	dead	ADJ
brj-22592	46	34	knots	knot	NOUN
brj-22592	46	35	,	,	PUNCT
brj-22592	46	36	and	and	CCONJ
brj-22592	46	37	adjust	adjust	VERB
brj-22592	46	38	the	the	DET
brj-22592	46	39	partial	partial	ADJ
brj-22592	46	40	convolutional	convolutional	ADJ
brj-22592	46	41	structure	structure	NOUN
brj-22592	46	42	of	of	ADP
brj-22592	46	43	the	the	DET
brj-22592	46	44	bottleneck	bottleneck	NOUN
brj-22592	46	45	network	network	NOUN
brj-22592	46	46	to	to	PART
brj-22592	46	47	further	far	ADV
brj-22592	46	48	improve	improve	VERB
brj-22592	46	49	the	the	DET
brj-22592	46	50	classification	classification	NOUN
brj-22592	46	51	accuracy	accuracy	NOUN
brj-22592	46	52	of	of	ADP
brj-22592	46	53	wood	wood	NOUN
brj-22592	46	54	defects	defect	NOUN
brj-22592	46	55	,	,	PUNCT
brj-22592	46	56	reduce	reduce	VERB
brj-22592	46	57	the	the	DET
brj-22592	46	58	detection	detection	NOUN
brj-22592	46	59	time	time	NOUN
brj-22592	46	60	and	and	CCONJ
brj-22592	46	61	reduce	reduce	VERB
brj-22592	46	62	the	the	DET
brj-22592	46	63	algorithm	algorithm	NOUN
brj-22592	46	64	parameters	parameter	NOUN
brj-22592	46	65	.	.	PUNCT
brj-22592	47	1	based	base	VERB
brj-22592	47	2	on	on	ADP
brj-22592	47	3	the	the	DET
brj-22592	47	4	requirements	requirement	NOUN
brj-22592	47	5	of	of	ADP
brj-22592	47	6	computer	computer	NOUN
brj-22592	47	7	hardware	hardware	NOUN
brj-22592	47	8	,	,	PUNCT
brj-22592	47	9	in	in	ADP
brj-22592	47	10	this	this	DET
brj-22592	47	11	study	study	NOUN
brj-22592	47	12	,	,	PUNCT
brj-22592	47	13	peer	peer	NOUN
brj-22592	47	14	-	-	PUNCT
brj-22592	47	15	reviewed	review	VERB
brj-22592	47	16	article	article	NOUN
brj-22592	47	17	bioresources.com	bioresources.com	X
brj-22592	47	18	xie	xie	PROPN
brj-22592	47	19	&	&	CCONJ
brj-22592	47	20	ling	ling	PROPN
brj-22592	47	21	(	(	PUNCT
brj-22592	47	22	2023	2023	NUM
brj-22592	47	23	)	)	PUNCT
brj-22592	47	24	.	.	PUNCT
brj-22592	48	1	“	"	PUNCT
brj-22592	48	2	wood	wood	NOUN
brj-22592	48	3	defect	defect	NOUN
brj-22592	48	4	classification	classification	NOUN
brj-22592	48	5	,	,	PUNCT
brj-22592	48	6	”	"	PUNCT
brj-22592	48	7	bioresources	bioresource	NOUN
brj-22592	48	8	18(4	18(4	NUM
brj-22592	48	9	)	)	PUNCT
brj-22592	48	10	,	,	PUNCT
brj-22592	48	11	7663	7663	NUM
brj-22592	48	12	-	-	SYM
brj-22592	48	13	7680	7680	NUM
brj-22592	48	14	.	.	PUNCT
brj-22592	49	1	7665	7665	NUM
brj-22592	49	2	a	a	DET
brj-22592	49	3	reference	reference	NOUN
brj-22592	49	4	is	be	AUX
brj-22592	49	5	provided	provide	VERB
brj-22592	49	6	for	for	ADP
brj-22592	49	7	the	the	DET
brj-22592	49	8	industrial	industrial	ADJ
brj-22592	49	9	application	application	NOUN
brj-22592	49	10	of	of	ADP
brj-22592	49	11	the	the	DET
brj-22592	49	12	rapid	rapid	ADJ
brj-22592	49	13	classification	classification	NOUN
brj-22592	49	14	of	of	ADP
brj-22592	49	15	wood	wood	NOUN
brj-22592	49	16	defects	defect	NOUN
brj-22592	49	17	.	.	PUNCT
brj-22592	50	1	experimental	experimental	ADJ
brj-22592	50	2	data	datum	NOUN
brj-22592	50	3	set	set	VERB
brj-22592	50	4	construction	construction	NOUN
brj-22592	50	5	data	datum	NOUN
brj-22592	50	6	set	set	VERB
brj-22592	50	7	construction	construction	NOUN
brj-22592	50	8	was	be	AUX
brj-22592	50	9	based	base	VERB
brj-22592	50	10	on	on	ADP
brj-22592	50	11	the	the	DET
brj-22592	50	12	early	early	ADJ
brj-22592	50	13	wood	wood	NOUN
brj-22592	50	14	defect	defect	NOUN
brj-22592	50	15	research	research	NOUN
brj-22592	50	16	sample	sample	NOUN
brj-22592	50	17	set	set	NOUN
brj-22592	50	18	of	of	ADP
brj-22592	50	19	the	the	DET
brj-22592	50	20	project	project	NOUN
brj-22592	50	21	team	team	NOUN
brj-22592	50	22	(	(	PUNCT
brj-22592	50	23	xie	xie	PROPN
brj-22592	50	24	2014	2014	NUM
brj-22592	50	25	)	)	PUNCT
brj-22592	50	26	,	,	PUNCT
brj-22592	50	27	which	which	PRON
brj-22592	50	28	was	be	AUX
brj-22592	50	29	supplemented	supplement	VERB
brj-22592	50	30	with	with	ADP
brj-22592	50	31	common	common	ADJ
brj-22592	50	32	coniferous	coniferous	ADJ
brj-22592	50	33	and	and	CCONJ
brj-22592	50	34	broadleaved	broadleave	VERB
brj-22592	50	35	samples	sample	NOUN
brj-22592	50	36	from	from	ADP
brj-22592	50	37	northeast	northeast	ADJ
brj-22592	50	38	china	china	PROPN
brj-22592	50	39	.	.	PUNCT
brj-22592	51	1	the	the	DET
brj-22592	51	2	team	team	NOUN
brj-22592	51	3	collected	collect	VERB
brj-22592	51	4	wormhole	wormhole	NOUN
brj-22592	51	5	,	,	PUNCT
brj-22592	51	6	live	live	ADJ
brj-22592	51	7	knot	knot	NOUN
brj-22592	51	8	,	,	PUNCT
brj-22592	51	9	and	and	CCONJ
brj-22592	51	10	dead	dead	ADJ
brj-22592	51	11	knot	knot	NOUN
brj-22592	51	12	samples	sample	NOUN
brj-22592	51	13	to	to	PART
brj-22592	51	14	establish	establish	VERB
brj-22592	51	15	a	a	DET
brj-22592	51	16	small	small	ADJ
brj-22592	51	17	database	database	NOUN
brj-22592	51	18	that	that	PRON
brj-22592	51	19	was	be	AUX
brj-22592	51	20	divided	divide	VERB
brj-22592	51	21	into	into	ADP
brj-22592	51	22	training	training	NOUN
brj-22592	51	23	,	,	PUNCT
brj-22592	51	24	validation	validation	NOUN
brj-22592	51	25	,	,	PUNCT
brj-22592	51	26	and	and	CCONJ
brj-22592	51	27	test	test	NOUN
brj-22592	51	28	sets	set	NOUN
brj-22592	51	29	in	in	ADP
brj-22592	51	30	a	a	DET
brj-22592	51	31	6:2:2	6:2:2	NUM
brj-22592	51	32	ratio	ratio	NOUN
brj-22592	51	33	.	.	PUNCT
brj-22592	52	1	owing	owe	VERB
brj-22592	52	2	to	to	ADP
brj-22592	52	3	the	the	DET
brj-22592	52	4	small	small	ADJ
brj-22592	52	5	number	number	NOUN
brj-22592	52	6	of	of	ADP
brj-22592	52	7	samples	sample	NOUN
brj-22592	52	8	in	in	ADP
brj-22592	52	9	the	the	DET
brj-22592	52	10	dataset	dataset	NOUN
brj-22592	52	11	,	,	PUNCT
brj-22592	52	12	it	it	PRON
brj-22592	52	13	was	be	AUX
brj-22592	52	14	not	not	PART
brj-22592	52	15	suitable	suitable	ADJ
brj-22592	52	16	for	for	ADP
brj-22592	52	17	deep	deep	ADJ
brj-22592	52	18	learning	learning	NOUN
brj-22592	52	19	.	.	PUNCT
brj-22592	53	1	the	the	DET
brj-22592	53	2	samples	sample	NOUN
brj-22592	53	3	in	in	ADP
brj-22592	53	4	the	the	DET
brj-22592	53	5	training	training	NOUN
brj-22592	53	6	,	,	PUNCT
brj-22592	53	7	validation	validation	NOUN
brj-22592	53	8	and	and	CCONJ
brj-22592	53	9	test	test	NOUN
brj-22592	53	10	sets	set	NOUN
brj-22592	53	11	were	be	AUX
brj-22592	53	12	processed	process	VERB
brj-22592	53	13	using	use	VERB
brj-22592	53	14	rotation	rotation	NOUN
brj-22592	53	15	,	,	PUNCT
brj-22592	53	16	translation	translation	NOUN
brj-22592	53	17	,	,	PUNCT
brj-22592	53	18	scale	scale	NOUN
brj-22592	53	19	transformation	transformation	NOUN
brj-22592	53	20	,	,	PUNCT
brj-22592	53	21	and	and	CCONJ
brj-22592	53	22	gray	gray	ADJ
brj-22592	53	23	transformation	transformation	NOUN
brj-22592	53	24	to	to	PART
brj-22592	53	25	expand	expand	VERB
brj-22592	53	26	the	the	DET
brj-22592	53	27	sample	sample	NOUN
brj-22592	53	28	database	database	NOUN
brj-22592	53	29	(	(	PUNCT
brj-22592	53	30	ling	ling	NOUN
brj-22592	53	31	and	and	CCONJ
brj-22592	53	32	xie	xie	PROPN
brj-22592	53	33	2022	2022	NUM
brj-22592	53	34	)	)	PUNCT
brj-22592	53	35	.	.	PUNCT
brj-22592	54	1	after	after	ADP
brj-22592	54	2	capacity	capacity	NOUN
brj-22592	54	3	expansion	expansion	NOUN
brj-22592	54	4	,	,	PUNCT
brj-22592	54	5	the	the	DET
brj-22592	54	6	training	training	NOUN
brj-22592	54	7	set	set	NOUN
brj-22592	54	8	contained	contain	VERB
brj-22592	54	9	3,310	3,310	NUM
brj-22592	54	10	sample	sample	NOUN
brj-22592	54	11	data	datum	NOUN
brj-22592	54	12	,	,	PUNCT
brj-22592	54	13	the	the	DET
brj-22592	54	14	validation	validation	NOUN
brj-22592	54	15	set	set	NOUN
brj-22592	54	16	contained	contain	VERB
brj-22592	54	17	1,054	1,054	NUM
brj-22592	54	18	sample	sample	NOUN
brj-22592	54	19	data	datum	NOUN
brj-22592	54	20	,	,	PUNCT
brj-22592	54	21	and	and	CCONJ
brj-22592	54	22	the	the	DET
brj-22592	54	23	test	test	NOUN
brj-22592	54	24	set	set	NOUN
brj-22592	54	25	contained	contain	VERB
brj-22592	54	26	1,054	1,054	NUM
brj-22592	54	27	sample	sample	NOUN
brj-22592	54	28	data	datum	NOUN
brj-22592	54	29	.	.	PUNCT
brj-22592	55	1	some	some	DET
brj-22592	55	2	samples	sample	NOUN
brj-22592	55	3	of	of	ADP
brj-22592	55	4	the	the	DET
brj-22592	55	5	dataset	dataset	NOUN
brj-22592	55	6	are	be	AUX
brj-22592	55	7	shown	show	VERB
brj-22592	55	8	in	in	ADP
brj-22592	55	9	fig	fig	NOUN
brj-22592	55	10	.	.	PUNCT
brj-22592	56	1	1	1	X
brj-22592	56	2	.	.	X
brj-22592	56	3	fig	fig	NOUN
brj-22592	56	4	.	.	PUNCT
brj-22592	57	1	1	1	X
brj-22592	57	2	.	.	X
brj-22592	57	3	samples	sample	NOUN
brj-22592	57	4	of	of	ADP
brj-22592	57	5	defective	defective	ADJ
brj-22592	57	6	part	part	NOUN
brj-22592	57	7	of	of	ADP
brj-22592	57	8	a	a	DET
brj-22592	57	9	wood	wood	NOUN
brj-22592	57	10	nas	nas	NOUN
brj-22592	57	11	algorithm	algorithm	NOUN
brj-22592	57	12	prior	prior	ADV
brj-22592	57	13	to	to	ADP
brj-22592	57	14	its	its	PRON
brj-22592	57	15	introduction	introduction	NOUN
brj-22592	57	16	,	,	PUNCT
brj-22592	57	17	the	the	DET
brj-22592	57	18	regnet	regnet	NOUN
brj-22592	57	19	model	model	NOUN
brj-22592	57	20	was	be	AUX
brj-22592	57	21	used	use	VERB
brj-22592	57	22	.	.	PUNCT
brj-22592	58	1	first	first	ADV
brj-22592	58	2	,	,	PUNCT
brj-22592	58	3	the	the	DET
brj-22592	58	4	neural	neural	ADJ
brj-22592	58	5	architecture	architecture	NOUN
brj-22592	58	6	search	search	NOUN
brj-22592	58	7	algorithm	algorithm	NOUN
brj-22592	58	8	(	(	PUNCT
brj-22592	58	9	nas	nas	PROPN
brj-22592	58	10	)	)	PUNCT
brj-22592	58	11	was	be	AUX
brj-22592	58	12	introduced	introduce	VERB
brj-22592	58	13	.	.	PUNCT
brj-22592	59	1	in	in	ADP
brj-22592	59	2	deep	deep	ADJ
brj-22592	59	3	learning	learning	NOUN
brj-22592	59	4	,	,	PUNCT
brj-22592	59	5	hyperparameters	hyperparameter	NOUN
brj-22592	59	6	are	be	AUX
brj-22592	59	7	divided	divide	VERB
brj-22592	59	8	into	into	ADP
brj-22592	59	9	two	two	NUM
brj-22592	59	10	categories	category	NOUN
brj-22592	59	11	:	:	PUNCT
brj-22592	59	12	training	training	NOUN
brj-22592	59	13	parameters	parameter	NOUN
brj-22592	59	14	and	and	CCONJ
brj-22592	59	15	parameters	parameter	NOUN
brj-22592	59	16	that	that	PRON
brj-22592	59	17	define	define	VERB
brj-22592	59	18	the	the	DET
brj-22592	59	19	network	network	NOUN
brj-22592	59	20	structure	structure	NOUN
brj-22592	59	21	.	.	PUNCT
brj-22592	60	1	the	the	DET
brj-22592	60	2	training	training	NOUN
brj-22592	60	3	parameters	parameter	NOUN
brj-22592	60	4	include	include	VERB
brj-22592	60	5	the	the	DET
brj-22592	60	6	learning	learning	NOUN
brj-22592	60	7	rate	rate	NOUN
brj-22592	60	8	,	,	PUNCT
brj-22592	60	9	batch	batch	NOUN
brj-22592	60	10	size	size	NOUN
brj-22592	60	11	,	,	PUNCT
brj-22592	60	12	attenuation	attenuation	NOUN
brj-22592	60	13	weight	weight	NOUN
brj-22592	60	14	,	,	PUNCT
brj-22592	60	15	and	and	CCONJ
brj-22592	60	16	other	other	ADJ
brj-22592	60	17	parameters	parameter	NOUN
brj-22592	60	18	.	.	PUNCT
brj-22592	61	1	the	the	DET
brj-22592	61	2	automatic	automatic	ADJ
brj-22592	61	3	tuning	tuning	NOUN
brj-22592	61	4	of	of	ADP
brj-22592	61	5	the	the	DET
brj-22592	61	6	training	training	NOUN
brj-22592	61	7	parameters	parameter	NOUN
brj-22592	61	8	belongs	belong	VERB
brj-22592	61	9	to	to	PART
brj-22592	61	10	hyperparameter	hyperparameter	VERB
brj-22592	61	11	optimization	optimization	NOUN
brj-22592	61	12	(	(	PUNCT
brj-22592	61	13	ho	ho	PROPN
brj-22592	61	14	)	)	PUNCT
brj-22592	61	15	.	.	PUNCT
brj-22592	62	1	the	the	DET
brj-22592	62	2	parameters	parameter	NOUN
brj-22592	62	3	defining	define	VERB
brj-22592	62	4	the	the	DET
brj-22592	62	5	network	network	NOUN
brj-22592	62	6	structure	structure	NOUN
brj-22592	62	7	include	include	VERB
brj-22592	62	8	the	the	DET
brj-22592	62	9	layer	layer	NOUN
brj-22592	62	10	structure	structure	NOUN
brj-22592	62	11	operator	operator	NOUN
brj-22592	62	12	,	,	PUNCT
brj-22592	62	13	convolution	convolution	NOUN
brj-22592	62	14	kernel	kernel	NOUN
brj-22592	62	15	size	size	NOUN
brj-22592	62	16	,	,	PUNCT
brj-22592	62	17	dimension	dimension	NOUN
brj-22592	62	18	,	,	PUNCT
brj-22592	62	19	and	and	CCONJ
brj-22592	62	20	dispersion	dispersion	NOUN
brj-22592	62	21	degree	degree	NOUN
brj-22592	62	22	.	.	PUNCT
brj-22592	63	1	automatic	automatic	ADJ
brj-22592	63	2	optimization	optimization	NOUN
brj-22592	63	3	of	of	ADP
brj-22592	63	4	network	network	NOUN
brj-22592	63	5	structure	structure	NOUN
brj-22592	63	6	parameters	parameter	NOUN
brj-22592	63	7	is	be	AUX
brj-22592	63	8	generally	generally	ADV
brj-22592	63	9	termed	term	VERB
brj-22592	63	10	neural	neural	ADJ
brj-22592	63	11	architecture	architecture	NOUN
brj-22592	63	12	search	search	NOUN
brj-22592	63	13	(	(	PUNCT
brj-22592	63	14	nas	nas	PROPN
brj-22592	63	15	)	)	PUNCT
brj-22592	63	16	.	.	PUNCT
brj-22592	64	1	the	the	DET
brj-22592	64	2	classic	classic	ADJ
brj-22592	64	3	nas	nas	NOUN
brj-22592	64	4	algorithm	algorithm	NOUN
brj-22592	64	5	includes	include	VERB
brj-22592	64	6	three	three	NUM
brj-22592	64	7	aspects	aspect	NOUN
brj-22592	64	8	:	:	PUNCT
brj-22592	64	9	search	search	NOUN
brj-22592	64	10	space	space	NOUN
brj-22592	64	11	,	,	PUNCT
brj-22592	64	12	search	search	NOUN
brj-22592	64	13	strategy	strategy	NOUN
brj-22592	64	14	,	,	PUNCT
brj-22592	64	15	evaluation	evaluation	NOUN
brj-22592	64	16	,	,	PUNCT
brj-22592	64	17	and	and	CCONJ
brj-22592	64	18	prediction	prediction	NOUN
brj-22592	64	19	.	.	PUNCT
brj-22592	65	1	the	the	DET
brj-22592	65	2	search	search	NOUN
brj-22592	65	3	strategy	strategy	NOUN
brj-22592	65	4	is	be	AUX
brj-22592	65	5	selected	select	VERB
brj-22592	65	6	based	base	VERB
brj-22592	65	7	on	on	ADP
brj-22592	65	8	search	search	NOUN
brj-22592	65	9	space	space	NOUN
brj-22592	65	10	.	.	PUNCT
brj-22592	66	1	when	when	SCONJ
brj-22592	66	2	the	the	DET
brj-22592	66	3	strategy	strategy	NOUN
brj-22592	66	4	is	be	AUX
brj-22592	66	5	determined	determine	VERB
brj-22592	66	6	,	,	PUNCT
brj-22592	66	7	the	the	DET
brj-22592	66	8	designed	design	VERB
brj-22592	66	9	network	network	NOUN
brj-22592	66	10	structure	structure	NOUN
brj-22592	66	11	is	be	AUX
brj-22592	66	12	evaluated	evaluate	VERB
brj-22592	66	13	and	and	CCONJ
brj-22592	66	14	predicted	predict	VERB
brj-22592	66	15	,	,	PUNCT
brj-22592	66	16	and	and	CCONJ
brj-22592	66	17	it	it	PRON
brj-22592	66	18	continues	continue	VERB
brj-22592	66	19	to	to	PART
brj-22592	66	20	strengthen	strengthen	VERB
brj-22592	66	21	the	the	DET
brj-22592	66	22	search	search	NOUN
brj-22592	66	23	for	for	ADP
brj-22592	66	24	a	a	DET
brj-22592	66	25	better	well	ADJ
brj-22592	66	26	network	network	NOUN
brj-22592	66	27	model	model	NOUN
brj-22592	66	28	,	,	PUNCT
brj-22592	66	29	which	which	PRON
brj-22592	66	30	corresponds	correspond	VERB
brj-22592	66	31	to	to	ADP
brj-22592	66	32	the	the	DET
brj-22592	66	33	nas	nas	PROPN
brj-22592	66	34	algorithm	algorithm	NOUN
brj-22592	66	35	.	.	PUNCT
brj-22592	67	1	figure	figure	NOUN
brj-22592	67	2	2	2	NUM
brj-22592	67	3	shows	show	VERB
brj-22592	67	4	the	the	DET
brj-22592	67	5	rules	rule	NOUN
brj-22592	67	6	of	of	ADP
brj-22592	67	7	the	the	DET
brj-22592	67	8	nas	nas	PROPN
brj-22592	67	9	algorithm	algorithm	NOUN
brj-22592	67	10	.	.	PUNCT
brj-22592	68	1	(	(	PUNCT
brj-22592	68	2	a	a	X
brj-22592	68	3	)	)	PUNCT
brj-22592	68	4	wormhole	wormhole	NOUN
brj-22592	68	5	(	(	PUNCT
brj-22592	68	6	b	b	X
brj-22592	68	7	)	)	PUNCT
brj-22592	68	8	slip	slip	NOUN
brj-22592	68	9	knots	knot	NOUN
brj-22592	68	10	(	(	PUNCT
brj-22592	68	11	c	c	NOUN
brj-22592	68	12	)	)	PUNCT
brj-22592	68	13	dead	dead	ADJ
brj-22592	68	14	knots	knot	NOUN
brj-22592	68	15	peer	peer	NOUN
brj-22592	68	16	-	-	PUNCT
brj-22592	68	17	reviewed	review	VERB
brj-22592	68	18	article	article	NOUN
brj-22592	68	19	bioresources.com	bioresources.com	X
brj-22592	68	20	xie	xie	PROPN
brj-22592	68	21	&	&	CCONJ
brj-22592	68	22	ling	ling	PROPN
brj-22592	68	23	(	(	PUNCT
brj-22592	68	24	2023	2023	NUM
brj-22592	68	25	)	)	PUNCT
brj-22592	68	26	.	.	PUNCT
brj-22592	69	1	“	"	PUNCT
brj-22592	69	2	wood	wood	NOUN
brj-22592	69	3	defect	defect	NOUN
brj-22592	69	4	classification	classification	NOUN
brj-22592	69	5	,	,	PUNCT
brj-22592	69	6	”	"	PUNCT
brj-22592	69	7	bioresources	bioresource	NOUN
brj-22592	69	8	18(4	18(4	NUM
brj-22592	69	9	)	)	PUNCT
brj-22592	69	10	,	,	PUNCT
brj-22592	69	11	7663	7663	NUM
brj-22592	69	12	-	-	SYM
brj-22592	69	13	7680	7680	NUM
brj-22592	69	14	.	.	PUNCT
brj-22592	70	1	7666	7666	NUM
brj-22592	70	2	fig	fig	NOUN
brj-22592	70	3	.	.	PUNCT
brj-22592	71	1	2	2	X
brj-22592	71	2	.	.	X
brj-22592	71	3	nas	nas	PROPN
brj-22592	71	4	algorithm	algorithm	NOUN
brj-22592	71	5	search	search	NOUN
brj-22592	71	6	space	space	NOUN
brj-22592	71	7	in	in	ADP
brj-22592	71	8	principle	principle	NOUN
brj-22592	71	9	,	,	PUNCT
brj-22592	71	10	using	use	VERB
brj-22592	71	11	the	the	DET
brj-22592	71	12	nas	nas	PROPN
brj-22592	71	13	algorithm	algorithm	NOUN
brj-22592	71	14	,	,	PUNCT
brj-22592	71	15	a	a	DET
brj-22592	71	16	space	space	NOUN
brj-22592	71	17	composed	compose	VERB
brj-22592	71	18	of	of	ADP
brj-22592	71	19	all	all	DET
brj-22592	71	20	potential	potential	ADJ
brj-22592	71	21	network	network	NOUN
brj-22592	71	22	structure	structure	NOUN
brj-22592	71	23	models	model	NOUN
brj-22592	71	24	is	be	AUX
brj-22592	71	25	defined	define	VERB
brj-22592	71	26	as	as	ADP
brj-22592	71	27	the	the	DET
brj-22592	71	28	search	search	NOUN
brj-22592	71	29	space	space	NOUN
brj-22592	71	30	.	.	PUNCT
brj-22592	72	1	the	the	DET
brj-22592	72	2	neural	neural	ADJ
brj-22592	72	3	network	network	NOUN
brj-22592	72	4	architecture	architecture	NOUN
brj-22592	72	5	searches	search	VERB
brj-22592	72	6	the	the	DET
brj-22592	72	7	network	network	NOUN
brj-22592	72	8	model	model	NOUN
brj-22592	72	9	structure	structure	NOUN
brj-22592	72	10	parameters	parameter	NOUN
brj-22592	72	11	that	that	PRON
brj-22592	72	12	can	can	AUX
brj-22592	72	13	be	be	AUX
brj-22592	72	14	optimized	optimize	VERB
brj-22592	72	15	,	,	PUNCT
brj-22592	72	16	including	include	VERB
brj-22592	72	17	the	the	DET
brj-22592	72	18	number	number	NOUN
brj-22592	72	19	of	of	ADP
brj-22592	72	20	layers	layer	NOUN
brj-22592	72	21	n	n	CCONJ
brj-22592	72	22	,	,	PUNCT
brj-22592	72	23	type	type	NOUN
brj-22592	72	24	of	of	ADP
brj-22592	72	25	operation	operation	NOUN
brj-22592	72	26	performed	perform	VERB
brj-22592	72	27	by	by	ADP
brj-22592	72	28	layers	layer	NOUN
brj-22592	72	29	,	,	PUNCT
brj-22592	72	30	and	and	CCONJ
brj-22592	72	31	the	the	DET
brj-22592	72	32	super	super	ADJ
brj-22592	72	33	parameters	parameter	NOUN
brj-22592	72	34	in	in	ADP
brj-22592	72	35	the	the	DET
brj-22592	72	36	layer	layer	NOUN
brj-22592	72	37	structure	structure	NOUN
brj-22592	72	38	.	.	PUNCT
brj-22592	73	1	search	search	NOUN
brj-22592	73	2	strategy	strategy	NOUN
brj-22592	73	3	given	give	VERB
brj-22592	73	4	the	the	DET
brj-22592	73	5	search	search	NOUN
brj-22592	73	6	space	space	NOUN
brj-22592	73	7	,	,	PUNCT
brj-22592	73	8	the	the	DET
brj-22592	73	9	best	good	ADJ
brj-22592	73	10	neural	neural	ADJ
brj-22592	73	11	network	network	NOUN
brj-22592	73	12	structure	structure	NOUN
brj-22592	73	13	can	can	AUX
brj-22592	73	14	be	be	AUX
brj-22592	73	15	determined	determine	VERB
brj-22592	73	16	using	use	VERB
brj-22592	73	17	the	the	DET
brj-22592	73	18	corresponding	corresponding	ADJ
brj-22592	73	19	search	search	NOUN
brj-22592	73	20	strategy	strategy	NOUN
brj-22592	73	21	.	.	PUNCT
brj-22592	74	1	the	the	DET
brj-22592	74	2	common	common	ADJ
brj-22592	74	3	search	search	NOUN
brj-22592	74	4	strategies	strategy	NOUN
brj-22592	74	5	include	include	VERB
brj-22592	74	6	reinforcementbased	reinforcementbased	ADJ
brj-22592	74	7	learning	learning	NOUN
brj-22592	74	8	,	,	PUNCT
brj-22592	74	9	evolutionary	evolutionary	ADJ
brj-22592	74	10	algorithms	algorithm	NOUN
brj-22592	74	11	,	,	PUNCT
brj-22592	74	12	bayesian	bayesian	NOUN
brj-22592	74	13	optimization	optimization	NOUN
brj-22592	74	14	algorithms	algorithm	NOUN
brj-22592	74	15	,	,	PUNCT
brj-22592	74	16	and	and	CCONJ
brj-22592	74	17	gradientbased	gradientbased	ADJ
brj-22592	74	18	methods	method	NOUN
brj-22592	74	19	(	(	PUNCT
brj-22592	74	20	jin	jin	NOUN
brj-22592	74	21	et	et	PROPN
brj-22592	74	22	al	al	PROPN
brj-22592	74	23	.	.	PROPN
brj-22592	74	24	2019	2019	NUM
brj-22592	74	25	)	)	PUNCT
brj-22592	74	26	.	.	PUNCT
brj-22592	75	1	evaluation	evaluation	NOUN
brj-22592	75	2	and	and	CCONJ
brj-22592	75	3	prediction	prediction	VERB
brj-22592	75	4	the	the	DET
brj-22592	75	5	purpose	purpose	NOUN
brj-22592	75	6	of	of	ADP
brj-22592	75	7	the	the	DET
brj-22592	75	8	search	search	NOUN
brj-22592	75	9	strategy	strategy	NOUN
brj-22592	75	10	involves	involve	VERB
brj-22592	75	11	designing	design	VERB
brj-22592	75	12	a	a	DET
brj-22592	75	13	good	good	ADJ
brj-22592	75	14	network	network	NOUN
brj-22592	75	15	structure	structure	NOUN
brj-22592	75	16	and	and	CCONJ
brj-22592	75	17	ensuring	ensure	VERB
brj-22592	75	18	that	that	SCONJ
brj-22592	75	19	the	the	DET
brj-22592	75	20	network	network	NOUN
brj-22592	75	21	structure	structure	NOUN
brj-22592	75	22	exhibits	exhibit	VERB
brj-22592	75	23	the	the	DET
brj-22592	75	24	highest	high	ADJ
brj-22592	75	25	accuracy	accuracy	NOUN
brj-22592	75	26	in	in	ADP
brj-22592	75	27	the	the	DET
brj-22592	75	28	test	test	NOUN
brj-22592	75	29	set	set	VERB
brj-22592	75	30	.	.	PUNCT
brj-22592	76	1	to	to	PART
brj-22592	76	2	strengthen	strengthen	VERB
brj-22592	76	3	the	the	DET
brj-22592	76	4	search	search	NOUN
brj-22592	76	5	process	process	NOUN
brj-22592	76	6	,	,	PUNCT
brj-22592	76	7	the	the	DET
brj-22592	76	8	search	search	NOUN
brj-22592	76	9	strategy	strategy	NOUN
brj-22592	76	10	must	must	AUX
brj-22592	76	11	evaluate	evaluate	VERB
brj-22592	76	12	the	the	DET
brj-22592	76	13	performance	performance	NOUN
brj-22592	76	14	of	of	ADP
brj-22592	76	15	a	a	DET
brj-22592	76	16	given	give	VERB
brj-22592	76	17	network	network	NOUN
brj-22592	76	18	structure	structure	NOUN
brj-22592	76	19	.	.	PUNCT
brj-22592	77	1	regnet	regnet	PROPN
brj-22592	77	2	model	model	NOUN
brj-22592	77	3	the	the	DET
brj-22592	77	4	regnet	regnet	PROPN
brj-22592	77	5	model	model	NOUN
brj-22592	77	6	uses	use	VERB
brj-22592	77	7	the	the	DET
brj-22592	77	8	neural	neural	ADJ
brj-22592	77	9	architecture	architecture	NOUN
brj-22592	77	10	search	search	NOUN
brj-22592	77	11	algorithm	algorithm	NOUN
brj-22592	77	12	(	(	PUNCT
brj-22592	77	13	nas	nas	PROPN
brj-22592	77	14	)	)	PUNCT
brj-22592	77	15	to	to	PART
brj-22592	77	16	obtain	obtain	VERB
brj-22592	77	17	a	a	DET
brj-22592	77	18	low	low	ADJ
brj-22592	77	19	-	-	PUNCT
brj-22592	77	20	dimensional	dimensional	ADJ
brj-22592	77	21	design	design	NOUN
brj-22592	77	22	space	space	NOUN
brj-22592	77	23	composed	compose	VERB
brj-22592	77	24	of	of	ADP
brj-22592	77	25	a	a	DET
brj-22592	77	26	simple	simple	ADJ
brj-22592	77	27	rule	rule	NOUN
brj-22592	77	28	network	network	NOUN
brj-22592	77	29	in	in	ADP
brj-22592	77	30	a	a	DET
brj-22592	77	31	relatively	relatively	ADV
brj-22592	77	32	unconstrained	unconstrained	ADJ
brj-22592	77	33	design	design	NOUN
brj-22592	77	34	space	space	NOUN
brj-22592	77	35	by	by	ADP
brj-22592	77	36	using	use	VERB
brj-22592	77	37	the	the	DET
brj-22592	77	38	man	man	NOUN
brj-22592	77	39	-	-	PUNCT
brj-22592	77	40	machine	machine	NOUN
brj-22592	77	41	rotation	rotation	NOUN
brj-22592	77	42	method	method	NOUN
brj-22592	77	43	.	.	PUNCT
brj-22592	78	1	this	this	DET
brj-22592	78	2	design	design	NOUN
brj-22592	78	3	space	space	NOUN
brj-22592	78	4	is	be	AUX
brj-22592	78	5	termed	term	VERB
brj-22592	78	6	as	as	ADP
brj-22592	78	7	regnet	regnet	NOUN
brj-22592	78	8	space	space	NOUN
brj-22592	78	9	(	(	PUNCT
brj-22592	78	10	han	han	PROPN
brj-22592	78	11	et	et	PROPN
brj-22592	78	12	al	al	PROPN
brj-22592	78	13	.	.	PROPN
brj-22592	78	14	2020	2020	NUM
brj-22592	78	15	)	)	PUNCT
brj-22592	78	16	.	.	PUNCT
brj-22592	79	1	the	the	DET
brj-22592	79	2	central	central	ADJ
brj-22592	79	3	idea	idea	NOUN
brj-22592	79	4	of	of	ADP
brj-22592	79	5	the	the	DET
brj-22592	79	6	regnet	regnet	NOUN
brj-22592	79	7	design	design	NOUN
brj-22592	79	8	space	space	NOUN
brj-22592	79	9	is	be	AUX
brj-22592	79	10	that	that	SCONJ
brj-22592	79	11	the	the	DET
brj-22592	79	12	width	width	ADJ
brj-22592	79	13	and	and	CCONJ
brj-22592	79	14	depth	depth	NOUN
brj-22592	79	15	of	of	ADP
brj-22592	79	16	the	the	DET
brj-22592	79	17	model	model	NOUN
brj-22592	79	18	are	be	AUX
brj-22592	79	19	determined	determine	VERB
brj-22592	79	20	by	by	ADP
brj-22592	79	21	a	a	DET
brj-22592	79	22	quantitative	quantitative	ADJ
brj-22592	79	23	linear	linear	NOUN
brj-22592	79	24	function	function	NOUN
brj-22592	79	25	.	.	PUNCT
brj-22592	80	1	the	the	DET
brj-22592	80	2	structure	structure	NOUN
brj-22592	80	3	block	block	NOUN
brj-22592	80	4	diagram	diagram	NOUN
brj-22592	80	5	of	of	ADP
brj-22592	80	6	regnet	regnet	NOUN
brj-22592	80	7	network	network	NOUN
brj-22592	80	8	is	be	AUX
brj-22592	80	9	shown	show	VERB
brj-22592	80	10	in	in	ADP
brj-22592	80	11	fig	fig	NOUN
brj-22592	80	12	.	.	PUNCT
brj-22592	81	1	3	3	X
brj-22592	81	2	.	.	X
brj-22592	81	3	the	the	DET
brj-22592	81	4	network	network	NOUN
brj-22592	81	5	is	be	AUX
brj-22592	81	6	mainly	mainly	ADV
brj-22592	81	7	composed	compose	VERB
brj-22592	81	8	of	of	ADP
brj-22592	81	9	stem	stem	NOUN
brj-22592	81	10	,	,	PUNCT
brj-22592	81	11	body	body	NOUN
brj-22592	81	12	,	,	PUNCT
brj-22592	81	13	and	and	CCONJ
brj-22592	81	14	head	head	NOUN
brj-22592	81	15	.	.	PUNCT
brj-22592	82	1	stem	stem	VERB
brj-22592	82	2	the	the	DET
brj-22592	82	3	stem	stem	NOUN
brj-22592	82	4	is	be	AUX
brj-22592	82	5	composed	compose	VERB
brj-22592	82	6	of	of	ADP
brj-22592	82	7	a	a	DET
brj-22592	82	8	convolution	convolution	NOUN
brj-22592	82	9	layer	layer	NOUN
brj-22592	82	10	containing	contain	VERB
brj-22592	82	11	a	a	DET
brj-22592	82	12	bn	bn	NOUN
brj-22592	82	13	layer	layer	NOUN
brj-22592	82	14	and	and	CCONJ
brj-22592	82	15	relu	relu	NOUN
brj-22592	82	16	activation	activation	NOUN
brj-22592	82	17	function	function	NOUN
brj-22592	82	18	.	.	PUNCT
brj-22592	83	1	the	the	DET
brj-22592	83	2	size	size	NOUN
brj-22592	83	3	of	of	ADP
brj-22592	83	4	the	the	DET
brj-22592	83	5	convolution	convolution	NOUN
brj-22592	83	6	kernel	kernel	NOUN
brj-22592	83	7	is	be	AUX
brj-22592	83	8	3×3	3×3	NUM
brj-22592	83	9	.	.	PUNCT
brj-22592	84	1	the	the	DET
brj-22592	84	2	step	step	NOUN
brj-22592	84	3	length	length	NOUN
brj-22592	84	4	is	be	AUX
brj-22592	84	5	two	two	NUM
brj-22592	84	6	and	and	CCONJ
brj-22592	84	7	the	the	DET
brj-22592	84	8	number	number	NOUN
brj-22592	84	9	of	of	ADP
brj-22592	84	10	convolution	convolution	NOUN
brj-22592	84	11	kernels	kernel	NOUN
brj-22592	84	12	is	be	AUX
brj-22592	84	13	32	32	NUM
brj-22592	84	14	.	.	PUNCT
brj-22592	85	1	body	body	NOUN
brj-22592	85	2	the	the	DET
brj-22592	85	3	body	body	NOUN
brj-22592	85	4	is	be	AUX
brj-22592	85	5	composed	compose	VERB
brj-22592	85	6	of	of	ADP
brj-22592	85	7	four	four	NUM
brj-22592	85	8	-	-	PUNCT
brj-22592	85	9	stage	stage	NOUN
brj-22592	85	10	stacks	stack	NOUN
brj-22592	85	11	,	,	PUNCT
brj-22592	85	12	and	and	CCONJ
brj-22592	85	13	each	each	DET
brj-22592	85	14	stage	stage	NOUN
brj-22592	85	15	is	be	AUX
brj-22592	85	16	stacked	stack	VERB
brj-22592	85	17	with	with	ADP
brj-22592	85	18	a	a	DET
brj-22592	85	19	large	large	ADJ
brj-22592	85	20	number	number	NOUN
brj-22592	85	21	of	of	ADP
brj-22592	85	22	blocks	block	NOUN
brj-22592	85	23	.	.	PUNCT
brj-22592	86	1	in	in	ADP
brj-22592	86	2	this	this	DET
brj-22592	86	3	stage	stage	NOUN
brj-22592	86	4	,	,	PUNCT
brj-22592	86	5	in	in	ADP
brj-22592	86	6	the	the	DET
brj-22592	86	7	main	main	ADJ
brj-22592	86	8	branch	branch	NOUN
brj-22592	86	9	and	and	CCONJ
brj-22592	86	10	shortcut	shortcut	VERB
brj-22592	86	11	branch	branch	NOUN
brj-22592	86	12	of	of	ADP
brj-22592	86	13	the	the	DET
brj-22592	86	14	first	first	ADJ
brj-22592	86	15	block	block	NOUN
brj-22592	86	16	part	part	NOUN
brj-22592	86	17	of	of	ADP
brj-22592	86	18	the	the	DET
brj-22592	86	19	stage	stage	NOUN
brj-22592	86	20	,	,	PUNCT
brj-22592	86	21	there	there	PRON
brj-22592	86	22	is	be	VERB
brj-22592	86	23	a	a	DET
brj-22592	86	24	group	group	NOUN
brj-22592	86	25	convolution	convolution	NOUN
brj-22592	86	26	with	with	ADP
brj-22592	86	27	two	two	NUM
brj-22592	86	28	steps	step	NOUN
brj-22592	86	29	,	,	PUNCT
brj-22592	86	30	and	and	CCONJ
brj-22592	86	31	the	the	DET
brj-22592	86	32	number	number	NOUN
brj-22592	86	33	of	of	ADP
brj-22592	86	34	convolution	convolution	NOUN
brj-22592	86	35	steps	step	NOUN
brj-22592	86	36	for	for	ADP
brj-22592	86	37	the	the	DET
brj-22592	86	38	other	other	ADJ
brj-22592	86	39	block	block	NOUN
brj-22592	86	40	parts	part	NOUN
brj-22592	86	41	corresponds	correspond	VERB
brj-22592	86	42	to	to	ADP
brj-22592	86	43	one	one	NUM
brj-22592	86	44	.	.	PUNCT
brj-22592	87	1	peer	peer	NOUN
brj-22592	87	2	-	-	PUNCT
brj-22592	87	3	reviewed	review	VERB
brj-22592	87	4	article	article	NOUN
brj-22592	87	5	bioresources.com	bioresources.com	X
brj-22592	87	6	xie	xie	PROPN
brj-22592	87	7	&	&	CCONJ
brj-22592	87	8	ling	ling	PROPN
brj-22592	87	9	(	(	PUNCT
brj-22592	87	10	2023	2023	NUM
brj-22592	87	11	)	)	PUNCT
brj-22592	87	12	.	.	PUNCT
brj-22592	88	1	“	"	PUNCT
brj-22592	88	2	wood	wood	NOUN
brj-22592	88	3	defect	defect	NOUN
brj-22592	88	4	classification	classification	NOUN
brj-22592	88	5	,	,	PUNCT
brj-22592	88	6	”	"	PUNCT
brj-22592	88	7	bioresources	bioresource	NOUN
brj-22592	88	8	18(4	18(4	NUM
brj-22592	88	9	)	)	PUNCT
brj-22592	88	10	,	,	PUNCT
brj-22592	88	11	7663	7663	NUM
brj-22592	88	12	-	-	SYM
brj-22592	88	13	7680	7680	NUM
brj-22592	88	14	.	.	PUNCT
brj-22592	88	15	7667	7667	NUM
brj-22592	88	16	head	head	NOUN
brj-22592	88	17	the	the	DET
brj-22592	88	18	head	head	NOUN
brj-22592	88	19	is	be	AUX
brj-22592	88	20	composed	compose	VERB
brj-22592	88	21	of	of	ADP
brj-22592	88	22	a	a	DET
brj-22592	88	23	global	global	ADJ
brj-22592	88	24	average	average	ADJ
brj-22592	88	25	pooling	pooling	NOUN
brj-22592	88	26	layer	layer	NOUN
brj-22592	88	27	and	and	CCONJ
brj-22592	88	28	full	full	ADJ
brj-22592	88	29	connection	connection	NOUN
brj-22592	88	30	layer	layer	NOUN
brj-22592	88	31	as	as	ADP
brj-22592	88	32	part	part	NOUN
brj-22592	88	33	of	of	ADP
brj-22592	88	34	the	the	DET
brj-22592	88	35	classifier	classifier	NOUN
brj-22592	88	36	in	in	ADP
brj-22592	88	37	the	the	DET
brj-22592	88	38	classification	classification	NOUN
brj-22592	88	39	network	network	NOUN
brj-22592	88	40	.	.	PUNCT
brj-22592	89	1	fig	fig	NOUN
brj-22592	89	2	.	.	PUNCT
brj-22592	90	1	3	3	X
brj-22592	90	2	.	.	X
brj-22592	90	3	processing	processing	NOUN
brj-22592	90	4	flow	flow	NOUN
brj-22592	90	5	of	of	ADP
brj-22592	90	6	the	the	DET
brj-22592	90	7	regnet	regnet	NOUN
brj-22592	90	8	network	network	NOUN
brj-22592	90	9	model	model	NOUN
brj-22592	90	10	the	the	DET
brj-22592	90	11	main	main	ADJ
brj-22592	90	12	branch	branch	NOUN
brj-22592	90	13	of	of	ADP
brj-22592	90	14	the	the	DET
brj-22592	90	15	block	block	NOUN
brj-22592	90	16	module	module	NOUN
brj-22592	90	17	of	of	ADP
brj-22592	90	18	the	the	DET
brj-22592	90	19	regnet	regnet	NOUN
brj-22592	90	20	model	model	NOUN
brj-22592	90	21	is	be	AUX
brj-22592	90	22	a	a	DET
brj-22592	90	23	1	1	NUM
brj-22592	90	24	×	×	NOUN
brj-22592	90	25	1	1	NUM
brj-22592	90	26	convolution	convolution	NOUN
brj-22592	90	27	kernel	kernel	NOUN
brj-22592	90	28	.	.	PUNCT
brj-22592	91	1	the	the	DET
brj-22592	91	2	group	group	NOUN
brj-22592	91	3	convolution	convolution	NOUN
brj-22592	91	4	of	of	ADP
brj-22592	91	5	1	1	NUM
brj-22592	91	6	(	(	PUNCT
brj-22592	91	7	including	include	VERB
brj-22592	91	8	bn	bn	NUM
brj-22592	91	9	and	and	CCONJ
brj-22592	91	10	relu	relu	NOUN
brj-22592	91	11	)	)	PUNCT
brj-22592	91	12	and	and	CCONJ
brj-22592	91	13	one	one	NUM
brj-22592	91	14	convolution	convolution	NOUN
brj-22592	91	15	kernel	kernel	NOUN
brj-22592	91	16	is	be	AUX
brj-22592	91	17	3×3	3×3	NUM
brj-22592	91	18	.	.	PUNCT
brj-22592	92	1	the	the	DET
brj-22592	92	2	group	group	NOUN
brj-22592	92	3	convolution	convolution	NOUN
brj-22592	92	4	of	of	ADP
brj-22592	92	5	three	three	NUM
brj-22592	92	6	(	(	PUNCT
brj-22592	92	7	including	include	VERB
brj-22592	92	8	bn	bn	NUM
brj-22592	92	9	and	and	CCONJ
brj-22592	92	10	relu	relu	NOUN
brj-22592	92	11	)	)	PUNCT
brj-22592	92	12	and	and	CCONJ
brj-22592	92	13	one	one	NUM
brj-22592	92	14	convolution	convolution	NOUN
brj-22592	92	15	kernel	kernel	NOUN
brj-22592	92	16	is	be	AUX
brj-22592	92	17	1×1	1×1	NUM
brj-22592	92	18	(	(	PUNCT
brj-22592	92	19	including	include	VERB
brj-22592	92	20	bn	bn	NUM
brj-22592	92	21	)	)	PUNCT
brj-22592	92	22	.	.	PUNCT
brj-22592	93	1	on	on	ADP
brj-22592	93	2	the	the	DET
brj-22592	93	3	shortcut	shortcut	NOUN
brj-22592	93	4	branch	branch	NOUN
brj-22592	93	5	,	,	PUNCT
brj-22592	93	6	when	when	SCONJ
brj-22592	93	7	stride	stride	NOUN
brj-22592	93	8	=	=	SYM
brj-22592	93	9	1	1	NUM
brj-22592	93	10	,	,	PUNCT
brj-22592	93	11	the	the	DET
brj-22592	93	12	input	input	NOUN
brj-22592	93	13	data	datum	NOUN
brj-22592	93	14	information	information	NOUN
brj-22592	93	15	is	be	AUX
brj-22592	93	16	not	not	PART
brj-22592	93	17	processed	process	VERB
brj-22592	93	18	.	.	PUNCT
brj-22592	94	1	when	when	SCONJ
brj-22592	94	2	stride	stride	NOUN
brj-22592	94	3	=	=	SYM
brj-22592	94	4	2	2	NUM
brj-22592	94	5	,	,	PUNCT
brj-22592	94	6	the	the	DET
brj-22592	94	7	input	input	NOUN
brj-22592	94	8	data	data	NOUN
brj-22592	94	9	is	be	AUX
brj-22592	94	10	processed	process	VERB
brj-22592	94	11	via	via	ADP
brj-22592	94	12	a	a	DET
brj-22592	94	13	convolution	convolution	NOUN
brj-22592	94	14	kernel	kernel	NOUN
brj-22592	94	15	1	1	NUM
brj-22592	94	16	×	×	NOUN
brj-22592	94	17	1	1	NUM
brj-22592	94	18	(	(	PUNCT
brj-22592	94	19	including	include	VERB
brj-22592	94	20	bn	bn	NOUN
brj-22592	94	21	)	)	PUNCT
brj-22592	94	22	to	to	PART
brj-22592	94	23	down	down	ADP
brj-22592	94	24	sample	sample	VERB
brj-22592	94	25	the	the	DET
brj-22592	94	26	information	information	NOUN
brj-22592	94	27	.	.	PUNCT
brj-22592	95	1	the	the	DET
brj-22592	95	2	block	block	NOUN
brj-22592	95	3	module	module	NOUN
brj-22592	95	4	structure	structure	NOUN
brj-22592	95	5	of	of	ADP
brj-22592	95	6	the	the	DET
brj-22592	95	7	regnet	regnet	NOUN
brj-22592	95	8	model	model	NOUN
brj-22592	95	9	is	be	AUX
brj-22592	95	10	shown	show	VERB
brj-22592	95	11	in	in	ADP
brj-22592	95	12	fig	fig	NOUN
brj-22592	95	13	.	.	PUNCT
brj-22592	96	1	4	4	X
brj-22592	96	2	.	.	X
brj-22592	97	1	in	in	ADP
brj-22592	97	2	fig	fig	NOUN
brj-22592	97	3	.	.	PUNCT
brj-22592	98	1	4	4	NUM
brj-22592	98	2	,	,	PUNCT
brj-22592	98	3	r	r	NOUN
brj-22592	98	4	denotes	denote	VERB
brj-22592	98	5	the	the	DET
brj-22592	98	6	resolution	resolution	NOUN
brj-22592	98	7	,	,	PUNCT
brj-22592	98	8	which	which	PRON
brj-22592	98	9	can	can	AUX
brj-22592	98	10	be	be	AUX
brj-22592	98	11	understood	understand	VERB
brj-22592	98	12	as	as	ADP
brj-22592	98	13	the	the	DET
brj-22592	98	14	height	height	NOUN
brj-22592	98	15	and	and	CCONJ
brj-22592	98	16	width	width	NOUN
brj-22592	98	17	of	of	ADP
brj-22592	98	18	the	the	DET
brj-22592	98	19	characteristic	characteristic	ADJ
brj-22592	98	20	matrix	matrix	NOUN
brj-22592	98	21	.	.	PUNCT
brj-22592	99	1	when	when	SCONJ
brj-22592	99	2	stride	stride	NOUN
brj-22592	99	3	=	=	SYM
brj-22592	99	4	1	1	NUM
brj-22592	99	5	,	,	PUNCT
brj-22592	99	6	the	the	DET
brj-22592	99	7	input	input	NOUN
brj-22592	99	8	and	and	CCONJ
brj-22592	99	9	output	output	NOUN
brj-22592	99	10	r	r	NOUN
brj-22592	99	11	are	be	AUX
brj-22592	99	12	equal	equal	ADJ
brj-22592	99	13	,	,	PUNCT
brj-22592	99	14	and	and	CCONJ
brj-22592	99	15	when	when	SCONJ
brj-22592	99	16	stride	stride	NOUN
brj-22592	99	17	=	=	SYM
brj-22592	99	18	2	2	NUM
brj-22592	99	19	,	,	PUNCT
brj-22592	99	20	the	the	DET
brj-22592	99	21	output	output	NOUN
brj-22592	99	22	r	r	NOUN
brj-22592	99	23	is	be	AUX
brj-22592	99	24	half	half	DET
brj-22592	99	25	that	that	PRON
brj-22592	99	26	of	of	ADP
brj-22592	99	27	the	the	DET
brj-22592	99	28	input	input	NOUN
brj-22592	99	29	.	.	PUNCT
brj-22592	100	1	specifically	specifically	ADV
brj-22592	100	2	,	,	PUNCT
brj-22592	100	3	w	w	PROPN
brj-22592	100	4	denotes	denote	NOUN
brj-22592	100	5	the	the	DET
brj-22592	100	6	number	number	NOUN
brj-22592	100	7	of	of	ADP
brj-22592	100	8	channels	channel	NOUN
brj-22592	100	9	in	in	ADP
brj-22592	100	10	the	the	DET
brj-22592	100	11	characteristic	characteristic	ADJ
brj-22592	100	12	matrix	matrix	NOUN
brj-22592	100	13	,	,	PUNCT
brj-22592	100	14	g	g	PROPN
brj-22592	100	15	denotes	denote	VERB
brj-22592	100	16	the	the	DET
brj-22592	100	17	group	group	NOUN
brj-22592	100	18	width	width	NOUN
brj-22592	100	19	in	in	ADP
brj-22592	100	20	the	the	DET
brj-22592	100	21	group	group	NOUN
brj-22592	100	22	convolution	convolution	NOUN
brj-22592	100	23	,	,	PUNCT
brj-22592	100	24	b	b	PROPN
brj-22592	100	25	denotes	denote	VERB
brj-22592	100	26	the	the	DET
brj-22592	100	27	bottleneck	bottleneck	NOUN
brj-22592	100	28	ratio	ratio	NOUN
brj-22592	100	29	,	,	PUNCT
brj-22592	100	30	and	and	CCONJ
brj-22592	100	31	the	the	DET
brj-22592	100	32	number	number	NOUN
brj-22592	100	33	of	of	ADP
brj-22592	100	34	channels	channel	NOUN
brj-22592	100	35	in	in	ADP
brj-22592	100	36	the	the	DET
brj-22592	100	37	output	output	NOUN
brj-22592	100	38	characteristic	characteristic	ADJ
brj-22592	100	39	matrix	matrix	NOUN
brj-22592	100	40	is	be	AUX
brj-22592	100	41	reduced	reduce	VERB
brj-22592	100	42	to	to	ADP
brj-22592	100	43	the	the	DET
brj-22592	100	44	number	number	NOUN
brj-22592	100	45	of	of	ADP
brj-22592	100	46	channels	channel	NOUN
brj-22592	100	47	1	1	NUM
brj-22592	100	48	/	/	SYM
brj-22592	100	49	b	b	NOUN
brj-22592	100	50	of	of	ADP
brj-22592	100	51	the	the	DET
brj-22592	100	52	output	output	NOUN
brj-22592	100	53	characteristic	characteristic	ADJ
brj-22592	100	54	matrix	matrix	NOUN
brj-22592	100	55	.	.	PUNCT
brj-22592	101	1	the	the	DET
brj-22592	101	2	image	image	NOUN
brj-22592	101	3	data	datum	NOUN
brj-22592	101	4	of	of	ADP
brj-22592	101	5	wood	wood	NOUN
brj-22592	101	6	defects	defect	NOUN
brj-22592	101	7	are	be	AUX
brj-22592	101	8	input	input	VERB
brj-22592	101	9	into	into	ADP
brj-22592	101	10	the	the	DET
brj-22592	101	11	group	group	NOUN
brj-22592	101	12	convolution	convolution	NOUN
brj-22592	101	13	module	module	NOUN
brj-22592	101	14	of	of	ADP
brj-22592	101	15	the	the	DET
brj-22592	101	16	stem	stem	NOUN
brj-22592	101	17	part	part	NOUN
brj-22592	101	18	of	of	ADP
brj-22592	101	19	the	the	DET
brj-22592	101	20	regnet	regnet	NOUN
brj-22592	101	21	network	network	NOUN
brj-22592	101	22	,	,	PUNCT
brj-22592	101	23	and	and	CCONJ
brj-22592	101	24	feature	feature	NOUN
brj-22592	101	25	extraction	extraction	NOUN
brj-22592	101	26	,	,	PUNCT
brj-22592	101	27	recognition	recognition	NOUN
brj-22592	101	28	,	,	PUNCT
brj-22592	101	29	and	and	CCONJ
brj-22592	101	30	classification	classification	NOUN
brj-22592	101	31	were	be	AUX
brj-22592	101	32	performed	perform	VERB
brj-22592	101	33	according	accord	VERB
brj-22592	101	34	to	to	ADP
brj-22592	101	35	the	the	DET
brj-22592	101	36	network	network	NOUN
brj-22592	101	37	structure	structure	NOUN
brj-22592	101	38	.	.	PUNCT
brj-22592	102	1	(	(	PUNCT
brj-22592	102	2	a	a	X
brj-22592	102	3	)	)	PUNCT
brj-22592	102	4	network	network	NOUN
brj-22592	102	5	(	(	PUNCT
brj-22592	102	6	b	b	NOUN
brj-22592	102	7	)	)	PUNCT
brj-22592	102	8	body	body	NOUN
brj-22592	102	9	(	(	PUNCT
brj-22592	102	10	c	c	NOUN
brj-22592	102	11	)	)	PUNCT
brj-22592	102	12	stage	stage	NOUN
brj-22592	102	13	i	i	PRON
brj-22592	102	14	peer	peer	NOUN
brj-22592	102	15	-	-	PUNCT
brj-22592	102	16	reviewed	review	VERB
brj-22592	102	17	article	article	NOUN
brj-22592	102	18	bioresources.com	bioresources.com	X
brj-22592	102	19	xie	xie	PROPN
brj-22592	102	20	&	&	CCONJ
brj-22592	102	21	ling	ling	PROPN
brj-22592	102	22	(	(	PUNCT
brj-22592	102	23	2023	2023	NUM
brj-22592	102	24	)	)	PUNCT
brj-22592	102	25	.	.	PUNCT
brj-22592	103	1	“	"	PUNCT
brj-22592	103	2	wood	wood	NOUN
brj-22592	103	3	defect	defect	NOUN
brj-22592	103	4	classification	classification	NOUN
brj-22592	103	5	,	,	PUNCT
brj-22592	103	6	”	"	PUNCT
brj-22592	103	7	bioresources	bioresource	NOUN
brj-22592	103	8	18(4	18(4	NUM
brj-22592	103	9	)	)	PUNCT
brj-22592	103	10	,	,	PUNCT
brj-22592	103	11	7663	7663	NUM
brj-22592	103	12	-	-	SYM
brj-22592	103	13	7680	7680	NUM
brj-22592	103	14	.	.	PUNCT
brj-22592	103	15	7668	7668	NUM
brj-22592	103	16	fig	fig	NOUN
brj-22592	103	17	.	.	PUNCT
brj-22592	104	1	4	4	X
brj-22592	104	2	.	.	NOUN
brj-22592	104	3	block	block	NOUN
brj-22592	104	4	module	module	NOUN
brj-22592	104	5	structure	structure	NOUN
brj-22592	104	6	in	in	ADP
brj-22592	104	7	regnet	regnet	PROPN
brj-22592	104	8	mode	mode	NOUN
brj-22592	104	9	improved	improve	VERB
brj-22592	104	10	regnet	regnet	NOUN
brj-22592	104	11	model	model	NOUN
brj-22592	104	12	to	to	PART
brj-22592	104	13	solve	solve	VERB
brj-22592	104	14	the	the	DET
brj-22592	104	15	problem	problem	NOUN
brj-22592	104	16	of	of	ADP
brj-22592	104	17	limited	limited	ADJ
brj-22592	104	18	computing	computing	NOUN
brj-22592	104	19	power	power	NOUN
brj-22592	104	20	,	,	PUNCT
brj-22592	104	21	computing	compute	VERB
brj-22592	104	22	resources	resource	NOUN
brj-22592	104	23	are	be	AUX
brj-22592	104	24	allocated	allocate	VERB
brj-22592	104	25	to	to	ADP
brj-22592	104	26	more	more	ADV
brj-22592	104	27	important	important	ADJ
brj-22592	104	28	tasks	task	NOUN
brj-22592	104	29	,	,	PUNCT
brj-22592	104	30	and	and	CCONJ
brj-22592	104	31	the	the	DET
brj-22592	104	32	problem	problem	NOUN
brj-22592	104	33	of	of	ADP
brj-22592	104	34	information	information	NOUN
brj-22592	104	35	overload	overload	NOUN
brj-22592	104	36	is	be	AUX
brj-22592	104	37	solved	solve	VERB
brj-22592	104	38	simultaneously	simultaneously	ADV
brj-22592	104	39	.	.	PUNCT
brj-22592	105	1	in	in	ADP
brj-22592	105	2	this	this	DET
brj-22592	105	3	study	study	NOUN
brj-22592	105	4	,	,	PUNCT
brj-22592	105	5	an	an	DET
brj-22592	105	6	attention	attention	NOUN
brj-22592	105	7	mechanism	mechanism	NOUN
brj-22592	105	8	module	module	NOUN
brj-22592	105	9	was	be	AUX
brj-22592	105	10	added	add	VERB
brj-22592	105	11	to	to	ADP
brj-22592	105	12	the	the	DET
brj-22592	105	13	regnet	regnet	NOUN
brj-22592	105	14	network	network	NOUN
brj-22592	105	15	.	.	PUNCT
brj-22592	106	1	in	in	ADP
brj-22592	106	2	computer	computer	NOUN
brj-22592	106	3	vision	vision	NOUN
brj-22592	106	4	,	,	PUNCT
brj-22592	106	5	the	the	DET
brj-22592	106	6	attention	attention	NOUN
brj-22592	106	7	mechanism	mechanism	NOUN
brj-22592	106	8	module	module	NOUN
brj-22592	106	9	was	be	AUX
brj-22592	106	10	mainly	mainly	ADV
brj-22592	106	11	divided	divide	VERB
brj-22592	106	12	into	into	ADP
brj-22592	106	13	a	a	DET
brj-22592	106	14	spatial	spatial	ADJ
brj-22592	106	15	attention	attention	NOUN
brj-22592	106	16	module	module	NOUN
brj-22592	106	17	,	,	PUNCT
brj-22592	106	18	channel	channel	NOUN
brj-22592	106	19	attention	attention	NOUN
brj-22592	106	20	module	module	NOUN
brj-22592	106	21	,	,	PUNCT
brj-22592	106	22	position	position	NOUN
brj-22592	106	23	pixel	pixel	ADJ
brj-22592	106	24	attention	attention	NOUN
brj-22592	106	25	module	module	NOUN
brj-22592	106	26	,	,	PUNCT
brj-22592	106	27	and	and	CCONJ
brj-22592	106	28	hybrid	hybrid	ADJ
brj-22592	106	29	attention	attention	NOUN
brj-22592	106	30	module	module	NOUN
brj-22592	106	31	.	.	PUNCT
brj-22592	107	1	the	the	DET
brj-22592	107	2	spatial	spatial	ADJ
brj-22592	107	3	attention	attention	NOUN
brj-22592	107	4	module	module	NOUN
brj-22592	107	5	adjusts	adjust	VERB
brj-22592	107	6	the	the	DET
brj-22592	107	7	self	self	NOUN
brj-22592	107	8	-	-	PUNCT
brj-22592	107	9	attention	attention	NOUN
brj-22592	107	10	of	of	ADP
brj-22592	107	11	each	each	DET
brj-22592	107	12	position	position	NOUN
brj-22592	107	13	of	of	ADP
brj-22592	107	14	the	the	DET
brj-22592	107	15	feature	feature	NOUN
brj-22592	107	16	map	map	NOUN
brj-22592	107	17	,	,	PUNCT
brj-22592	107	18	(	(	PUNCT
brj-22592	107	19	x	x	NOUN
brj-22592	107	20	,	,	PUNCT
brj-22592	107	21	y	y	NOUN
brj-22592	107	22	)	)	PUNCT
brj-22592	107	23	two	two	NUM
brj-22592	107	24	-	-	PUNCT
brj-22592	107	25	dimensional	dimensional	ADJ
brj-22592	107	26	adjustment	adjustment	NOUN
brj-22592	107	27	,	,	PUNCT
brj-22592	107	28	to	to	PART
brj-22592	107	29	ensure	ensure	VERB
brj-22592	107	30	that	that	SCONJ
brj-22592	107	31	the	the	DET
brj-22592	107	32	model	model	NOUN
brj-22592	107	33	focuses	focus	VERB
brj-22592	107	34	on	on	ADP
brj-22592	107	35	the	the	DET
brj-22592	107	36	areas	area	NOUN
brj-22592	107	37	worthy	worthy	ADJ
brj-22592	107	38	of	of	ADP
brj-22592	107	39	higher	high	ADJ
brj-22592	107	40	attention	attention	NOUN
brj-22592	107	41	.	.	PUNCT
brj-22592	108	1	the	the	DET
brj-22592	108	2	channel	channel	NOUN
brj-22592	108	3	attention	attention	NOUN
brj-22592	108	4	module	module	NOUN
brj-22592	108	5	allocates	allocate	VERB
brj-22592	108	6	resources	resource	NOUN
brj-22592	108	7	to	to	ADP
brj-22592	108	8	each	each	DET
brj-22592	108	9	convolution	convolution	NOUN
brj-22592	108	10	channel	channel	NOUN
brj-22592	108	11	and	and	CCONJ
brj-22592	108	12	adjusts	adjust	VERB
brj-22592	108	13	the	the	DET
brj-22592	108	14	z	z	NOUN
brj-22592	108	15	-	-	PUNCT
brj-22592	108	16	axis	axis	NOUN
brj-22592	108	17	in	in	ADP
brj-22592	108	18	one	one	NUM
brj-22592	108	19	dimension	dimension	NOUN
brj-22592	108	20	.	.	PUNCT
brj-22592	109	1	the	the	DET
brj-22592	109	2	position	position	NOUN
brj-22592	109	3	pixel	pixel	VERB
brj-22592	109	4	attention	attention	NOUN
brj-22592	109	5	module	module	NOUN
brj-22592	109	6	focuses	focus	VERB
brj-22592	109	7	more	more	ADJ
brj-22592	109	8	on	on	ADP
brj-22592	109	9	the	the	DET
brj-22592	109	10	correlation	correlation	NOUN
brj-22592	109	11	between	between	ADP
brj-22592	109	12	the	the	DET
brj-22592	109	13	pixels	pixel	NOUN
brj-22592	109	14	on	on	ADP
brj-22592	109	15	the	the	DET
brj-22592	109	16	feature	feature	NOUN
brj-22592	109	17	map	map	NOUN
brj-22592	109	18	and	and	CCONJ
brj-22592	109	19	other	other	ADJ
brj-22592	109	20	pixels	pixel	NOUN
brj-22592	109	21	and	and	CCONJ
brj-22592	109	22	realizes	realize	VERB
brj-22592	109	23	a	a	DET
brj-22592	109	24	global	global	ADJ
brj-22592	109	25	response	response	NOUN
brj-22592	109	26	to	to	ADP
brj-22592	109	27	the	the	DET
brj-22592	109	28	output	output	NOUN
brj-22592	109	29	.	.	PUNCT
brj-22592	110	1	squeeze	squeeze	NOUN
brj-22592	110	2	-	-	PUNCT
brj-22592	110	3	and	and	CCONJ
brj-22592	110	4	-	-	PUNCT
brj-22592	110	5	excitation	excitation	NOUN
brj-22592	110	6	networks	network	NOUN
brj-22592	110	7	belong	belong	VERB
brj-22592	110	8	to	to	ADP
brj-22592	110	9	the	the	DET
brj-22592	110	10	channel	channel	NOUN
brj-22592	110	11	attention	attention	NOUN
brj-22592	110	12	module	module	NOUN
brj-22592	110	13	.	.	PUNCT
brj-22592	111	1	the	the	DET
brj-22592	111	2	attention	attention	NOUN
brj-22592	111	3	mechanism	mechanism	NOUN
brj-22592	111	4	was	be	AUX
brj-22592	111	5	divided	divide	VERB
brj-22592	111	6	into	into	ADP
brj-22592	111	7	two	two	NUM
brj-22592	111	8	steps	step	NOUN
brj-22592	111	9	:	:	PUNCT
brj-22592	111	10	squeeze	squeeze	NOUN
brj-22592	111	11	and	and	CCONJ
brj-22592	111	12	exception	exception	NOUN
brj-22592	111	13	.	.	PUNCT
brj-22592	112	1	in	in	ADP
brj-22592	112	2	squeeze	squeeze	NOUN
brj-22592	112	3	step	step	NOUN
brj-22592	112	4	,	,	PUNCT
brj-22592	112	5	the	the	DET
brj-22592	112	6	global	global	ADJ
brj-22592	112	7	compressed	compress	VERB
brj-22592	112	8	feature	feature	NOUN
brj-22592	112	9	of	of	ADP
brj-22592	112	10	the	the	DET
brj-22592	112	11	current	current	ADJ
brj-22592	112	12	feature	feature	NOUN
brj-22592	112	13	map	map	NOUN
brj-22592	112	14	is	be	AUX
brj-22592	112	15	obtained	obtain	VERB
brj-22592	112	16	by	by	ADP
brj-22592	112	17	performing	perform	VERB
brj-22592	112	18	global	global	ADJ
brj-22592	112	19	average	average	ADJ
brj-22592	112	20	pooling	pooling	NOUN
brj-22592	112	21	on	on	ADP
brj-22592	112	22	the	the	DET
brj-22592	112	23	feature	feature	NOUN
brj-22592	112	24	map	map	NOUN
brj-22592	112	25	layer	layer	NOUN
brj-22592	112	26	.	.	PUNCT
brj-22592	113	1	the	the	DET
brj-22592	113	2	excitation	excitation	NOUN
brj-22592	113	3	step	step	NOUN
brj-22592	113	4	is	be	AUX
brj-22592	113	5	used	use	VERB
brj-22592	113	6	to	to	PART
brj-22592	113	7	obtain	obtain	VERB
brj-22592	113	8	the	the	DET
brj-22592	113	9	weight	weight	NOUN
brj-22592	113	10	of	of	ADP
brj-22592	113	11	each	each	DET
brj-22592	113	12	channel	channel	NOUN
brj-22592	113	13	in	in	ADP
brj-22592	113	14	the	the	DET
brj-22592	113	15	feature	feature	NOUN
brj-22592	113	16	map	map	NOUN
brj-22592	113	17	via	via	ADP
brj-22592	113	18	the	the	DET
brj-22592	113	19	two	two	NUM
brj-22592	113	20	-	-	PUNCT
brj-22592	113	21	layer	layer	NOUN
brj-22592	113	22	fully	fully	ADV
brj-22592	113	23	connected	connected	ADJ
brj-22592	113	24	bottleneck	bottleneck	NOUN
brj-22592	113	25	structure	structure	NOUN
brj-22592	113	26	.	.	PUNCT
brj-22592	114	1	furthermore	furthermore	ADV
brj-22592	114	2	,	,	PUNCT
brj-22592	114	3	weighted	weight	VERB
brj-22592	114	4	feature	feature	NOUN
brj-22592	114	5	map	map	NOUN
brj-22592	114	6	is	be	AUX
brj-22592	114	7	used	use	VERB
brj-22592	114	8	as	as	ADP
brj-22592	114	9	the	the	DET
brj-22592	114	10	input	input	NOUN
brj-22592	114	11	of	of	ADP
brj-22592	114	12	the	the	DET
brj-22592	114	13	next	next	ADJ
brj-22592	114	14	layer	layer	NOUN
brj-22592	114	15	network	network	NOUN
brj-22592	114	16	.	.	PUNCT
brj-22592	115	1	efficient	efficient	ADJ
brj-22592	115	2	channel	channel	PROPN
brj-22592	115	3	attention	attention	NOUN
brj-22592	115	4	also	also	ADV
brj-22592	115	5	belongs	belong	VERB
brj-22592	115	6	to	to	ADP
brj-22592	115	7	the	the	DET
brj-22592	115	8	channel	channel	NOUN
brj-22592	115	9	attention	attention	NOUN
brj-22592	115	10	mechanism	mechanism	NOUN
brj-22592	115	11	.	.	PUNCT
brj-22592	116	1	efficient	efficient	ADJ
brj-22592	116	2	channel	channel	PROPN
brj-22592	116	3	attention	attention	NOUN
brj-22592	116	4	avoids	avoid	VERB
brj-22592	116	5	the	the	DET
brj-22592	116	6	dimensionality	dimensionality	NOUN
brj-22592	116	7	reduction	reduction	NOUN
brj-22592	116	8	due	due	ADP
brj-22592	116	9	to	to	ADP
brj-22592	116	10	channel	channel	NOUN
brj-22592	116	11	compression	compression	NOUN
brj-22592	116	12	,	,	PUNCT
brj-22592	116	13	uses	use	VERB
brj-22592	116	14	1d	1d	NUM
brj-22592	116	15	convolution	convolution	NOUN
brj-22592	116	16	to	to	PART
brj-22592	116	17	efficiently	efficiently	ADV
brj-22592	116	18	realize	realize	VERB
brj-22592	116	19	local	local	ADJ
brj-22592	116	20	cross	cross	ADJ
brj-22592	116	21	-	-	ADJ
brj-22592	116	22	channel	channel	ADJ
brj-22592	116	23	interaction	interaction	NOUN
brj-22592	116	24	,	,	PUNCT
brj-22592	116	25	and	and	CCONJ
brj-22592	116	26	extracts	extract	VERB
brj-22592	116	27	the	the	DET
brj-22592	116	28	dependencies	dependency	NOUN
brj-22592	116	29	between	between	ADP
brj-22592	116	30	channels	channel	NOUN
brj-22592	116	31	.	.	PUNCT
brj-22592	117	1	after	after	SCONJ
brj-22592	117	2	the	the	DET
brj-22592	117	3	group	group	NOUN
brj-22592	117	4	convolution	convolution	NOUN
brj-22592	117	5	structure	structure	NOUN
brj-22592	117	6	in	in	ADP
brj-22592	117	7	the	the	DET
brj-22592	117	8	block	block	NOUN
brj-22592	117	9	module	module	NOUN
brj-22592	117	10	of	of	ADP
brj-22592	117	11	the	the	DET
brj-22592	117	12	regnet	regnet	NOUN
brj-22592	117	13	model	model	NOUN
brj-22592	117	14	network	network	NOUN
brj-22592	117	15	,	,	PUNCT
brj-22592	117	16	in	in	ADP
brj-22592	117	17	this	this	DET
brj-22592	117	18	study	study	NOUN
brj-22592	117	19	,	,	PUNCT
brj-22592	117	20	the	the	DET
brj-22592	117	21	attention	attention	NOUN
brj-22592	117	22	mechanism	mechanism	NOUN
brj-22592	117	23	module	module	NOUN
brj-22592	117	24	nam	nam	NOUN
brj-22592	117	25	(	(	PUNCT
brj-22592	117	26	liu	liu	PROPN
brj-22592	117	27	et	et	PROPN
brj-22592	117	28	al	al	PROPN
brj-22592	117	29	.	.	PROPN
brj-22592	117	30	2021	2021	NUM
brj-22592	117	31	)	)	PUNCT
brj-22592	117	32	was	be	AUX
brj-22592	117	33	added	add	VERB
brj-22592	117	34	.	.	PUNCT
brj-22592	118	1	specifically	specifically	ADV
brj-22592	118	2	,	,	PUNCT
brj-22592	118	3	nam	nam	NOUN
brj-22592	118	4	adopts	adopt	VERB
brj-22592	118	5	the	the	DET
brj-22592	118	6	module	module	NOUN
brj-22592	118	7	integration	integration	NOUN
brj-22592	118	8	mode	mode	NOUN
brj-22592	118	9	,	,	PUNCT
brj-22592	118	10	including	include	VERB
brj-22592	118	11	the	the	DET
brj-22592	118	12	channel	channel	NOUN
brj-22592	118	13	attention	attention	NOUN
brj-22592	118	14	submodule	submodule	NOUN
brj-22592	118	15	and	and	CCONJ
brj-22592	118	16	spatial	spatial	ADJ
brj-22592	118	17	attention	attention	NOUN
brj-22592	118	18	submodule	submodule	NOUN
brj-22592	118	19	.	.	PUNCT
brj-22592	119	1	a	a	DET
brj-22592	119	2	schematic	schematic	ADJ
brj-22592	119	3	diagram	diagram	NOUN
brj-22592	119	4	of	of	ADP
brj-22592	119	5	the	the	DET
brj-22592	119	6	channel	channel	NOUN
brj-22592	119	7	attention	attention	NOUN
brj-22592	119	8	of	of	ADP
brj-22592	119	9	the	the	DET
brj-22592	119	10	nam	nam	NOUN
brj-22592	119	11	module	module	NOUN
brj-22592	119	12	is	be	AUX
brj-22592	119	13	shown	show	VERB
brj-22592	119	14	in	in	ADP
brj-22592	119	15	fig	fig	NOUN
brj-22592	119	16	.	.	PUNCT
brj-22592	120	1	5(a	5(a	NUM
brj-22592	120	2	)	)	PUNCT
brj-22592	120	3	,	,	PUNCT
brj-22592	120	4	and	and	CCONJ
brj-22592	120	5	a	a	DET
brj-22592	120	6	spatial	spatial	ADJ
brj-22592	120	7	attention	attention	NOUN
brj-22592	120	8	schematic	schematic	ADJ
brj-22592	120	9	diagram	diagram	NOUN
brj-22592	120	10	is	be	AUX
brj-22592	120	11	shown	show	VERB
brj-22592	120	12	in	in	ADP
brj-22592	120	13	fig	fig	NOUN
brj-22592	120	14	.	.	PUNCT
brj-22592	121	1	5(b	5(b	NUM
brj-22592	121	2	)	)	PUNCT
brj-22592	121	3	.	.	PUNCT
brj-22592	122	1	(	(	PUNCT
brj-22592	122	2	a	a	X
brj-22592	122	3	)	)	PUNCT
brj-22592	122	4	x	x	SYM
brj-22592	122	5	block	block	NOUN
brj-22592	122	6	,	,	PUNCT
brj-22592	122	7	s=1	s=1	X
brj-22592	122	8	(	(	PUNCT
brj-22592	122	9	b	b	NOUN
brj-22592	122	10	)	)	PUNCT
brj-22592	122	11	x	x	SYM
brj-22592	122	12	block	block	NOUN
brj-22592	122	13	,	,	PUNCT
brj-22592	122	14	s=2	s=2	PRON
brj-22592	122	15	peer	peer	NOUN
brj-22592	122	16	-	-	PUNCT
brj-22592	122	17	reviewed	review	VERB
brj-22592	122	18	article	article	NOUN
brj-22592	122	19	bioresources.com	bioresources.com	X
brj-22592	122	20	xie	xie	PROPN
brj-22592	122	21	&	&	CCONJ
brj-22592	122	22	ling	ling	PROPN
brj-22592	122	23	(	(	PUNCT
brj-22592	122	24	2023	2023	NUM
brj-22592	122	25	)	)	PUNCT
brj-22592	122	26	.	.	PUNCT
brj-22592	123	1	“	"	PUNCT
brj-22592	123	2	wood	wood	NOUN
brj-22592	123	3	defect	defect	NOUN
brj-22592	123	4	classification	classification	NOUN
brj-22592	123	5	,	,	PUNCT
brj-22592	123	6	”	"	PUNCT
brj-22592	123	7	bioresources	bioresource	NOUN
brj-22592	123	8	18(4	18(4	NUM
brj-22592	123	9	)	)	PUNCT
brj-22592	123	10	,	,	PUNCT
brj-22592	123	11	7663	7663	NUM
brj-22592	123	12	-	-	SYM
brj-22592	123	13	7680	7680	NUM
brj-22592	123	14	.	.	PUNCT
brj-22592	124	1	7669	7669	NUM
brj-22592	124	2	fig	fig	NOUN
brj-22592	124	3	.	.	PUNCT
brj-22592	125	1	5	5	X
brj-22592	125	2	.	.	X
brj-22592	125	3	attention	attention	NOUN
brj-22592	125	4	sub	sub	NOUN
brj-22592	125	5	module	module	NOUN
brj-22592	125	6	of	of	ADP
brj-22592	125	7	nam	nam	NOUN
brj-22592	125	8	module	module	NOUN
brj-22592	125	9	in	in	ADP
brj-22592	125	10	the	the	DET
brj-22592	125	11	channel	channel	NOUN
brj-22592	125	12	attention	attention	NOUN
brj-22592	125	13	submodule	submodule	NOUN
brj-22592	125	14	,	,	PUNCT
brj-22592	125	15	the	the	DET
brj-22592	125	16	scaling	scaling	ADJ
brj-22592	125	17	factor	factor	NOUN
brj-22592	125	18	is	be	AUX
brj-22592	125	19	added	add	VERB
brj-22592	125	20	in	in	ADP
brj-22592	125	21	batch	batch	NOUN
brj-22592	125	22	normalization	normalization	NOUN
brj-22592	125	23	.	.	PUNCT
brj-22592	126	1	the	the	DET
brj-22592	126	2	scaling	scale	VERB
brj-22592	126	3	factor	factor	NOUN
brj-22592	126	4	reflects	reflect	VERB
brj-22592	126	5	the	the	DET
brj-22592	126	6	degree	degree	NOUN
brj-22592	126	7	of	of	ADP
brj-22592	126	8	change	change	NOUN
brj-22592	126	9	in	in	ADP
brj-22592	126	10	each	each	DET
brj-22592	126	11	channel	channel	NOUN
brj-22592	126	12	and	and	CCONJ
brj-22592	126	13	its	its	PRON
brj-22592	126	14	importance	importance	NOUN
brj-22592	126	15	.	.	PUNCT
brj-22592	127	1	the	the	DET
brj-22592	127	2	mathematical	mathematical	ADJ
brj-22592	127	3	relationship	relationship	NOUN
brj-22592	127	4	of	of	ADP
brj-22592	127	5	the	the	DET
brj-22592	127	6	scaling	scale	VERB
brj-22592	127	7	factor	factor	NOUN
brj-22592	127	8	is	be	AUX
brj-22592	127	9	as	as	SCONJ
brj-22592	127	10	follows	follow	VERB
brj-22592	127	11	,	,	PUNCT
brj-22592	127	12	β	β	X
brj-22592	127	13	εσ	εσ	VERB
brj-22592	127	14	μb	μb	PROPN
brj-22592	127	15	γ)b(bnb	γ)b(bnb	PROPN
brj-22592	127	16	2	2	NUM
brj-22592	127	17	b	b	PROPN
brj-22592	127	18	bin	bin	PROPN
brj-22592	127	19	inout	inout	PROPN
brj-22592	128	1	+	+	CCONJ
brj-22592	129	1	+	+	NUM
brj-22592	129	2	−	−	PROPN
brj-22592	129	3	=	=	SYM
brj-22592	129	4	=	=	SYM
brj-22592	129	5	(	(	PUNCT
brj-22592	129	6	1	1	NUM
brj-22592	129	7	)	)	PUNCT
brj-22592	129	8	where	where	SCONJ
brj-22592	129	9	bin	bin	NOUN
brj-22592	129	10	and	and	CCONJ
brj-22592	129	11	bout	bout	NOUN
brj-22592	129	12	denote	denote	NOUN
brj-22592	129	13	batch	batch	NOUN
brj-22592	129	14	input	input	NOUN
brj-22592	129	15	and	and	CCONJ
brj-22592	129	16	output	output	NOUN
brj-22592	129	17	feature	feature	NOUN
brj-22592	129	18	data	datum	NOUN
brj-22592	129	19	;	;	PUNCT
brj-22592	129	20	)	)	PUNCT
brj-22592	129	21	(	(	PUNCT
brj-22592	129	22	bn	bn	X
brj-22592	129	23			PROPN
brj-22592	129	24	denotes	denote	NOUN
brj-22592	129	25	the	the	DET
brj-22592	129	26	batch	batch	NOUN
brj-22592	129	27	normalization	normalization	NOUN
brj-22592	129	28	function	function	NOUN
brj-22592	129	29	;	;	PUNCT
brj-22592	129	30	bμ	bμ	CCONJ
brj-22592	129	31	and	and	CCONJ
brj-22592	129	32	bσ	bσ	PROPN
brj-22592	129	33	denote	denote	VERB
brj-22592	129	34	the	the	DET
brj-22592	129	35	average	average	ADJ
brj-22592	129	36	value	value	NOUN
brj-22592	129	37	and	and	CCONJ
brj-22592	129	38	standard	standard	ADJ
brj-22592	129	39	deviation	deviation	NOUN
brj-22592	129	40	of	of	ADP
brj-22592	129	41	batch	batch	NOUN
brj-22592	129	42	b	b	NOUN
brj-22592	129	43	,	,	PUNCT
brj-22592	129	44	respectively	respectively	ADV
brj-22592	129	45	;	;	PUNCT
brj-22592	129	46	γ	γ	PROPN
brj-22592	129	47	denotes	denote	VERB
brj-22592	129	48	the	the	DET
brj-22592	129	49	scale	scale	NOUN
brj-22592	129	50	transformation	transformation	NOUN
brj-22592	129	51	coefficient	coefficient	NOUN
brj-22592	129	52	of	of	ADP
brj-22592	129	53	affine	affine	NOUN
brj-22592	129	54	transformation	transformation	NOUN
brj-22592	129	55	;	;	PUNCT
brj-22592	129	56	β	β	X
brj-22592	129	57	denotes	denote	VERB
brj-22592	129	58	the	the	DET
brj-22592	129	59	displacement	displacement	ADJ
brj-22592	129	60	transformation	transformation	NOUN
brj-22592	129	61	coefficient	coefficient	NOUN
brj-22592	129	62	of	of	ADP
brj-22592	129	63	affine	affine	NOUN
brj-22592	129	64	transformation	transformation	NOUN
brj-22592	129	65	;	;	PUNCT
brj-22592	129	66	and	and	CCONJ
brj-22592	129	67	ε	ε	PROPN
brj-22592	129	68	prevents	prevent	VERB
brj-22592	129	69	the	the	DET
brj-22592	129	70	variance	variance	NOUN
brj-22592	129	71	from	from	ADP
brj-22592	129	72	being	be	AUX
brj-22592	129	73	0	0	NUM
brj-22592	129	74	,	,	PUNCT
brj-22592	129	75	resulting	result	VERB
brj-22592	129	76	in	in	ADP
brj-22592	129	77	invalid	invalid	ADJ
brj-22592	129	78	calculation	calculation	NOUN
brj-22592	129	79	.	.	PUNCT
brj-22592	130	1	the	the	DET
brj-22592	130	2	calculation	calculation	NOUN
brj-22592	130	3	principle	principle	NOUN
brj-22592	130	4	of	of	ADP
brj-22592	130	5	the	the	DET
brj-22592	130	6	channel	channel	NOUN
brj-22592	130	7	attention	attention	NOUN
brj-22592	130	8	submodule	submodule	NOUN
brj-22592	130	9	is	be	AUX
brj-22592	130	10	as	as	SCONJ
brj-22592	130	11	follows	follow	VERB
brj-22592	130	12	:	:	PUNCT
brj-22592	130	13	)	)	PUNCT
brj-22592	130	14	)	)	PUNCT
brj-22592	130	15	)	)	PUNCT
brj-22592	131	1	f(bn(w(sigmoidm	f(bn(w(sigmoidm	NOUN
brj-22592	131	2	1γc	1γc	NOUN
brj-22592	131	3	=	=	PUNCT
brj-22592	131	4	(	(	PUNCT
brj-22592	131	5	2	2	NUM
brj-22592	131	6	)	)	PUNCT
brj-22592	131	7	peer	peer	NOUN
brj-22592	131	8	-	-	PUNCT
brj-22592	131	9	reviewed	review	VERB
brj-22592	131	10	article	article	NOUN
brj-22592	131	11	bioresources.com	bioresources.com	X
brj-22592	131	12	xie	xie	PROPN
brj-22592	131	13	&	&	CCONJ
brj-22592	131	14	ling	ling	PROPN
brj-22592	131	15	(	(	PUNCT
brj-22592	131	16	2023	2023	NUM
brj-22592	131	17	)	)	PUNCT
brj-22592	131	18	.	.	PUNCT
brj-22592	132	1	“	"	PUNCT
brj-22592	132	2	wood	wood	NOUN
brj-22592	132	3	defect	defect	NOUN
brj-22592	132	4	classification	classification	NOUN
brj-22592	132	5	,	,	PUNCT
brj-22592	132	6	”	"	PUNCT
brj-22592	132	7	bioresources	bioresource	NOUN
brj-22592	132	8	18(4	18(4	NUM
brj-22592	132	9	)	)	PUNCT
brj-22592	132	10	,	,	PUNCT
brj-22592	132	11	7663	7663	NUM
brj-22592	132	12	-	-	SYM
brj-22592	132	13	7680	7680	NUM
brj-22592	132	14	.	.	PUNCT
brj-22592	133	1	7670	7670	NUM
brj-22592	133	2	where	where	SCONJ
brj-22592	133	3	f1	f1	NOUN
brj-22592	133	4	denotes	denote	VERB
brj-22592	133	5	the	the	DET
brj-22592	133	6	input	input	NOUN
brj-22592	133	7	feature	feature	NOUN
brj-22592	134	1	;	;	PUNCT
brj-22592	134	2	mc	mc	PROPN
brj-22592	134	3	denotes	denote	VERB
brj-22592	134	4	the	the	DET
brj-22592	134	5	output	output	NOUN
brj-22592	134	6	feature	feature	NOUN
brj-22592	134	7	;	;	PUNCT
brj-22592	134	8	γ	γ	PROPN
brj-22592	134	9	denotes	denote	VERB
brj-22592	134	10	the	the	DET
brj-22592	134	11	scale	scale	NOUN
brj-22592	134	12	factor	factor	NOUN
brj-22592	134	13	of	of	ADP
brj-22592	134	14	each	each	DET
brj-22592	134	15	channel	channel	NOUN
brj-22592	134	16	;	;	PUNCT
brj-22592	134	17	)	)	PUNCT
brj-22592	134	18	(	(	PUNCT
brj-22592	134	19	bn	bn	X
brj-22592	134	20			NUM
brj-22592	134	21	denote	denote	VERB
brj-22592	134	22	the	the	DET
brj-22592	134	23	batch	batch	NOUN
brj-22592	134	24	normalization	normalization	NOUN
brj-22592	134	25	function	function	NOUN
brj-22592	134	26	;	;	PUNCT
brj-22592	134	27	)	)	PUNCT
brj-22592	134	28	(	(	PUNCT
brj-22592	134	29	sigmoid	sigmoid	NOUN
brj-22592	134	30			PROPN
brj-22592	134	31	denotes	denote	VERB
brj-22592	134	32	the	the	DET
brj-22592	134	33	sigmoid	sigmoid	NOUN
brj-22592	134	34	function	function	NOUN
brj-22592	134	35	;	;	PUNCT
brj-22592	134	36	and	and	CCONJ
brj-22592	134	37	the	the	DET
brj-22592	134	38	weight	weight	NOUN
brj-22592	134	39	γw	γw	PRON
brj-22592	134	40	represents	represent	VERB
brj-22592	134	41	the	the	DET
brj-22592	134	42	scale	scale	NOUN
brj-22592	134	43	factor	factor	NOUN
brj-22592	134	44	of	of	ADP
brj-22592	134	45	the	the	DET
brj-22592	134	46	bn	bn	NOUN
brj-22592	134	47	and	and	CCONJ
brj-22592	134	48	importance	importance	NOUN
brj-22592	134	49	of	of	ADP
brj-22592	134	50	the	the	DET
brj-22592	134	51	pixels	pixel	NOUN
brj-22592	134	52	.	.	PUNCT
brj-22592	135	1	the	the	DET
brj-22592	135	2	calculation	calculation	NOUN
brj-22592	135	3	formula	formula	NOUN
brj-22592	135	4	can	can	AUX
brj-22592	135	5	be	be	AUX
brj-22592	135	6	expressed	express	VERB
brj-22592	135	7	as	as	SCONJ
brj-22592	135	8	follows	follow	VERB
brj-22592	135	9	,	,	PUNCT
brj-22592	135	10			X
brj-22592	135	11	=	=	PUNCT
brj-22592	136	1	=	=	SYM
brj-22592	136	2	0j	0j	NOUN
brj-22592	137	1	j	j	PROPN
brj-22592	137	2	i	i	PRON
brj-22592	137	3	γ	γ	PROPN
brj-22592	137	4	γ	γ	X
brj-22592	137	5	γ	γ	X
brj-22592	137	6	w	w	PROPN
brj-22592	137	7	(	(	PUNCT
brj-22592	137	8	3	3	NUM
brj-22592	137	9	)	)	PUNCT
brj-22592	137	10	where	where	SCONJ
brj-22592	137	11	ji	ji	PROPN
brj-22592	137	12	γ	γ	PROPN
brj-22592	137	13	,	,	PUNCT
brj-22592	137	14	γ	γ	PROPN
brj-22592	137	15	denotes	denote	VERB
brj-22592	137	16	the	the	DET
brj-22592	137	17	scale	scale	NOUN
brj-22592	137	18	transformation	transformation	NOUN
brj-22592	137	19	coefficient	coefficient	NOUN
brj-22592	137	20	of	of	ADP
brj-22592	137	21	affine	affine	NOUN
brj-22592	137	22	transformation	transformation	NOUN
brj-22592	137	23	.	.	PUNCT
brj-22592	138	1	the	the	DET
brj-22592	138	2	calculation	calculation	NOUN
brj-22592	138	3	principle	principle	NOUN
brj-22592	138	4	of	of	ADP
brj-22592	138	5	the	the	DET
brj-22592	138	6	spatial	spatial	ADJ
brj-22592	138	7	attention	attention	NOUN
brj-22592	138	8	submodule	submodule	NOUN
brj-22592	138	9	is	be	AUX
brj-22592	138	10	shown	show	VERB
brj-22592	138	11	in	in	ADP
brj-22592	138	12	as	as	SCONJ
brj-22592	138	13	follows	follow	VERB
brj-22592	138	14	,	,	PUNCT
brj-22592	138	15	)	)	PUNCT
brj-22592	138	16	)	)	PUNCT
brj-22592	138	17	)	)	PUNCT
brj-22592	138	18	f(bn(w(sigmoidm	f(bn(w(sigmoidm	NOUN
brj-22592	138	19	2sλs	2sλs	NUM
brj-22592	138	20	=	=	SYM
brj-22592	138	21	(	(	PUNCT
brj-22592	138	22	4	4	NUM
brj-22592	138	23	)	)	PUNCT
brj-22592	138	24	where	where	SCONJ
brj-22592	138	25	f2	f2	PROPN
brj-22592	138	26	denotes	denote	VERB
brj-22592	138	27	the	the	DET
brj-22592	138	28	input	input	NOUN
brj-22592	138	29	feature	feature	NOUN
brj-22592	138	30	;	;	PUNCT
brj-22592	138	31	ms	ms	PROPN
brj-22592	138	32	denotes	denote	VERB
brj-22592	138	33	the	the	DET
brj-22592	138	34	output	output	NOUN
brj-22592	138	35	feature	feature	NOUN
brj-22592	138	36	;	;	PUNCT
brj-22592	138	37	)	)	PUNCT
brj-22592	138	38	(	(	PUNCT
brj-22592	138	39	bns	bns	PROPN
brj-22592	138	40			PROPN
brj-22592	138	41	denotes	denote	NOUN
brj-22592	138	42	the	the	DET
brj-22592	138	43	batch	batch	NOUN
brj-22592	138	44	normalization	normalization	NOUN
brj-22592	138	45	function	function	NOUN
brj-22592	138	46	;	;	PUNCT
brj-22592	138	47	)	)	PUNCT
brj-22592	138	48	(	(	PUNCT
brj-22592	138	49	sigmoid	sigmoid	NOUN
brj-22592	138	50			PROPN
brj-22592	138	51	denotes	denote	VERB
brj-22592	138	52	the	the	DET
brj-22592	138	53	sigmoid	sigmoid	NOUN
brj-22592	138	54	function	function	NOUN
brj-22592	138	55	;	;	PUNCT
brj-22592	138	56	λ	λ	PROPN
brj-22592	138	57	denotes	denote	VERB
brj-22592	138	58	the	the	DET
brj-22592	138	59	scale	scale	NOUN
brj-22592	138	60	factor	factor	NOUN
brj-22592	138	61	of	of	ADP
brj-22592	138	62	each	each	DET
brj-22592	138	63	channel	channel	NOUN
brj-22592	138	64	,	,	PUNCT
brj-22592	138	65	and	and	CCONJ
brj-22592	138	66	the	the	DET
brj-22592	138	67	weight	weight	NOUN
brj-22592	138	68	λw	λw	X
brj-22592	138	69	is	be	AUX
brj-22592	138	70	as	as	SCONJ
brj-22592	138	71	follows	follow	VERB
brj-22592	138	72	:	:	PUNCT
brj-22592	138	73			X
brj-22592	138	74	=	=	PUNCT
brj-22592	139	1	=	=	SYM
brj-22592	139	2	0j	0j	NOUN
brj-22592	140	1	j	j	NOUN
brj-22592	141	1	i	i	PRON
brj-22592	141	2	λ	λ	X
brj-22592	141	3	λ	λ	X
brj-22592	141	4	λ	λ	X
brj-22592	141	5	w	w	PROPN
brj-22592	141	6	(	(	PUNCT
brj-22592	141	7	5	5	NUM
brj-22592	141	8	)	)	PUNCT
brj-22592	142	1	to	to	PART
brj-22592	142	2	suppress	suppress	VERB
brj-22592	142	3	the	the	DET
brj-22592	142	4	small	small	ADJ
brj-22592	142	5	weight	weight	NOUN
brj-22592	142	6	and	and	CCONJ
brj-22592	142	7	prevent	prevent	VERB
brj-22592	142	8	overfitting	overfitting	NOUN
brj-22592	142	9	,	,	PUNCT
brj-22592	142	10	a	a	DET
brj-22592	142	11	regularization	regularization	NOUN
brj-22592	142	12	term	term	NOUN
brj-22592	142	13	is	be	AUX
brj-22592	142	14	added	add	VERB
brj-22592	142	15	to	to	ADP
brj-22592	142	16	the	the	DET
brj-22592	142	17	loss	loss	NOUN
brj-22592	142	18	function	function	NOUN
brj-22592	142	19	,	,	PUNCT
brj-22592	142	20	as	as	SCONJ
brj-22592	142	21	shown	show	VERB
brj-22592	142	22	in	in	ADP
brj-22592	142	23	eq	eq	ADJ
brj-22592	142	24	.	.	PROPN
brj-22592	142	25	6	6	NUM
brj-22592	142	26	.	.	X
brj-22592	143	1	in	in	ADP
brj-22592	143	2	the	the	DET
brj-22592	143	3	equation	equation	NOUN
brj-22592	143	4	,	,	PUNCT
brj-22592	143	5	x	x	PRON
brj-22592	143	6	denotes	denote	VERB
brj-22592	143	7	the	the	DET
brj-22592	143	8	input	input	NOUN
brj-22592	143	9	,	,	PUNCT
brj-22592	143	10	y	y	PROPN
brj-22592	143	11	denotes	denote	VERB
brj-22592	143	12	the	the	DET
brj-22592	143	13	output	output	NOUN
brj-22592	143	14	,	,	PUNCT
brj-22592	143	15	w	w	PROPN
brj-22592	143	16	denotes	denote	NOUN
brj-22592	143	17	the	the	DET
brj-22592	143	18	network	network	NOUN
brj-22592	143	19	weight	weight	NOUN
brj-22592	143	20	,	,	PUNCT
brj-22592	143	21	loss	loss	NOUN
brj-22592	143	22	denotes	denote	VERB
brj-22592	143	23	the	the	DET
brj-22592	143	24	loss	loss	NOUN
brj-22592	143	25	data	datum	NOUN
brj-22592	143	26	,	,	PUNCT
brj-22592	143	27	)	)	PUNCT
brj-22592	143	28	(	(	PUNCT
brj-22592	143	29	l	l	X
brj-22592	143	30			PROPN
brj-22592	143	31	denotes	denote	VERB
brj-22592	143	32	the	the	DET
brj-22592	143	33	loss	loss	NOUN
brj-22592	143	34	function	function	NOUN
brj-22592	143	35	,	,	PUNCT
brj-22592	143	36	)	)	PUNCT
brj-22592	144	1	(	(	PUNCT
brj-22592	144	2	f	f	PROPN
brj-22592	144	3			PROPN
brj-22592	144	4	denotes	denote	VERB
brj-22592	144	5	the	the	DET
brj-22592	144	6	activate	activate	NOUN
brj-22592	144	7	function	function	NOUN
brj-22592	144	8	,	,	PUNCT
brj-22592	144	9	)	)	PUNCT
brj-22592	144	10	(	(	PUNCT
brj-22592	144	11	g	g	PROPN
brj-22592	144	12			PROPN
brj-22592	144	13	denotes	denote	VERB
brj-22592	144	14	the	the	DET
brj-22592	144	15	l1	l1	PROPN
brj-22592	144	16	norm	norm	PROPN
brj-22592	144	17	penalty	penalty	NOUN
brj-22592	144	18	function	function	NOUN
brj-22592	144	19	,	,	PUNCT
brj-22592	144	20	and	and	CCONJ
brj-22592	144	21	p	p	NOUN
brj-22592	144	22	denotes	denote	VERB
brj-22592	144	23	the	the	DET
brj-22592	144	24	punishment	punishment	NOUN
brj-22592	144	25	for	for	ADP
brj-22592	144	26	balancing	balance	VERB
brj-22592	144	27	)	)	PUNCT
brj-22592	144	28	γ(g	γ(g	PROPN
brj-22592	144	29	and	and	CCONJ
brj-22592	144	30	)	)	PUNCT
brj-22592	144	31	λ(g	λ(g	NOUN
brj-22592	144	32	follows	follow	VERB
brj-22592	144	33	:	:	PUNCT
brj-22592	144	34			X
brj-22592	144	35			X
brj-22592	145	1	++=	++=	PUNCT
brj-22592	145	2	)	)	PUNCT
brj-22592	145	3	y	y	PROPN
brj-22592	145	4	,	,	PUNCT
brj-22592	145	5	x	x	X
brj-22592	145	6	(	(	PUNCT
brj-22592	145	7	)	)	PUNCT
brj-22592	145	8	λ(gp)γ(gp)y),w	λ(gp)γ(gp)y),w	NOUN
brj-22592	145	9	,	,	PUNCT
brj-22592	145	10	x(f(lloss	x(f(lloss	PROPN
brj-22592	145	11	(	(	PUNCT
brj-22592	145	12	6	6	NUM
brj-22592	145	13	)	)	PUNCT
brj-22592	145	14	the	the	DET
brj-22592	145	15	output	output	NOUN
brj-22592	145	16	feature	feature	NOUN
brj-22592	145	17	map	map	NOUN
brj-22592	145	18	of	of	ADP
brj-22592	145	19	the	the	DET
brj-22592	145	20	convolution	convolution	NOUN
brj-22592	145	21	layer	layer	NOUN
brj-22592	145	22	usually	usually	ADV
brj-22592	145	23	includes	include	VERB
brj-22592	145	24	considerable	considerable	ADJ
brj-22592	145	25	redundancy	redundancy	NOUN
brj-22592	145	26	.	.	PUNCT
brj-22592	146	1	a	a	DET
brj-22592	146	2	large	large	ADJ
brj-22592	146	3	number	number	NOUN
brj-22592	146	4	of	of	ADP
brj-22592	146	5	experiments	experiment	NOUN
brj-22592	146	6	demonstrated	demonstrate	VERB
brj-22592	146	7	that	that	SCONJ
brj-22592	146	8	the	the	DET
brj-22592	146	9	generation	generation	NOUN
brj-22592	146	10	of	of	ADP
brj-22592	146	11	these	these	DET
brj-22592	146	12	redundant	redundant	ADJ
brj-22592	146	13	feature	feature	NOUN
brj-22592	146	14	maps	map	NOUN
brj-22592	146	15	individually	individually	ADV
brj-22592	146	16	with	with	ADP
brj-22592	146	17	a	a	DET
brj-22592	146	18	large	large	ADJ
brj-22592	146	19	amount	amount	NOUN
brj-22592	146	20	of	of	ADP
brj-22592	146	21	traffic	traffic	NOUN
brj-22592	146	22	and	and	CCONJ
brj-22592	146	23	parameters	parameter	NOUN
brj-22592	146	24	is	be	AUX
brj-22592	146	25	a	a	DET
brj-22592	146	26	waste	waste	NOUN
brj-22592	146	27	of	of	ADP
brj-22592	146	28	computing	computing	NOUN
brj-22592	146	29	and	and	CCONJ
brj-22592	146	30	hardware	hardware	NOUN
brj-22592	146	31	storage	storage	NOUN
brj-22592	146	32	resources	resource	NOUN
brj-22592	146	33	(	(	PUNCT
brj-22592	146	34	han	han	PROPN
brj-22592	146	35	et	et	PROPN
brj-22592	146	36	al	al	PROPN
brj-22592	146	37	.	.	PROPN
brj-22592	146	38	2020	2020	NUM
brj-22592	146	39	)	)	PUNCT
brj-22592	146	40	.	.	PUNCT
brj-22592	147	1	to	to	PART
brj-22592	147	2	solve	solve	VERB
brj-22592	147	3	the	the	DET
brj-22592	147	4	problem	problem	NOUN
brj-22592	147	5	of	of	ADP
brj-22592	147	6	redundancy	redundancy	NOUN
brj-22592	147	7	in	in	ADP
brj-22592	147	8	the	the	DET
brj-22592	147	9	output	output	NOUN
brj-22592	147	10	characteristic	characteristic	ADJ
brj-22592	147	11	diagram	diagram	NOUN
brj-22592	147	12	of	of	ADP
brj-22592	147	13	the	the	DET
brj-22592	147	14	convolution	convolution	NOUN
brj-22592	147	15	layer	layer	NOUN
brj-22592	147	16	,	,	PUNCT
brj-22592	147	17	in	in	ADP
brj-22592	147	18	this	this	DET
brj-22592	147	19	study	study	NOUN
brj-22592	147	20	,	,	PUNCT
brj-22592	147	21	the	the	DET
brj-22592	147	22	ghostmodule	ghostmodule	NOUN
brj-22592	147	23	was	be	AUX
brj-22592	147	24	introduced	introduce	VERB
brj-22592	147	25	into	into	ADP
brj-22592	147	26	the	the	DET
brj-22592	147	27	body	body	NOUN
brj-22592	147	28	part	part	NOUN
brj-22592	147	29	of	of	ADP
brj-22592	147	30	the	the	DET
brj-22592	147	31	regnet	regnet	NOUN
brj-22592	147	32	network	network	NOUN
brj-22592	147	33	.	.	PUNCT
brj-22592	148	1	the	the	DET
brj-22592	148	2	ghostmodule	ghostmodule	NOUN
brj-22592	148	3	is	be	AUX
brj-22592	148	4	a	a	DET
brj-22592	148	5	method	method	NOUN
brj-22592	148	6	of	of	ADP
brj-22592	148	7	model	model	NOUN
brj-22592	148	8	compression	compression	NOUN
brj-22592	148	9	,	,	PUNCT
brj-22592	148	10	i.e.	i.e.	X
brj-22592	148	11	,	,	PUNCT
brj-22592	148	12	while	while	SCONJ
brj-22592	148	13	ensuring	ensure	VERB
brj-22592	148	14	network	network	NOUN
brj-22592	148	15	accuracy	accuracy	NOUN
brj-22592	148	16	,	,	PUNCT
brj-22592	148	17	it	it	PRON
brj-22592	148	18	can	can	AUX
brj-22592	148	19	effectively	effectively	ADV
brj-22592	148	20	reduce	reduce	VERB
brj-22592	148	21	the	the	DET
brj-22592	148	22	network	network	NOUN
brj-22592	148	23	parameters	parameter	NOUN
brj-22592	148	24	and	and	CCONJ
brj-22592	148	25	amount	amount	NOUN
brj-22592	148	26	of	of	ADP
brj-22592	148	27	calculation	calculation	NOUN
brj-22592	148	28	for	for	ADP
brj-22592	148	29	improving	improve	VERB
brj-22592	148	30	the	the	DET
brj-22592	148	31	calculation	calculation	NOUN
brj-22592	148	32	speed	speed	NOUN
brj-22592	148	33	and	and	CCONJ
brj-22592	148	34	reducing	reduce	VERB
brj-22592	148	35	the	the	DET
brj-22592	148	36	delay	delay	NOUN
brj-22592	148	37	effect	effect	NOUN
brj-22592	148	38	.	.	PUNCT
brj-22592	149	1	in	in	ADP
brj-22592	149	2	the	the	DET
brj-22592	149	3	improved	improved	ADJ
brj-22592	149	4	model	model	NOUN
brj-22592	149	5	,	,	PUNCT
brj-22592	149	6	the	the	DET
brj-22592	149	7	ghostmodule	ghostmodule	NOUN
brj-22592	149	8	was	be	AUX
brj-22592	149	9	embedded	embed	VERB
brj-22592	149	10	in	in	ADP
brj-22592	149	11	the	the	DET
brj-22592	149	12	bottleneck	bottleneck	NOUN
brj-22592	149	13	part	part	NOUN
brj-22592	149	14	of	of	ADP
brj-22592	149	15	the	the	DET
brj-22592	149	16	regnet	regnet	NOUN
brj-22592	149	17	network	network	NOUN
brj-22592	149	18	to	to	PART
brj-22592	149	19	replace	replace	VERB
brj-22592	149	20	some	some	DET
brj-22592	149	21	group	group	NOUN
brj-22592	149	22	convolution	convolution	NOUN
brj-22592	149	23	structures	structure	NOUN
brj-22592	149	24	in	in	ADP
brj-22592	149	25	the	the	DET
brj-22592	149	26	network	network	NOUN
brj-22592	149	27	.	.	PUNCT
brj-22592	150	1	the	the	DET
brj-22592	150	2	operating	operate	VERB
brj-22592	150	3	principle	principle	NOUN
brj-22592	150	4	of	of	ADP
brj-22592	150	5	the	the	DET
brj-22592	150	6	ghostmodule	ghostmodule	NOUN
brj-22592	150	7	convolution	convolution	NOUN
brj-22592	150	8	is	be	AUX
brj-22592	150	9	shown	show	VERB
brj-22592	150	10	in	in	ADP
brj-22592	150	11	fig	fig	NOUN
brj-22592	150	12	.	.	PUNCT
brj-22592	151	1	6	6	NUM
brj-22592	151	2	.	.	PUNCT
brj-22592	151	3	the	the	DET
brj-22592	151	4	calculation	calculation	NOUN
brj-22592	151	5	of	of	ADP
brj-22592	151	6	the	the	DET
brj-22592	151	7	module	module	NOUN
brj-22592	151	8	includes	include	VERB
brj-22592	151	9	two	two	NUM
brj-22592	151	10	steps	step	NOUN
brj-22592	151	11	.	.	PUNCT
brj-22592	152	1	the	the	DET
brj-22592	152	2	first	first	ADJ
brj-22592	152	3	part	part	NOUN
brj-22592	152	4	is	be	AUX
brj-22592	152	5	the	the	DET
brj-22592	152	6	ordinary	ordinary	ADJ
brj-22592	152	7	convolution	convolution	NOUN
brj-22592	152	8	of	of	ADP
brj-22592	152	9	the	the	DET
brj-22592	152	10	input	input	NOUN
brj-22592	152	11	features	feature	NOUN
brj-22592	152	12	and	and	CCONJ
brj-22592	152	13	the	the	DET
brj-22592	152	14	second	second	ADJ
brj-22592	152	15	part	part	NOUN
brj-22592	152	16	is	be	AUX
brj-22592	152	17	the	the	DET
brj-22592	152	18	separation	separation	NOUN
brj-22592	152	19	convolution	convolution	NOUN
brj-22592	152	20	of	of	ADP
brj-22592	152	21	the	the	DET
brj-22592	152	22	results	result	NOUN
brj-22592	152	23	of	of	ADP
brj-22592	152	24	the	the	DET
brj-22592	152	25	first	first	ADJ
brj-22592	152	26	part	part	NOUN
brj-22592	152	27	.	.	PUNCT
brj-22592	153	1	based	base	VERB
brj-22592	153	2	on	on	ADP
brj-22592	153	3	the	the	DET
brj-22592	153	4	calculation	calculation	NOUN
brj-22592	153	5	results	result	NOUN
brj-22592	153	6	of	of	ADP
brj-22592	153	7	the	the	DET
brj-22592	153	8	two	two	NUM
brj-22592	153	9	parts	part	NOUN
brj-22592	153	10	,	,	PUNCT
brj-22592	153	11	the	the	DET
brj-22592	153	12	feature	feature	NOUN
brj-22592	153	13	fusion	fusion	NOUN
brj-22592	153	14	between	between	ADP
brj-22592	153	15	channels	channel	NOUN
brj-22592	153	16	is	be	AUX
brj-22592	153	17	conducted	conduct	VERB
brj-22592	153	18	via	via	ADP
brj-22592	153	19	concat	concat	NOUN
brj-22592	153	20	.	.	PUNCT
brj-22592	154	1	the	the	DET
brj-22592	154	2	block	block	NOUN
brj-22592	154	3	module	module	NOUN
brj-22592	154	4	structure	structure	NOUN
brj-22592	154	5	in	in	ADP
brj-22592	154	6	the	the	DET
brj-22592	154	7	improved	improved	ADJ
brj-22592	154	8	regnet	regnet	NOUN
brj-22592	154	9	model	model	NOUN
brj-22592	154	10	is	be	AUX
brj-22592	154	11	shown	show	VERB
brj-22592	154	12	in	in	ADP
brj-22592	154	13	fig	fig	NOUN
brj-22592	154	14	.	.	PUNCT
brj-22592	155	1	7	7	X
brj-22592	155	2	.	.	X
brj-22592	155	3	peer	peer	NOUN
brj-22592	155	4	-	-	PUNCT
brj-22592	155	5	reviewed	review	VERB
brj-22592	155	6	article	article	NOUN
brj-22592	155	7	bioresources.com	bioresources.com	X
brj-22592	155	8	xie	xie	PROPN
brj-22592	155	9	&	&	CCONJ
brj-22592	155	10	ling	ling	PROPN
brj-22592	155	11	(	(	PUNCT
brj-22592	155	12	2023	2023	NUM
brj-22592	155	13	)	)	PUNCT
brj-22592	155	14	.	.	PUNCT
brj-22592	156	1	“	"	PUNCT
brj-22592	156	2	wood	wood	NOUN
brj-22592	156	3	defect	defect	NOUN
brj-22592	156	4	classification	classification	NOUN
brj-22592	156	5	,	,	PUNCT
brj-22592	156	6	”	"	PUNCT
brj-22592	156	7	bioresources	bioresource	NOUN
brj-22592	156	8	18(4	18(4	NUM
brj-22592	156	9	)	)	PUNCT
brj-22592	156	10	,	,	PUNCT
brj-22592	156	11	7663	7663	NUM
brj-22592	156	12	-	-	SYM
brj-22592	156	13	7680	7680	NUM
brj-22592	156	14	.	.	PUNCT
brj-22592	157	1	7671	7671	NUM
brj-22592	157	2	fig	fig	NOUN
brj-22592	157	3	.	.	PUNCT
brj-22592	158	1	6	6	NUM
brj-22592	158	2	.	.	NOUN
brj-22592	158	3	ghost	ghost	NOUN
brj-22592	158	4	module	module	NOUN
brj-22592	158	5	principle	principle	ADJ
brj-22592	158	6	fig	fig	NOUN
brj-22592	158	7	.	.	PUNCT
brj-22592	159	1	7	7	X
brj-22592	159	2	.	.	NOUN
brj-22592	159	3	block	block	NOUN
brj-22592	159	4	module	module	NOUN
brj-22592	159	5	structure	structure	NOUN
brj-22592	159	6	in	in	ADP
brj-22592	159	7	the	the	DET
brj-22592	159	8	improved	improved	ADJ
brj-22592	159	9	regnet	regnet	NOUN
brj-22592	159	10	model	model	NOUN
brj-22592	159	11	experimental	experimental	ADJ
brj-22592	159	12	process	process	NOUN
brj-22592	159	13	the	the	DET
brj-22592	159	14	hardware	hardware	NOUN
brj-22592	159	15	and	and	CCONJ
brj-22592	159	16	software	software	NOUN
brj-22592	159	17	environment	environment	NOUN
brj-22592	159	18	of	of	ADP
brj-22592	159	19	the	the	DET
brj-22592	159	20	experiment	experiment	NOUN
brj-22592	159	21	is	be	AUX
brj-22592	159	22	shown	show	VERB
brj-22592	159	23	in	in	ADP
brj-22592	159	24	table	table	NOUN
brj-22592	159	25	1	1	NUM
brj-22592	159	26	.	.	PUNCT
brj-22592	160	1	the	the	DET
brj-22592	160	2	training	training	NOUN
brj-22592	160	3	set	set	VERB
brj-22592	160	4	samples	sample	NOUN
brj-22592	160	5	in	in	ADP
brj-22592	160	6	the	the	DET
brj-22592	160	7	wood	wood	NOUN
brj-22592	160	8	defect	defect	NOUN
brj-22592	160	9	sample	sample	NOUN
brj-22592	160	10	dataset	dataset	NOUN
brj-22592	160	11	were	be	AUX
brj-22592	160	12	input	input	VERB
brj-22592	160	13	into	into	ADP
brj-22592	160	14	the	the	DET
brj-22592	160	15	above	above	ADJ
brj-22592	160	16	test	test	NOUN
brj-22592	160	17	network	network	NOUN
brj-22592	160	18	to	to	PART
brj-22592	160	19	obtain	obtain	VERB
brj-22592	160	20	the	the	DET
brj-22592	160	21	training	training	NOUN
brj-22592	160	22	weight	weight	NOUN
brj-22592	160	23	.	.	PUNCT
brj-22592	161	1	the	the	DET
brj-22592	161	2	validation	validation	NOUN
brj-22592	161	3	set	set	VERB
brj-22592	161	4	samples	sample	NOUN
brj-22592	161	5	were	be	AUX
brj-22592	161	6	placed	place	VERB
brj-22592	161	7	in	in	ADP
brj-22592	161	8	the	the	DET
brj-22592	161	9	wood	wood	NOUN
brj-22592	161	10	defect	defect	NOUN
brj-22592	161	11	samples	sample	NOUN
brj-22592	161	12	in	in	ADP
brj-22592	161	13	each	each	DET
brj-22592	161	14	network	network	NOUN
brj-22592	161	15	to	to	PART
brj-22592	161	16	automatically	automatically	ADV
brj-22592	161	17	adjust	adjust	VERB
brj-22592	161	18	the	the	DET
brj-22592	161	19	appropriate	appropriate	ADJ
brj-22592	161	20	super	super	ADJ
brj-22592	161	21	parameters	parameter	NOUN
brj-22592	161	22	.	.	PUNCT
brj-22592	162	1	(	(	PUNCT
brj-22592	162	2	a	a	X
brj-22592	162	3	)	)	PUNCT
brj-22592	162	4	x	x	SYM
brj-22592	162	5	block	block	NOUN
brj-22592	162	6	,	,	PUNCT
brj-22592	162	7	s=1	s=1	X
brj-22592	162	8	(	(	PUNCT
brj-22592	162	9	b	b	NOUN
brj-22592	162	10	)	)	PUNCT
brj-22592	162	11	x	x	SYM
brj-22592	162	12	block	block	NOUN
brj-22592	162	13	,	,	PUNCT
brj-22592	162	14	s=2	s=2	PRON
brj-22592	162	15	peer	peer	NOUN
brj-22592	162	16	-	-	PUNCT
brj-22592	162	17	reviewed	review	VERB
brj-22592	162	18	article	article	NOUN
brj-22592	162	19	bioresources.com	bioresources.com	X
brj-22592	162	20	xie	xie	PROPN
brj-22592	162	21	&	&	CCONJ
brj-22592	162	22	ling	ling	PROPN
brj-22592	162	23	(	(	PUNCT
brj-22592	162	24	2023	2023	NUM
brj-22592	162	25	)	)	PUNCT
brj-22592	162	26	.	.	PUNCT
brj-22592	163	1	“	"	PUNCT
brj-22592	163	2	wood	wood	NOUN
brj-22592	163	3	defect	defect	NOUN
brj-22592	163	4	classification	classification	NOUN
brj-22592	163	5	,	,	PUNCT
brj-22592	163	6	”	"	PUNCT
brj-22592	163	7	bioresources	bioresource	NOUN
brj-22592	163	8	18(4	18(4	NUM
brj-22592	163	9	)	)	PUNCT
brj-22592	163	10	,	,	PUNCT
brj-22592	163	11	7663	7663	NUM
brj-22592	163	12	-	-	SYM
brj-22592	163	13	7680	7680	NUM
brj-22592	163	14	.	.	PUNCT
brj-22592	164	1	7672	7672	NUM
brj-22592	164	2	finally	finally	ADV
brj-22592	164	3	,	,	PUNCT
brj-22592	164	4	the	the	DET
brj-22592	164	5	test	test	NOUN
brj-22592	164	6	set	set	VERB
brj-22592	164	7	samples	sample	NOUN
brj-22592	164	8	were	be	AUX
brj-22592	164	9	placed	place	VERB
brj-22592	164	10	into	into	ADP
brj-22592	164	11	each	each	DET
brj-22592	164	12	network	network	NOUN
brj-22592	164	13	model	model	NOUN
brj-22592	164	14	with	with	ADP
brj-22592	164	15	the	the	DET
brj-22592	164	16	best	good	ADJ
brj-22592	164	17	training	training	NOUN
brj-22592	164	18	,	,	PUNCT
brj-22592	164	19	and	and	CCONJ
brj-22592	164	20	the	the	DET
brj-22592	164	21	most	most	ADV
brj-22592	164	22	suitable	suitable	ADJ
brj-22592	164	23	network	network	NOUN
brj-22592	164	24	model	model	NOUN
brj-22592	164	25	was	be	AUX
brj-22592	164	26	obtained	obtain	VERB
brj-22592	164	27	for	for	ADP
brj-22592	164	28	this	this	DET
brj-22592	164	29	dataset	dataset	NOUN
brj-22592	164	30	.	.	PUNCT
brj-22592	165	1	during	during	ADP
brj-22592	165	2	the	the	DET
brj-22592	165	3	experiment	experiment	NOUN
brj-22592	165	4	,	,	PUNCT
brj-22592	165	5	it	it	PRON
brj-22592	165	6	was	be	AUX
brj-22592	165	7	necessary	necessary	ADJ
brj-22592	165	8	to	to	PART
brj-22592	165	9	control	control	VERB
brj-22592	165	10	the	the	DET
brj-22592	165	11	consistency	consistency	NOUN
brj-22592	165	12	of	of	ADP
brj-22592	165	13	other	other	ADJ
brj-22592	165	14	super	super	ADJ
brj-22592	165	15	parameters	parameter	NOUN
brj-22592	165	16	,	,	PUNCT
brj-22592	165	17	and	and	CCONJ
brj-22592	165	18	finally	finally	ADV
brj-22592	165	19	,	,	PUNCT
brj-22592	165	20	to	to	PART
brj-22592	165	21	compare	compare	VERB
brj-22592	165	22	the	the	DET
brj-22592	165	23	test	test	NOUN
brj-22592	165	24	set	set	VERB
brj-22592	165	25	accuracy	accuracy	NOUN
brj-22592	165	26	of	of	ADP
brj-22592	165	27	each	each	DET
brj-22592	165	28	model	model	NOUN
brj-22592	165	29	,	,	PUNCT
brj-22592	165	30	number	number	NOUN
brj-22592	165	31	of	of	ADP
brj-22592	165	32	network	network	NOUN
brj-22592	165	33	parameters	parameter	NOUN
brj-22592	165	34	,	,	PUNCT
brj-22592	165	35	and	and	CCONJ
brj-22592	165	36	various	various	ADJ
brj-22592	165	37	evaluation	evaluation	NOUN
brj-22592	165	38	performance	performance	NOUN
brj-22592	165	39	indexes	index	NOUN
brj-22592	165	40	of	of	ADP
brj-22592	165	41	the	the	DET
brj-22592	165	42	model	model	NOUN
brj-22592	165	43	.	.	PUNCT
brj-22592	166	1	table	table	NOUN
brj-22592	166	2	1	1	NUM
brj-22592	166	3	.	.	PUNCT
brj-22592	167	1	the	the	DET
brj-22592	167	2	hardware	hardware	NOUN
brj-22592	167	3	and	and	CCONJ
brj-22592	167	4	software	software	NOUN
brj-22592	167	5	environment	environment	NOUN
brj-22592	167	6	of	of	ADP
brj-22592	167	7	the	the	DET
brj-22592	167	8	experiment	experiment	NOUN
brj-22592	167	9	experimental	experimental	ADJ
brj-22592	167	10	environment	environment	NOUN
brj-22592	167	11	version	version	NOUN
brj-22592	167	12	details	detail	NOUN
brj-22592	167	13	gpu	gpu	PROPN
brj-22592	167	14	model	model	NOUN
brj-22592	167	15	nvidia	nvidia	PROPN
brj-22592	167	16	gtx980	gtx980	PROPN
brj-22592	167	17	cpu	cpu	PROPN
brj-22592	167	18	model	model	NOUN
brj-22592	167	19	intel(r)core	intel(r)core	X
brj-22592	167	20	(	(	PUNCT
brj-22592	167	21	tm	tm	NOUN
brj-22592	167	22	)	)	PUNCT
brj-22592	167	23	i5	i5	ADJ
brj-22592	167	24	-	-	PUNCT
brj-22592	167	25	6300hq	6300hq	ADJ
brj-22592	167	26	cpu@2.30ghz	cpu@2.30ghz	NOUN
brj-22592	167	27	memory	memory	NOUN
brj-22592	167	28	size	size	NOUN
brj-22592	167	29	4	4	NUM
brj-22592	167	30	g	g	NOUN
brj-22592	167	31	system	system	NOUN
brj-22592	167	32	version	version	NOUN
brj-22592	167	33	windows	window	NOUN
brj-22592	167	34	10	10	NUM
brj-22592	167	35	programming	programming	NOUN
brj-22592	167	36	language	language	NOUN
brj-22592	167	37	python	python	NOUN
brj-22592	167	38	3.8.0	3.8.0	NUM
brj-22592	167	39	python	python	NOUN
brj-22592	167	40	version	version	NOUN
brj-22592	167	41	pytorch	pytorch	NOUN
brj-22592	167	42	1.7.1	1.7.1	NUM
brj-22592	167	43	the	the	DET
brj-22592	167	44	experiment	experiment	NOUN
brj-22592	167	45	was	be	AUX
brj-22592	167	46	completed	complete	VERB
brj-22592	167	47	in	in	ADP
brj-22592	167	48	two	two	NUM
brj-22592	167	49	steps	step	NOUN
brj-22592	167	50	.	.	PUNCT
brj-22592	168	1	first	first	ADV
brj-22592	168	2	,	,	PUNCT
brj-22592	168	3	the	the	DET
brj-22592	168	4	classification	classification	NOUN
brj-22592	168	5	performance	performance	NOUN
brj-22592	168	6	of	of	ADP
brj-22592	168	7	regnet	regnet	NOUN
brj-22592	168	8	,	,	PUNCT
brj-22592	168	9	regnet+se	regnet+se	ADJ
brj-22592	168	10	,	,	PUNCT
brj-22592	168	11	regnet+eca	regnet+eca	PROPN
brj-22592	168	12	,	,	PUNCT
brj-22592	168	13	regnet+nam	regnet+nam	PROPN
brj-22592	168	14	,	,	PUNCT
brj-22592	168	15	and	and	CCONJ
brj-22592	168	16	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	168	17	networks	network	NOUN
brj-22592	168	18	were	be	AUX
brj-22592	168	19	compared	compare	VERB
brj-22592	168	20	to	to	PART
brj-22592	168	21	select	select	VERB
brj-22592	168	22	high	high	ADJ
brj-22592	168	23	-	-	PUNCT
brj-22592	168	24	performance	performance	NOUN
brj-22592	168	25	classification	classification	NOUN
brj-22592	168	26	networks	network	NOUN
brj-22592	168	27	.	.	PUNCT
brj-22592	169	1	then	then	ADV
brj-22592	169	2	,	,	PUNCT
brj-22592	169	3	the	the	DET
brj-22592	169	4	selected	select	VERB
brj-22592	169	5	network	network	NOUN
brj-22592	169	6	,	,	PUNCT
brj-22592	169	7	mobilenet	mobilenet	NOUN
brj-22592	169	8	-	-	PUNCT
brj-22592	169	9	v2	v2	NOUN
brj-22592	169	10	,	,	PUNCT
brj-22592	169	11	efficientnet	efficientnet	NOUN
brj-22592	169	12	,	,	PUNCT
brj-22592	169	13	and	and	CCONJ
brj-22592	169	14	vision	vision	NOUN
brj-22592	169	15	-	-	PUNCT
brj-22592	169	16	transformer	transformer	NOUN
brj-22592	169	17	networks	network	NOUN
brj-22592	169	18	were	be	AUX
brj-22592	169	19	selected	select	VERB
brj-22592	169	20	twice	twice	ADV
brj-22592	169	21	to	to	PART
brj-22592	169	22	find	find	VERB
brj-22592	169	23	the	the	DET
brj-22592	169	24	best	good	ADJ
brj-22592	169	25	network	network	NOUN
brj-22592	169	26	model	model	NOUN
brj-22592	169	27	suitable	suitable	ADJ
brj-22592	169	28	for	for	ADP
brj-22592	169	29	the	the	DET
brj-22592	169	30	wood	wood	NOUN
brj-22592	169	31	defect	defect	NOUN
brj-22592	169	32	classification	classification	NOUN
brj-22592	169	33	.	.	PUNCT
brj-22592	170	1	results	result	NOUN
brj-22592	170	2	and	and	CCONJ
brj-22592	170	3	discussion	discussion	NOUN
brj-22592	170	4	to	to	PART
brj-22592	170	5	demonstrate	demonstrate	VERB
brj-22592	170	6	the	the	DET
brj-22592	170	7	role	role	NOUN
brj-22592	170	8	of	of	ADP
brj-22592	170	9	the	the	DET
brj-22592	170	10	attention	attention	NOUN
brj-22592	170	11	mechanism	mechanism	NOUN
brj-22592	170	12	in	in	ADP
brj-22592	170	13	network	network	NOUN
brj-22592	170	14	applications	application	NOUN
brj-22592	170	15	,	,	PUNCT
brj-22592	170	16	the	the	DET
brj-22592	170	17	test	test	NOUN
brj-22592	170	18	set	set	VERB
brj-22592	170	19	samples	sample	NOUN
brj-22592	170	20	of	of	ADP
brj-22592	170	21	various	various	ADJ
brj-22592	170	22	wood	wood	NOUN
brj-22592	170	23	defects	defect	NOUN
brj-22592	170	24	were	be	AUX
brj-22592	170	25	input	input	VERB
brj-22592	170	26	into	into	ADP
brj-22592	170	27	the	the	DET
brj-22592	170	28	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	170	29	network	network	NOUN
brj-22592	170	30	model	model	NOUN
brj-22592	170	31	,	,	PUNCT
brj-22592	170	32	and	and	CCONJ
brj-22592	170	33	the	the	DET
brj-22592	170	34	visual	visual	ADJ
brj-22592	170	35	thermal	thermal	ADJ
brj-22592	170	36	diagram	diagram	NOUN
brj-22592	170	37	of	of	ADP
brj-22592	170	38	the	the	DET
brj-22592	170	39	nam	nam	NOUN
brj-22592	170	40	attention	attention	NOUN
brj-22592	170	41	mechanism	mechanism	NOUN
brj-22592	170	42	was	be	AUX
brj-22592	170	43	drawn	draw	VERB
brj-22592	170	44	by	by	ADP
brj-22592	170	45	the	the	DET
brj-22592	170	46	grad	grad	NOUN
brj-22592	170	47	-	-	PUNCT
brj-22592	170	48	cam	cam	NOUN
brj-22592	170	49	network	network	NOUN
brj-22592	170	50	,	,	PUNCT
brj-22592	170	51	as	as	SCONJ
brj-22592	170	52	shown	show	VERB
brj-22592	170	53	in	in	ADP
brj-22592	170	54	fig	fig	NOUN
brj-22592	170	55	.	.	PUNCT
brj-22592	171	1	8	8	NUM
brj-22592	171	2	.	.	X
brj-22592	171	3	fig	fig	NOUN
brj-22592	171	4	.	.	PUNCT
brj-22592	172	1	8	8	NUM
brj-22592	172	2	.	.	X
brj-22592	172	3	thermodynamic	thermodynamic	ADJ
brj-22592	172	4	diagram	diagram	NOUN
brj-22592	172	5	of	of	ADP
brj-22592	172	6	attention	attention	NOUN
brj-22592	172	7	mechanism	mechanism	NOUN
brj-22592	172	8	of	of	ADP
brj-22592	172	9	various	various	ADJ
brj-22592	172	10	samples	sample	NOUN
brj-22592	172	11	in	in	ADP
brj-22592	172	12	the	the	DET
brj-22592	172	13	test	test	NOUN
brj-22592	172	14	set	set	VERB
brj-22592	172	15	experimental	experimental	ADJ
brj-22592	172	16	data	datum	NOUN
brj-22592	172	17	and	and	CCONJ
brj-22592	172	18	results	result	VERB
brj-22592	172	19	the	the	DET
brj-22592	172	20	training	training	NOUN
brj-22592	172	21	set	set	NOUN
brj-22592	172	22	samples	sample	NOUN
brj-22592	172	23	were	be	AUX
brj-22592	172	24	placed	place	VERB
brj-22592	172	25	into	into	ADP
brj-22592	172	26	the	the	DET
brj-22592	172	27	test	test	NOUN
brj-22592	172	28	network	network	NOUN
brj-22592	172	29	and	and	CCONJ
brj-22592	172	30	the	the	DET
brj-22592	172	31	number	number	NOUN
brj-22592	172	32	of	of	ADP
brj-22592	172	33	experimental	experimental	ADJ
brj-22592	172	34	epochs	epoch	NOUN
brj-22592	172	35	was	be	AUX
brj-22592	172	36	set	set	VERB
brj-22592	172	37	to	to	ADP
brj-22592	172	38	120	120	NUM
brj-22592	172	39	,	,	PUNCT
brj-22592	172	40	the	the	DET
brj-22592	172	41	number	number	NOUN
brj-22592	172	42	of	of	ADP
brj-22592	172	43	learning	learn	VERB
brj-22592	172	44	rate	rate	NOUN
brj-22592	172	45	was	be	AUX
brj-22592	172	46	set	set	VERB
brj-22592	172	47	to	to	ADP
brj-22592	172	48	0.001	0.001	NUM
brj-22592	172	49	,	,	PUNCT
brj-22592	172	50	and	and	CCONJ
brj-22592	172	51	the	the	DET
brj-22592	172	52	number	number	NOUN
brj-22592	172	53	of	of	ADP
brj-22592	172	54	batch	batch	NOUN
brj-22592	172	55	size	size	NOUN
brj-22592	172	56	was	be	AUX
brj-22592	172	57	set	set	VERB
brj-22592	172	58	to	to	ADP
brj-22592	172	59	32	32	NUM
brj-22592	172	60	for	for	ADP
brj-22592	172	61	comparative	comparative	ADJ
brj-22592	172	62	experiments	experiment	NOUN
brj-22592	172	63	.	.	PUNCT
brj-22592	173	1	(	(	PUNCT
brj-22592	173	2	a	a	X
brj-22592	173	3	)	)	PUNCT
brj-22592	173	4	dead	dead	ADJ
brj-22592	173	5	knot	knot	NOUN
brj-22592	173	6	(	(	PUNCT
brj-22592	173	7	b	b	NOUN
brj-22592	173	8	)	)	PUNCT
brj-22592	173	9	wormhole	wormhole	NOUN
brj-22592	173	10	(	(	PUNCT
brj-22592	173	11	c	c	NOUN
brj-22592	173	12	)	)	PUNCT
brj-22592	173	13	slip	slip	NOUN
brj-22592	173	14	knot	knot	VERB
brj-22592	173	15	peer	peer	NOUN
brj-22592	173	16	-	-	PUNCT
brj-22592	173	17	reviewed	review	VERB
brj-22592	173	18	article	article	NOUN
brj-22592	173	19	bioresources.com	bioresources.com	X
brj-22592	173	20	xie	xie	PROPN
brj-22592	173	21	&	&	CCONJ
brj-22592	173	22	ling	ling	PROPN
brj-22592	173	23	(	(	PUNCT
brj-22592	173	24	2023	2023	NUM
brj-22592	173	25	)	)	PUNCT
brj-22592	173	26	.	.	PUNCT
brj-22592	174	1	“	"	PUNCT
brj-22592	174	2	wood	wood	NOUN
brj-22592	174	3	defect	defect	NOUN
brj-22592	174	4	classification	classification	NOUN
brj-22592	174	5	,	,	PUNCT
brj-22592	174	6	”	"	PUNCT
brj-22592	174	7	bioresources	bioresource	NOUN
brj-22592	174	8	18(4	18(4	NUM
brj-22592	174	9	)	)	PUNCT
brj-22592	174	10	,	,	PUNCT
brj-22592	174	11	7663	7663	NUM
brj-22592	174	12	-	-	SYM
brj-22592	174	13	7680	7680	NUM
brj-22592	174	14	.	.	PUNCT
brj-22592	175	1	7673	7673	NUM
brj-22592	175	2	(	(	PUNCT
brj-22592	175	3	a	a	NOUN
brj-22592	175	4	)	)	PUNCT
brj-22592	175	5	(	(	PUNCT
brj-22592	175	6	b	b	X
brj-22592	175	7	)	)	PUNCT
brj-22592	175	8	(	(	PUNCT
brj-22592	175	9	c	c	X
brj-22592	175	10	)	)	PUNCT
brj-22592	175	11	(	(	PUNCT
brj-22592	175	12	d	d	X
brj-22592	175	13	)	)	PUNCT
brj-22592	175	14	fig	fig	NOUN
brj-22592	175	15	.	.	PUNCT
brj-22592	176	1	9	9	X
brj-22592	176	2	.	.	X
brj-22592	176	3	loss	loss	NOUN
brj-22592	176	4	function	function	NOUN
brj-22592	176	5	and	and	CCONJ
brj-22592	176	6	accuracy	accuracy	NOUN
brj-22592	176	7	of	of	ADP
brj-22592	176	8	networks	network	NOUN
brj-22592	176	9	(	(	PUNCT
brj-22592	176	10	a	a	X
brj-22592	176	11	)	)	PUNCT
brj-22592	176	12	loss	loss	NOUN
brj-22592	176	13	function	function	NOUN
brj-22592	176	14	and	and	CCONJ
brj-22592	176	15	accuracy	accuracy	NOUN
brj-22592	176	16	of	of	ADP
brj-22592	176	17	ghostregnet	ghostregnet	NOUN
brj-22592	176	18	+	+	PROPN
brj-22592	176	19	nam	nam	NOUN
brj-22592	176	20	network	network	NOUN
brj-22592	176	21	(	(	PUNCT
brj-22592	176	22	b	b	NOUN
brj-22592	176	23	)	)	PUNCT
brj-22592	176	24	loss	loss	NOUN
brj-22592	176	25	function	function	NOUN
brj-22592	176	26	and	and	CCONJ
brj-22592	176	27	accuracy	accuracy	NOUN
brj-22592	176	28	of	of	ADP
brj-22592	176	29	regnet	regnet	NOUN
brj-22592	176	30	network	network	NOUN
brj-22592	176	31	(	(	PUNCT
brj-22592	176	32	c	c	NOUN
brj-22592	176	33	)	)	PUNCT
brj-22592	176	34	loss	loss	NOUN
brj-22592	176	35	function	function	NOUN
brj-22592	176	36	and	and	CCONJ
brj-22592	176	37	accuracy	accuracy	NOUN
brj-22592	176	38	of	of	ADP
brj-22592	176	39	regnet+se	regnet+se	ADJ
brj-22592	176	40	network	network	NOUN
brj-22592	176	41	(	(	PUNCT
brj-22592	176	42	d	d	NOUN
brj-22592	176	43	)	)	PUNCT
brj-22592	176	44	loss	loss	NOUN
brj-22592	176	45	function	function	NOUN
brj-22592	176	46	and	and	CCONJ
brj-22592	176	47	accuracy	accuracy	NOUN
brj-22592	176	48	of	of	ADP
brj-22592	176	49	regnet+eca	regnet+eca	NOUN
brj-22592	176	50	network	network	NOUN
brj-22592	176	51	(	(	PUNCT
brj-22592	176	52	1	1	NUM
brj-22592	176	53	)	)	PUNCT
brj-22592	176	54	regnet	regnet	NOUN
brj-22592	176	55	network	network	NOUN
brj-22592	176	56	realized	realize	VERB
brj-22592	176	57	wood	wood	NOUN
brj-22592	176	58	defect	defect	NOUN
brj-22592	176	59	classification	classification	NOUN
brj-22592	176	60	.	.	PUNCT
brj-22592	177	1	in	in	ADP
brj-22592	177	2	the	the	DET
brj-22592	177	3	108th	108th	ADJ
brj-22592	177	4	epoch	epoch	NOUN
brj-22592	177	5	,	,	PUNCT
brj-22592	177	6	the	the	DET
brj-22592	177	7	accuracy	accuracy	NOUN
brj-22592	177	8	of	of	ADP
brj-22592	177	9	the	the	DET
brj-22592	177	10	validation	validation	NOUN
brj-22592	177	11	set	set	NOUN
brj-22592	177	12	was	be	AUX
brj-22592	177	13	highest	high	ADJ
brj-22592	177	14	at	at	ADP
brj-22592	177	15	95.2	95.2	NUM
brj-22592	177	16	%	%	NOUN
brj-22592	177	17	.	.	PUNCT
brj-22592	178	1	the	the	DET
brj-22592	178	2	training	training	NOUN
brj-22592	178	3	set	set	VERB
brj-22592	178	4	loss	loss	NOUN
brj-22592	178	5	function	function	NOUN
brj-22592	178	6	and	and	CCONJ
brj-22592	178	7	validation	validation	NOUN
brj-22592	178	8	set	set	VERB
brj-22592	178	9	accuracy	accuracy	NOUN
brj-22592	178	10	of	of	ADP
brj-22592	178	11	the	the	DET
brj-22592	178	12	regnet	regnet	NOUN
brj-22592	178	13	network	network	NOUN
brj-22592	178	14	are	be	AUX
brj-22592	178	15	shown	show	VERB
brj-22592	178	16	in	in	ADP
brj-22592	178	17	fig	fig	NOUN
brj-22592	178	18	.	.	PUNCT
brj-22592	179	1	9(b	9(b	NUM
brj-22592	179	2	)	)	PUNCT
brj-22592	179	3	.	.	PUNCT
brj-22592	180	1	(	(	PUNCT
brj-22592	180	2	2	2	NUM
brj-22592	180	3	)	)	PUNCT
brj-22592	180	4	regnet+se	regnet+se	NOUN
brj-22592	180	5	network	network	NOUN
brj-22592	180	6	classified	classify	VERB
brj-22592	180	7	the	the	DET
brj-22592	180	8	classification	classification	NOUN
brj-22592	180	9	of	of	ADP
brj-22592	180	10	wood	wood	NOUN
brj-22592	180	11	defects	defect	NOUN
brj-22592	180	12	.	.	PUNCT
brj-22592	181	1	in	in	ADP
brj-22592	181	2	the	the	DET
brj-22592	181	3	105th	105th	ADJ
brj-22592	181	4	epoch	epoch	NOUN
brj-22592	181	5	,	,	PUNCT
brj-22592	181	6	the	the	DET
brj-22592	181	7	accuracy	accuracy	NOUN
brj-22592	181	8	of	of	ADP
brj-22592	181	9	the	the	DET
brj-22592	181	10	validation	validation	NOUN
brj-22592	181	11	set	set	NOUN
brj-22592	181	12	was	be	AUX
brj-22592	181	13	the	the	DET
brj-22592	181	14	highest	high	ADJ
brj-22592	181	15	,	,	PUNCT
brj-22592	181	16	with	with	ADP
brj-22592	181	17	a	a	DET
brj-22592	181	18	maximum	maximum	ADJ
brj-22592	181	19	accuracy	accuracy	NOUN
brj-22592	181	20	of	of	ADP
brj-22592	181	21	95.9	95.9	NUM
brj-22592	181	22	%	%	NOUN
brj-22592	181	23	.	.	PUNCT
brj-22592	182	1	the	the	DET
brj-22592	182	2	training	training	NOUN
brj-22592	182	3	set	set	VERB
brj-22592	182	4	loss	loss	NOUN
brj-22592	182	5	function	function	NOUN
brj-22592	182	6	and	and	CCONJ
brj-22592	182	7	validation	validation	NOUN
brj-22592	182	8	set	set	VERB
brj-22592	182	9	accuracy	accuracy	NOUN
brj-22592	182	10	of	of	ADP
brj-22592	182	11	the	the	DET
brj-22592	182	12	regnet+se	regnet+se	ADJ
brj-22592	182	13	network	network	NOUN
brj-22592	182	14	are	be	AUX
brj-22592	182	15	shown	show	VERB
brj-22592	182	16	in	in	ADP
brj-22592	182	17	fig	fig	NOUN
brj-22592	182	18	.	.	PUNCT
brj-22592	183	1	9(c	9(c	NUM
brj-22592	183	2	)	)	PUNCT
brj-22592	183	3	.	.	PUNCT
brj-22592	184	1	(	(	PUNCT
brj-22592	184	2	3	3	X
brj-22592	184	3	)	)	PUNCT
brj-22592	184	4	regnet+eca	regnet+eca	NOUN
brj-22592	184	5	network	network	NOUN
brj-22592	184	6	realized	realize	VERB
brj-22592	184	7	the	the	DET
brj-22592	184	8	classification	classification	NOUN
brj-22592	184	9	of	of	ADP
brj-22592	184	10	wood	wood	NOUN
brj-22592	184	11	defects	defect	NOUN
brj-22592	184	12	.	.	PUNCT
brj-22592	185	1	in	in	ADP
brj-22592	185	2	the	the	DET
brj-22592	185	3	116th	116th	ADJ
brj-22592	185	4	epoch	epoch	NOUN
brj-22592	185	5	,	,	PUNCT
brj-22592	185	6	the	the	DET
brj-22592	185	7	accuracy	accuracy	NOUN
brj-22592	185	8	of	of	ADP
brj-22592	185	9	the	the	DET
brj-22592	185	10	validation	validation	NOUN
brj-22592	185	11	set	set	NOUN
brj-22592	185	12	was	be	AUX
brj-22592	185	13	the	the	DET
brj-22592	185	14	highest	high	ADJ
brj-22592	185	15	,	,	PUNCT
brj-22592	185	16	with	with	ADP
brj-22592	185	17	a	a	DET
brj-22592	185	18	maximum	maximum	ADJ
brj-22592	185	19	accuracy	accuracy	NOUN
brj-22592	185	20	of	of	ADP
brj-22592	185	21	95.2	95.2	NUM
brj-22592	185	22	%	%	NOUN
brj-22592	185	23	.	.	PUNCT
brj-22592	186	1	the	the	DET
brj-22592	186	2	training	training	NOUN
brj-22592	186	3	set	set	VERB
brj-22592	186	4	loss	loss	NOUN
brj-22592	186	5	function	function	NOUN
brj-22592	186	6	and	and	CCONJ
brj-22592	186	7	validation	validation	NOUN
brj-22592	186	8	set	set	VERB
brj-22592	186	9	accuracy	accuracy	NOUN
brj-22592	186	10	of	of	ADP
brj-22592	186	11	the	the	DET
brj-22592	186	12	regnet+eca	regnet+eca	NOUN
brj-22592	186	13	network	network	NOUN
brj-22592	186	14	are	be	AUX
brj-22592	186	15	shown	show	VERB
brj-22592	186	16	in	in	ADP
brj-22592	186	17	fig	fig	NOUN
brj-22592	186	18	.	.	PUNCT
brj-22592	187	1	9(d	9(d	NUM
brj-22592	187	2	)	)	PUNCT
brj-22592	187	3	.	.	PUNCT
brj-22592	188	1	(	(	PUNCT
brj-22592	188	2	4	4	X
brj-22592	188	3	)	)	PUNCT
brj-22592	188	4	regnet+nam	regnet+nam	PROPN
brj-22592	188	5	network	network	NOUN
brj-22592	188	6	realized	realize	VERB
brj-22592	188	7	wood	wood	NOUN
brj-22592	188	8	defect	defect	NOUN
brj-22592	188	9	classification	classification	NOUN
brj-22592	188	10	.	.	PUNCT
brj-22592	189	1	in	in	ADP
brj-22592	189	2	the	the	DET
brj-22592	189	3	113th	113th	PROPN
brj-22592	189	4	epoch	epoch	NOUN
brj-22592	189	5	,	,	PUNCT
brj-22592	189	6	the	the	DET
brj-22592	189	7	accuracy	accuracy	NOUN
brj-22592	189	8	of	of	ADP
brj-22592	189	9	the	the	DET
brj-22592	189	10	validation	validation	NOUN
brj-22592	189	11	set	set	NOUN
brj-22592	189	12	was	be	AUX
brj-22592	189	13	the	the	DET
brj-22592	189	14	highest	high	ADJ
brj-22592	189	15	,	,	PUNCT
brj-22592	189	16	with	with	ADP
brj-22592	189	17	a	a	DET
brj-22592	189	18	maximum	maximum	ADJ
brj-22592	189	19	accuracy	accuracy	NOUN
brj-22592	189	20	of	of	ADP
brj-22592	189	21	96.2	96.2	NUM
brj-22592	189	22	%	%	NOUN
brj-22592	189	23	.	.	PUNCT
brj-22592	190	1	the	the	DET
brj-22592	190	2	training	training	NOUN
brj-22592	190	3	set	set	VERB
brj-22592	190	4	loss	loss	NOUN
brj-22592	190	5	function	function	NOUN
brj-22592	190	6	and	and	CCONJ
brj-22592	190	7	validation	validation	NOUN
brj-22592	190	8	set	set	VERB
brj-22592	190	9	accuracy	accuracy	NOUN
brj-22592	190	10	of	of	ADP
brj-22592	190	11	the	the	DET
brj-22592	190	12	regnet+nam	regnet+nam	PROPN
brj-22592	190	13	network	network	NOUN
brj-22592	190	14	are	be	AUX
brj-22592	190	15	shown	show	VERB
brj-22592	190	16	in	in	ADP
brj-22592	190	17	figure	figure	NOUN
brj-22592	190	18	10(a	10(a	NUM
brj-22592	190	19	)	)	PUNCT
brj-22592	190	20	.	.	PUNCT
brj-22592	191	1	(	(	PUNCT
brj-22592	191	2	5	5	X
brj-22592	191	3	)	)	PUNCT
brj-22592	191	4	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	191	5	network	network	NOUN
brj-22592	191	6	realized	realize	VERB
brj-22592	191	7	wood	wood	NOUN
brj-22592	191	8	defect	defect	NOUN
brj-22592	191	9	classification	classification	NOUN
brj-22592	191	10	.	.	PUNCT
brj-22592	192	1	in	in	ADP
brj-22592	192	2	the	the	DET
brj-22592	192	3	107th	107th	ADJ
brj-22592	192	4	epoch	epoch	NOUN
brj-22592	192	5	,	,	PUNCT
brj-22592	192	6	the	the	DET
brj-22592	192	7	accuracy	accuracy	NOUN
brj-22592	192	8	of	of	ADP
brj-22592	192	9	the	the	DET
brj-22592	192	10	validation	validation	NOUN
brj-22592	192	11	set	set	NOUN
brj-22592	192	12	was	be	AUX
brj-22592	192	13	the	the	DET
brj-22592	192	14	highest	high	ADJ
brj-22592	192	15	,	,	PUNCT
brj-22592	192	16	with	with	ADP
brj-22592	192	17	a	a	DET
brj-22592	192	18	maximum	maximum	ADJ
brj-22592	192	19	accuracy	accuracy	NOUN
brj-22592	192	20	of	of	ADP
brj-22592	192	21	96.3	96.3	NUM
brj-22592	192	22	%	%	NOUN
brj-22592	192	23	.	.	PUNCT
brj-22592	193	1	the	the	DET
brj-22592	193	2	training	training	NOUN
brj-22592	193	3	set	set	VERB
brj-22592	193	4	loss	loss	NOUN
brj-22592	193	5	function	function	NOUN
brj-22592	193	6	and	and	CCONJ
brj-22592	193	7	validation	validation	NOUN
brj-22592	193	8	set	set	VERB
brj-22592	193	9	accuracy	accuracy	NOUN
brj-22592	193	10	of	of	ADP
brj-22592	193	11	the	the	DET
brj-22592	193	12	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	193	13	network	network	NOUN
brj-22592	193	14	are	be	AUX
brj-22592	193	15	shown	show	VERB
brj-22592	193	16	in	in	ADP
brj-22592	193	17	fig	fig	NOUN
brj-22592	193	18	.	.	PUNCT
brj-22592	194	1	9(a	9(a	NUM
brj-22592	194	2	)	)	PUNCT
brj-22592	194	3	.	.	PUNCT
brj-22592	195	1	peer	peer	NOUN
brj-22592	195	2	-	-	PUNCT
brj-22592	195	3	reviewed	review	VERB
brj-22592	195	4	article	article	NOUN
brj-22592	195	5	bioresources.com	bioresources.com	X
brj-22592	195	6	xie	xie	PROPN
brj-22592	195	7	&	&	CCONJ
brj-22592	195	8	ling	ling	PROPN
brj-22592	195	9	(	(	PUNCT
brj-22592	195	10	2023	2023	NUM
brj-22592	195	11	)	)	PUNCT
brj-22592	195	12	.	.	PUNCT
brj-22592	196	1	“	"	PUNCT
brj-22592	196	2	wood	wood	NOUN
brj-22592	196	3	defect	defect	NOUN
brj-22592	196	4	classification	classification	NOUN
brj-22592	196	5	,	,	PUNCT
brj-22592	196	6	”	"	PUNCT
brj-22592	196	7	bioresources	bioresource	NOUN
brj-22592	196	8	18(4	18(4	NUM
brj-22592	196	9	)	)	PUNCT
brj-22592	196	10	,	,	PUNCT
brj-22592	196	11	7663	7663	NUM
brj-22592	196	12	-	-	SYM
brj-22592	196	13	7680	7680	NUM
brj-22592	196	14	.	.	PUNCT
brj-22592	197	1	7674	7674	NUM
brj-22592	197	2	(	(	PUNCT
brj-22592	197	3	a	a	NOUN
brj-22592	197	4	)	)	PUNCT
brj-22592	197	5	(	(	PUNCT
brj-22592	197	6	b	b	X
brj-22592	197	7	)	)	PUNCT
brj-22592	197	8	(	(	PUNCT
brj-22592	197	9	c	c	X
brj-22592	197	10	)	)	PUNCT
brj-22592	197	11	(	(	PUNCT
brj-22592	197	12	d	d	X
brj-22592	197	13	)	)	PUNCT
brj-22592	197	14	fig	fig	NOUN
brj-22592	197	15	.	.	PUNCT
brj-22592	198	1	10	10	NUM
brj-22592	198	2	.	.	PUNCT
brj-22592	198	3	loss	loss	NOUN
brj-22592	198	4	function	function	NOUN
brj-22592	198	5	and	and	CCONJ
brj-22592	198	6	accuracy	accuracy	NOUN
brj-22592	198	7	of	of	ADP
brj-22592	198	8	networks	network	NOUN
brj-22592	198	9	(	(	PUNCT
brj-22592	198	10	a	a	X
brj-22592	198	11	)	)	PUNCT
brj-22592	198	12	loss	loss	NOUN
brj-22592	198	13	function	function	NOUN
brj-22592	198	14	and	and	CCONJ
brj-22592	198	15	accuracy	accuracy	NOUN
brj-22592	198	16	of	of	ADP
brj-22592	198	17	regnet+nam	regnet+nam	PROPN
brj-22592	198	18	network	network	NOUN
brj-22592	198	19	(	(	PUNCT
brj-22592	198	20	b	b	NOUN
brj-22592	198	21	)	)	PUNCT
brj-22592	198	22	loss	loss	NOUN
brj-22592	198	23	function	function	NOUN
brj-22592	198	24	and	and	CCONJ
brj-22592	198	25	accuracy	accuracy	NOUN
brj-22592	198	26	of	of	ADP
brj-22592	198	27	mobilenet	mobilenet	NOUN
brj-22592	198	28	-	-	PUNCT
brj-22592	198	29	v2	v2	NOUN
brj-22592	198	30	network	network	NOUN
brj-22592	198	31	(	(	PUNCT
brj-22592	198	32	c	c	NOUN
brj-22592	198	33	)	)	PUNCT
brj-22592	198	34	loss	loss	NOUN
brj-22592	198	35	function	function	NOUN
brj-22592	198	36	and	and	CCONJ
brj-22592	198	37	accuracy	accuracy	NOUN
brj-22592	198	38	of	of	ADP
brj-22592	198	39	efficientnet	efficientnet	NOUN
brj-22592	198	40	network	network	NOUN
brj-22592	198	41	(	(	PUNCT
brj-22592	198	42	d	d	NOUN
brj-22592	198	43	)	)	PUNCT
brj-22592	198	44	loss	loss	NOUN
brj-22592	198	45	function	function	NOUN
brj-22592	198	46	and	and	CCONJ
brj-22592	198	47	accuracy	accuracy	NOUN
brj-22592	198	48	of	of	ADP
brj-22592	198	49	vision	vision	NOUN
brj-22592	198	50	-	-	PUNCT
brj-22592	198	51	transformer	transformer	NOUN
brj-22592	198	52	network	network	NOUN
brj-22592	198	53	the	the	DET
brj-22592	198	54	training	training	NOUN
brj-22592	198	55	set	set	NOUN
brj-22592	198	56	samples	sample	NOUN
brj-22592	198	57	were	be	AUX
brj-22592	198	58	placed	place	VERB
brj-22592	198	59	into	into	ADP
brj-22592	198	60	the	the	DET
brj-22592	198	61	mobilenet	mobilenet	NOUN
brj-22592	198	62	-	-	PUNCT
brj-22592	198	63	v2	v2	NOUN
brj-22592	198	64	,	,	PUNCT
brj-22592	198	65	efficientnet	efficientnet	NOUN
brj-22592	198	66	,	,	PUNCT
brj-22592	198	67	and	and	CCONJ
brj-22592	198	68	vision	vision	NOUN
brj-22592	198	69	-	-	PUNCT
brj-22592	198	70	transformer	transformer	NOUN
brj-22592	198	71	networks	network	NOUN
brj-22592	198	72	.	.	PUNCT
brj-22592	199	1	the	the	DET
brj-22592	199	2	number	number	NOUN
brj-22592	199	3	of	of	ADP
brj-22592	199	4	experimental	experimental	ADJ
brj-22592	199	5	epochs	epoch	NOUN
brj-22592	199	6	was	be	AUX
brj-22592	199	7	set	set	VERB
brj-22592	199	8	to	to	ADP
brj-22592	199	9	120	120	NUM
brj-22592	199	10	,	,	PUNCT
brj-22592	199	11	the	the	DET
brj-22592	199	12	number	number	NOUN
brj-22592	199	13	of	of	ADP
brj-22592	199	14	learning	learn	VERB
brj-22592	199	15	rate	rate	NOUN
brj-22592	199	16	was	be	AUX
brj-22592	199	17	set	set	VERB
brj-22592	199	18	to	to	ADP
brj-22592	199	19	0.001	0.001	NUM
brj-22592	199	20	,	,	PUNCT
brj-22592	199	21	and	and	CCONJ
brj-22592	199	22	the	the	DET
brj-22592	199	23	number	number	NOUN
brj-22592	199	24	of	of	ADP
brj-22592	199	25	batch	batch	NOUN
brj-22592	199	26	size	size	NOUN
brj-22592	199	27	was	be	AUX
brj-22592	199	28	set	set	VERB
brj-22592	199	29	to	to	ADP
brj-22592	199	30	32	32	NUM
brj-22592	199	31	for	for	ADP
brj-22592	199	32	comparative	comparative	ADJ
brj-22592	199	33	experiments	experiment	NOUN
brj-22592	199	34	.	.	PUNCT
brj-22592	200	1	(	(	PUNCT
brj-22592	200	2	1	1	X
brj-22592	200	3	)	)	PUNCT
brj-22592	200	4	mobilenet	mobilenet	NOUN
brj-22592	200	5	-	-	PUNCT
brj-22592	200	6	v2	v2	NOUN
brj-22592	200	7	network	network	NOUN
brj-22592	200	8	realized	realize	VERB
brj-22592	200	9	wood	wood	NOUN
brj-22592	200	10	defect	defect	NOUN
brj-22592	200	11	classification	classification	NOUN
brj-22592	200	12	.	.	PUNCT
brj-22592	201	1	in	in	ADP
brj-22592	201	2	the	the	DET
brj-22592	201	3	118th	118th	ADJ
brj-22592	201	4	epoch	epoch	NOUN
brj-22592	201	5	,	,	PUNCT
brj-22592	201	6	the	the	DET
brj-22592	201	7	accuracy	accuracy	NOUN
brj-22592	201	8	of	of	ADP
brj-22592	201	9	the	the	DET
brj-22592	201	10	validation	validation	NOUN
brj-22592	201	11	set	set	NOUN
brj-22592	201	12	was	be	AUX
brj-22592	201	13	the	the	DET
brj-22592	201	14	highest	high	ADJ
brj-22592	201	15	,	,	PUNCT
brj-22592	201	16	with	with	ADP
brj-22592	201	17	a	a	DET
brj-22592	201	18	maximum	maximum	ADJ
brj-22592	201	19	accuracy	accuracy	NOUN
brj-22592	201	20	of	of	ADP
brj-22592	201	21	92.4	92.4	NUM
brj-22592	201	22	%	%	NOUN
brj-22592	201	23	.	.	PUNCT
brj-22592	202	1	the	the	DET
brj-22592	202	2	training	training	NOUN
brj-22592	202	3	set	set	VERB
brj-22592	202	4	loss	loss	NOUN
brj-22592	202	5	function	function	NOUN
brj-22592	202	6	and	and	CCONJ
brj-22592	202	7	validation	validation	NOUN
brj-22592	202	8	set	set	VERB
brj-22592	202	9	accuracy	accuracy	NOUN
brj-22592	202	10	of	of	ADP
brj-22592	202	11	mobilenet	mobilenet	NOUN
brj-22592	202	12	-	-	PUNCT
brj-22592	202	13	v2	v2	NOUN
brj-22592	202	14	network	network	NOUN
brj-22592	202	15	are	be	AUX
brj-22592	202	16	shown	show	VERB
brj-22592	202	17	in	in	ADP
brj-22592	202	18	fig	fig	NOUN
brj-22592	202	19	.	.	PUNCT
brj-22592	203	1	10(b	10(b	NUM
brj-22592	203	2	)	)	PUNCT
brj-22592	203	3	.	.	PUNCT
brj-22592	204	1	(	(	PUNCT
brj-22592	204	2	2	2	X
brj-22592	204	3	)	)	PUNCT
brj-22592	204	4	efficientnet	efficientnet	NOUN
brj-22592	204	5	network	network	NOUN
brj-22592	204	6	realized	realize	VERB
brj-22592	204	7	wood	wood	NOUN
brj-22592	204	8	defect	defect	NOUN
brj-22592	204	9	classification	classification	NOUN
brj-22592	204	10	.	.	PUNCT
brj-22592	205	1	in	in	ADP
brj-22592	205	2	the	the	DET
brj-22592	205	3	103rd	103rd	ADJ
brj-22592	205	4	epoch	epoch	NOUN
brj-22592	205	5	,	,	PUNCT
brj-22592	205	6	the	the	DET
brj-22592	205	7	accuracy	accuracy	NOUN
brj-22592	205	8	of	of	ADP
brj-22592	205	9	the	the	DET
brj-22592	205	10	validation	validation	NOUN
brj-22592	205	11	set	set	NOUN
brj-22592	205	12	was	be	AUX
brj-22592	205	13	the	the	DET
brj-22592	205	14	highest	high	ADJ
brj-22592	205	15	,	,	PUNCT
brj-22592	205	16	with	with	ADP
brj-22592	205	17	a	a	DET
brj-22592	205	18	maximum	maximum	ADJ
brj-22592	205	19	accuracy	accuracy	NOUN
brj-22592	205	20	of	of	ADP
brj-22592	205	21	94.2	94.2	NUM
brj-22592	205	22	%	%	NOUN
brj-22592	205	23	.	.	PUNCT
brj-22592	206	1	the	the	DET
brj-22592	206	2	training	training	NOUN
brj-22592	206	3	set	set	VERB
brj-22592	206	4	loss	loss	NOUN
brj-22592	206	5	function	function	NOUN
brj-22592	206	6	and	and	CCONJ
brj-22592	206	7	validation	validation	NOUN
brj-22592	206	8	set	set	VERB
brj-22592	206	9	accuracy	accuracy	NOUN
brj-22592	206	10	of	of	ADP
brj-22592	206	11	efficientnet	efficientnet	NOUN
brj-22592	206	12	network	network	NOUN
brj-22592	206	13	are	be	AUX
brj-22592	206	14	shown	show	VERB
brj-22592	206	15	in	in	ADP
brj-22592	206	16	fig	fig	NOUN
brj-22592	206	17	.	.	PUNCT
brj-22592	207	1	10(c	10(c	NUM
brj-22592	207	2	)	)	PUNCT
brj-22592	207	3	.	.	PUNCT
brj-22592	208	1	(	(	PUNCT
brj-22592	208	2	3	3	X
brj-22592	208	3	)	)	PUNCT
brj-22592	208	4	vision	vision	NOUN
brj-22592	208	5	-	-	PUNCT
brj-22592	208	6	transformer	transformer	NOUN
brj-22592	208	7	networks	network	NOUN
brj-22592	208	8	realized	realize	VERB
brj-22592	208	9	wood	wood	NOUN
brj-22592	208	10	defect	defect	NOUN
brj-22592	208	11	classification	classification	NOUN
brj-22592	208	12	.	.	PUNCT
brj-22592	209	1	in	in	ADP
brj-22592	209	2	the	the	DET
brj-22592	209	3	115th	115th	ADJ
brj-22592	209	4	epoch	epoch	NOUN
brj-22592	209	5	,	,	PUNCT
brj-22592	209	6	the	the	DET
brj-22592	209	7	accuracy	accuracy	NOUN
brj-22592	209	8	of	of	ADP
brj-22592	209	9	the	the	DET
brj-22592	209	10	validation	validation	NOUN
brj-22592	209	11	set	set	NOUN
brj-22592	209	12	was	be	AUX
brj-22592	209	13	the	the	DET
brj-22592	209	14	highest	high	ADJ
brj-22592	209	15	,	,	PUNCT
brj-22592	209	16	with	with	ADP
brj-22592	209	17	a	a	DET
brj-22592	209	18	maximum	maximum	ADJ
brj-22592	209	19	accuracy	accuracy	NOUN
brj-22592	209	20	of	of	ADP
brj-22592	209	21	94.6	94.6	NUM
brj-22592	209	22	%	%	NOUN
brj-22592	209	23	.	.	PUNCT
brj-22592	210	1	the	the	DET
brj-22592	210	2	training	training	NOUN
brj-22592	210	3	set	set	VERB
brj-22592	210	4	loss	loss	NOUN
brj-22592	210	5	function	function	NOUN
brj-22592	210	6	and	and	CCONJ
brj-22592	210	7	validation	validation	NOUN
brj-22592	210	8	set	set	VERB
brj-22592	210	9	accuracy	accuracy	NOUN
brj-22592	210	10	of	of	ADP
brj-22592	210	11	vision	vision	NOUN
brj-22592	210	12	-	-	PUNCT
brj-22592	210	13	transformer	transformer	NOUN
brj-22592	210	14	network	network	NOUN
brj-22592	210	15	are	be	AUX
brj-22592	210	16	shown	show	VERB
brj-22592	210	17	in	in	ADP
brj-22592	210	18	fig	fig	NOUN
brj-22592	210	19	.	.	PUNCT
brj-22592	211	1	10(d	10(d	NUM
brj-22592	211	2	)	)	PUNCT
brj-22592	211	3	.	.	PUNCT
brj-22592	212	1	model	model	NOUN
brj-22592	212	2	evaluation	evaluation	NOUN
brj-22592	212	3	index	index	NOUN
brj-22592	212	4	the	the	DET
brj-22592	212	5	weight	weight	NOUN
brj-22592	212	6	of	of	ADP
brj-22592	212	7	the	the	DET
brj-22592	212	8	highest	high	ADJ
brj-22592	212	9	accuracy	accuracy	NOUN
brj-22592	212	10	on	on	ADP
brj-22592	212	11	each	each	DET
brj-22592	212	12	network	network	NOUN
brj-22592	212	13	model	model	NOUN
brj-22592	212	14	training	training	NOUN
brj-22592	212	15	set	set	NOUN
brj-22592	212	16	was	be	AUX
brj-22592	212	17	saved	save	VERB
brj-22592	212	18	,	,	PUNCT
brj-22592	212	19	which	which	PRON
brj-22592	212	20	was	be	AUX
brj-22592	212	21	convenient	convenient	ADJ
brj-22592	212	22	for	for	SCONJ
brj-22592	212	23	the	the	DET
brj-22592	212	24	test	test	NOUN
brj-22592	212	25	set	set	VERB
brj-22592	212	26	to	to	PART
brj-22592	212	27	identify	identify	VERB
brj-22592	212	28	and	and	CCONJ
brj-22592	212	29	predict	predict	VERB
brj-22592	212	30	the	the	DET
brj-22592	212	31	category	category	NOUN
brj-22592	212	32	of	of	ADP
brj-22592	212	33	wood	wood	NOUN
brj-22592	212	34	defect	defect	NOUN
brj-22592	212	35	peer	peer	NOUN
brj-22592	212	36	-	-	PUNCT
brj-22592	212	37	reviewed	review	VERB
brj-22592	212	38	article	article	NOUN
brj-22592	212	39	bioresources.com	bioresources.com	X
brj-22592	212	40	xie	xie	PROPN
brj-22592	212	41	&	&	CCONJ
brj-22592	212	42	ling	ling	PROPN
brj-22592	212	43	(	(	PUNCT
brj-22592	212	44	2023	2023	NUM
brj-22592	212	45	)	)	PUNCT
brj-22592	212	46	.	.	PUNCT
brj-22592	213	1	“	"	PUNCT
brj-22592	213	2	wood	wood	NOUN
brj-22592	213	3	defect	defect	NOUN
brj-22592	213	4	classification	classification	NOUN
brj-22592	213	5	,	,	PUNCT
brj-22592	213	6	”	"	PUNCT
brj-22592	213	7	bioresources	bioresource	NOUN
brj-22592	213	8	18(4	18(4	NUM
brj-22592	213	9	)	)	PUNCT
brj-22592	213	10	,	,	PUNCT
brj-22592	213	11	7663	7663	NUM
brj-22592	213	12	-	-	SYM
brj-22592	213	13	7680	7680	NUM
brj-22592	213	14	.	.	PUNCT
brj-22592	214	1	7675	7675	NUM
brj-22592	214	2	samples	sample	NOUN
brj-22592	214	3	.	.	PUNCT
brj-22592	215	1	tp	tp	NOUN
brj-22592	215	2	represents	represent	VERB
brj-22592	215	3	the	the	DET
brj-22592	215	4	positive	positive	ADJ
brj-22592	215	5	sample	sample	NOUN
brj-22592	215	6	,	,	PUNCT
brj-22592	215	7	predicted	predict	VERB
brj-22592	215	8	as	as	ADP
brj-22592	215	9	positive	positive	ADJ
brj-22592	215	10	by	by	ADP
brj-22592	215	11	the	the	DET
brj-22592	215	12	model	model	NOUN
brj-22592	215	13	,	,	PUNCT
brj-22592	215	14	tn	tn	PROPN
brj-22592	215	15	represents	represent	VERB
brj-22592	215	16	the	the	DET
brj-22592	215	17	negative	negative	ADJ
brj-22592	215	18	sample	sample	NOUN
brj-22592	215	19	,	,	PUNCT
brj-22592	215	20	predicted	predict	VERB
brj-22592	215	21	as	as	ADP
brj-22592	215	22	negative	negative	ADJ
brj-22592	215	23	by	by	ADP
brj-22592	215	24	the	the	DET
brj-22592	215	25	model	model	NOUN
brj-22592	215	26	,	,	PUNCT
brj-22592	215	27	fp	fp	PROPN
brj-22592	215	28	denotes	denote	VERB
brj-22592	215	29	the	the	DET
brj-22592	215	30	negative	negative	ADJ
brj-22592	215	31	sample	sample	NOUN
brj-22592	215	32	,	,	PUNCT
brj-22592	215	33	predicted	predict	VERB
brj-22592	215	34	as	as	ADP
brj-22592	215	35	positive	positive	ADJ
brj-22592	215	36	by	by	ADP
brj-22592	215	37	the	the	DET
brj-22592	215	38	model	model	NOUN
brj-22592	215	39	,	,	PUNCT
brj-22592	215	40	and	and	CCONJ
brj-22592	215	41	fn	fn	NOUN
brj-22592	215	42	denotes	denote	NOUN
brj-22592	215	43	the	the	DET
brj-22592	215	44	positive	positive	ADJ
brj-22592	215	45	sample	sample	NOUN
brj-22592	215	46	predicted	predict	VERB
brj-22592	215	47	as	as	ADP
brj-22592	215	48	negative	negative	ADJ
brj-22592	215	49	by	by	ADP
brj-22592	215	50	the	the	DET
brj-22592	215	51	model	model	NOUN
brj-22592	215	52	.	.	PUNCT
brj-22592	216	1	a	a	DET
brj-22592	216	2	confusion	confusion	NOUN
brj-22592	216	3	matrix	matrix	NOUN
brj-22592	216	4	diagram	diagram	NOUN
brj-22592	216	5	can	can	AUX
brj-22592	216	6	be	be	AUX
brj-22592	216	7	drawn	draw	VERB
brj-22592	216	8	based	base	VERB
brj-22592	216	9	on	on	ADP
brj-22592	216	10	the	the	DET
brj-22592	216	11	prediction	prediction	NOUN
brj-22592	216	12	and	and	CCONJ
brj-22592	216	13	classification	classification	NOUN
brj-22592	216	14	of	of	ADP
brj-22592	216	15	the	the	DET
brj-22592	216	16	various	various	ADJ
brj-22592	216	17	samples	sample	NOUN
brj-22592	216	18	.	.	PUNCT
brj-22592	217	1	using	use	VERB
brj-22592	217	2	tp	tp	ADP
brj-22592	217	3	,	,	PUNCT
brj-22592	217	4	tn	tn	PROPN
brj-22592	217	5	,	,	PUNCT
brj-22592	217	6	fp	fp	NOUN
brj-22592	217	7	,	,	PUNCT
brj-22592	217	8	and	and	CCONJ
brj-22592	217	9	fn	fn	NOUN
brj-22592	217	10	,	,	PUNCT
brj-22592	217	11	performance	performance	NOUN
brj-22592	217	12	indices	index	NOUN
brj-22592	217	13	,	,	PUNCT
brj-22592	217	14	such	such	ADJ
brj-22592	217	15	as	as	ADP
brj-22592	217	16	accuracy	accuracy	NOUN
brj-22592	217	17	,	,	PUNCT
brj-22592	217	18	recall	recall	NOUN
brj-22592	217	19	,	,	PUNCT
brj-22592	217	20	and	and	CCONJ
brj-22592	217	21	specificity	specificity	NOUN
brj-22592	217	22	were	be	AUX
brj-22592	217	23	obtained	obtain	VERB
brj-22592	217	24	,	,	PUNCT
brj-22592	217	25	where	where	SCONJ
brj-22592	217	26	the	the	DET
brj-22592	217	27	accuracy	accuracy	NOUN
brj-22592	217	28	rate	rate	NOUN
brj-22592	217	29	denotes	denote	VERB
brj-22592	217	30	the	the	DET
brj-22592	217	31	ratio	ratio	NOUN
brj-22592	217	32	of	of	ADP
brj-22592	217	33	the	the	DET
brj-22592	217	34	number	number	NOUN
brj-22592	217	35	of	of	ADP
brj-22592	217	36	positive	positive	ADJ
brj-22592	217	37	samples	sample	NOUN
brj-22592	217	38	correctly	correctly	ADV
brj-22592	217	39	predicted	predict	VERB
brj-22592	217	40	to	to	ADP
brj-22592	217	41	the	the	DET
brj-22592	217	42	number	number	NOUN
brj-22592	217	43	of	of	ADP
brj-22592	217	44	positive	positive	ADJ
brj-22592	217	45	samples	sample	NOUN
brj-22592	217	46	predicted	predict	VERB
brj-22592	217	47	as	as	SCONJ
brj-22592	217	48	follows	follow	VERB
brj-22592	217	49	:	:	PUNCT
brj-22592	217	50	fptp	fptp	PROPN
brj-22592	217	51	tp	tp	PART
brj-22592	217	52	precision	precision	VERB
brj-22592	217	53	+	+	X
brj-22592	218	1	=	=	SYM
brj-22592	218	2	(	(	PUNCT
brj-22592	218	3	7	7	X
brj-22592	218	4	)	)	PUNCT
brj-22592	218	5	the	the	DET
brj-22592	218	6	recall	recall	NOUN
brj-22592	218	7	rate	rate	NOUN
brj-22592	218	8	refers	refer	VERB
brj-22592	218	9	to	to	ADP
brj-22592	218	10	the	the	DET
brj-22592	218	11	ratio	ratio	NOUN
brj-22592	218	12	of	of	ADP
brj-22592	218	13	the	the	DET
brj-22592	218	14	number	number	NOUN
brj-22592	218	15	of	of	ADP
brj-22592	218	16	correctly	correctly	ADV
brj-22592	218	17	predicted	predict	VERB
brj-22592	218	18	positive	positive	ADJ
brj-22592	218	19	samples	sample	NOUN
brj-22592	218	20	to	to	ADP
brj-22592	218	21	the	the	DET
brj-22592	218	22	total	total	ADJ
brj-22592	218	23	number	number	NOUN
brj-22592	218	24	of	of	ADP
brj-22592	218	25	real	real	ADJ
brj-22592	218	26	samples	sample	NOUN
brj-22592	218	27	.	.	PUNCT
brj-22592	219	1	fntp	fntp	NOUN
brj-22592	219	2	tp	tp	PART
brj-22592	219	3	recall	recall	VERB
brj-22592	219	4	+	+	X
brj-22592	220	1	=	=	SYM
brj-22592	221	1	(	(	PUNCT
brj-22592	221	2	8)	8)	NUM
brj-22592	221	3	specificity	specificity	NOUN
brj-22592	221	4	as	as	SCONJ
brj-22592	221	5	the	the	DET
brj-22592	221	6	specificity	specificity	NOUN
brj-22592	221	7	increases	increase	VERB
brj-22592	221	8	,	,	PUNCT
brj-22592	221	9	the	the	DET
brj-22592	221	10	probability	probability	NOUN
brj-22592	221	11	of	of	ADP
brj-22592	221	12	false	false	ADJ
brj-22592	221	13	detections	detection	NOUN
brj-22592	221	14	decreases	decrease	VERB
brj-22592	221	15	.	.	PUNCT
brj-22592	222	1	fptn	fptn	VERB
brj-22592	222	2	tn	tn	NOUN
brj-22592	222	3	yspecificit	yspecificit	NOUN
brj-22592	223	1	+	+	CCONJ
brj-22592	224	1	=	=	SYM
brj-22592	224	2	(	(	PUNCT
brj-22592	224	3	9	9	X
brj-22592	224	4	)	)	PUNCT
brj-22592	224	5	fig	fig	NOUN
brj-22592	224	6	.	.	PUNCT
brj-22592	225	1	11	11	NUM
brj-22592	225	2	.	.	PUNCT
brj-22592	225	3	networks	network	NOUN
brj-22592	225	4	confusion	confusion	NOUN
brj-22592	225	5	matrix	matrix	NOUN
brj-22592	225	6	(	(	PUNCT
brj-22592	225	7	a	a	X
brj-22592	225	8	)	)	PUNCT
brj-22592	225	9	ghostregnet	ghostregnet	NOUN
brj-22592	225	10	+	+	CCONJ
brj-22592	225	11	nam	nam	NOUN
brj-22592	225	12	network	network	NOUN
brj-22592	225	13	confusion	confusion	NOUN
brj-22592	225	14	matrix	matrix	NOUN
brj-22592	225	15	(	(	PUNCT
brj-22592	225	16	b	b	NOUN
brj-22592	225	17	)	)	PUNCT
brj-22592	225	18	regnet	regnet	NOUN
brj-22592	225	19	network	network	NOUN
brj-22592	225	20	confusion	confusion	NOUN
brj-22592	225	21	matrix	matrix	NOUN
brj-22592	225	22	(	(	PUNCT
brj-22592	225	23	c	c	NOUN
brj-22592	225	24	)	)	PUNCT
brj-22592	225	25	regnet+se	regnet+se	ADJ
brj-22592	225	26	network	network	NOUN
brj-22592	225	27	confusion	confusion	NOUN
brj-22592	225	28	matrix	matrix	NOUN
brj-22592	225	29	(	(	PUNCT
brj-22592	225	30	d	d	NOUN
brj-22592	225	31	)	)	PUNCT
brj-22592	226	1	regnet	regnet	NOUN
brj-22592	226	2	+	+	CCONJ
brj-22592	226	3	eca	eca	NOUN
brj-22592	226	4	network	network	NOUN
brj-22592	226	5	confusion	confusion	NOUN
brj-22592	226	6	matrix	matrix	NOUN
brj-22592	226	7	(	(	PUNCT
brj-22592	226	8	a	a	X
brj-22592	226	9	)	)	PUNCT
brj-22592	226	10	(	(	PUNCT
brj-22592	226	11	b	b	X
brj-22592	226	12	)	)	PUNCT
brj-22592	226	13	(	(	PUNCT
brj-22592	226	14	c	c	X
brj-22592	226	15	)	)	PUNCT
brj-22592	226	16	(	(	PUNCT
brj-22592	226	17	d	d	X
brj-22592	226	18	)	)	PUNCT
brj-22592	226	19	peer	peer	NOUN
brj-22592	226	20	-	-	PUNCT
brj-22592	226	21	reviewed	review	VERB
brj-22592	226	22	article	article	NOUN
brj-22592	226	23	bioresources.com	bioresources.com	X
brj-22592	226	24	xie	xie	PROPN
brj-22592	226	25	&	&	CCONJ
brj-22592	226	26	ling	ling	PROPN
brj-22592	226	27	(	(	PUNCT
brj-22592	226	28	2023	2023	NUM
brj-22592	226	29	)	)	PUNCT
brj-22592	226	30	.	.	PUNCT
brj-22592	227	1	“	"	PUNCT
brj-22592	227	2	wood	wood	NOUN
brj-22592	227	3	defect	defect	NOUN
brj-22592	227	4	classification	classification	NOUN
brj-22592	227	5	,	,	PUNCT
brj-22592	227	6	”	"	PUNCT
brj-22592	227	7	bioresources	bioresource	NOUN
brj-22592	227	8	18(4	18(4	NUM
brj-22592	227	9	)	)	PUNCT
brj-22592	227	10	,	,	PUNCT
brj-22592	227	11	7663	7663	NUM
brj-22592	227	12	-	-	SYM
brj-22592	227	13	7680	7680	NUM
brj-22592	227	14	.	.	PUNCT
brj-22592	228	1	7676	7676	NUM
brj-22592	228	2	model	model	NOUN
brj-22592	228	3	performance	performance	NOUN
brj-22592	228	4	analysis	analysis	NOUN
brj-22592	228	5	the	the	DET
brj-22592	228	6	predictions	prediction	NOUN
brj-22592	228	7	of	of	ADP
brj-22592	228	8	regnet	regnet	NOUN
brj-22592	228	9	,	,	PUNCT
brj-22592	228	10	regnet+se	regnet+se	ADJ
brj-22592	228	11	,	,	PUNCT
brj-22592	228	12	regnet+eca	regnet+eca	PROPN
brj-22592	228	13	,	,	PUNCT
brj-22592	228	14	regnet+nam	regnet+nam	PROPN
brj-22592	228	15	,	,	PUNCT
brj-22592	228	16	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	228	17	,	,	PUNCT
brj-22592	228	18	mobilenet	mobilenet	NOUN
brj-22592	228	19	-	-	PUNCT
brj-22592	228	20	v2	v2	NOUN
brj-22592	228	21	,	,	PUNCT
brj-22592	228	22	efficientnet	efficientnet	NOUN
brj-22592	228	23	,	,	PUNCT
brj-22592	228	24	and	and	CCONJ
brj-22592	228	25	vision	vision	NOUN
brj-22592	228	26	-	-	PUNCT
brj-22592	228	27	transformer	transformer	NOUN
brj-22592	228	28	networks	network	NOUN
brj-22592	228	29	on	on	ADP
brj-22592	228	30	the	the	DET
brj-22592	228	31	test	test	NOUN
brj-22592	228	32	set	set	NOUN
brj-22592	228	33	of	of	ADP
brj-22592	228	34	wood	wood	NOUN
brj-22592	228	35	defects	defect	NOUN
brj-22592	228	36	are	be	AUX
brj-22592	228	37	shown	show	VERB
brj-22592	228	38	in	in	ADP
brj-22592	228	39	the	the	DET
brj-22592	228	40	confusion	confusion	NOUN
brj-22592	228	41	matrix	matrix	NOUN
brj-22592	228	42	figs	fig	NOUN
brj-22592	228	43	.	.	PUNCT
brj-22592	229	1	11	11	NUM
brj-22592	229	2	and	and	CCONJ
brj-22592	229	3	12	12	NUM
brj-22592	229	4	,	,	PUNCT
brj-22592	229	5	respectively	respectively	ADV
brj-22592	229	6	.	.	PUNCT
brj-22592	230	1	the	the	DET
brj-22592	230	2	abscissa	abscissa	NOUN
brj-22592	230	3	of	of	ADP
brj-22592	230	4	the	the	DET
brj-22592	230	5	confusion	confusion	NOUN
brj-22592	230	6	matrix	matrix	NOUN
brj-22592	230	7	figure	figure	NOUN
brj-22592	230	8	represents	represent	VERB
brj-22592	230	9	the	the	DET
brj-22592	230	10	real	real	ADJ
brj-22592	230	11	label	label	NOUN
brj-22592	230	12	of	of	ADP
brj-22592	230	13	the	the	DET
brj-22592	230	14	sample	sample	NOUN
brj-22592	230	15	and	and	CCONJ
brj-22592	230	16	the	the	DET
brj-22592	230	17	ordinate	ordinate	NOUN
brj-22592	230	18	represents	represent	VERB
brj-22592	230	19	the	the	DET
brj-22592	230	20	predicted	predict	VERB
brj-22592	230	21	label	label	NOUN
brj-22592	230	22	of	of	ADP
brj-22592	230	23	the	the	DET
brj-22592	230	24	sample	sample	NOUN
brj-22592	230	25	.	.	PUNCT
brj-22592	231	1	by	by	ADP
brj-22592	231	2	using	use	VERB
brj-22592	231	3	the	the	DET
brj-22592	231	4	confusion	confusion	NOUN
brj-22592	231	5	matrix	matrix	NOUN
brj-22592	231	6	predicted	predict	VERB
brj-22592	231	7	by	by	ADP
brj-22592	231	8	each	each	DET
brj-22592	231	9	model	model	NOUN
brj-22592	231	10	on	on	ADP
brj-22592	231	11	the	the	DET
brj-22592	231	12	test	test	NOUN
brj-22592	231	13	set	set	VERB
brj-22592	231	14	and	and	CCONJ
brj-22592	231	15	color	color	NOUN
brj-22592	231	16	depth	depth	NOUN
brj-22592	231	17	of	of	ADP
brj-22592	231	18	the	the	DET
brj-22592	231	19	color	color	NOUN
brj-22592	231	20	block	block	NOUN
brj-22592	231	21	of	of	ADP
brj-22592	231	22	the	the	DET
brj-22592	231	23	confusion	confusion	NOUN
brj-22592	231	24	matrix	matrix	NOUN
brj-22592	231	25	,	,	PUNCT
brj-22592	231	26	it	it	PRON
brj-22592	231	27	can	can	AUX
brj-22592	231	28	be	be	AUX
brj-22592	231	29	intuitively	intuitively	ADV
brj-22592	231	30	observed	observe	VERB
brj-22592	231	31	that	that	SCONJ
brj-22592	231	32	the	the	DET
brj-22592	231	33	prediction	prediction	NOUN
brj-22592	231	34	effect	effect	NOUN
brj-22592	231	35	of	of	ADP
brj-22592	231	36	the	the	DET
brj-22592	231	37	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	231	38	network	network	NOUN
brj-22592	231	39	is	be	AUX
brj-22592	231	40	better	well	ADJ
brj-22592	231	41	,	,	PUNCT
brj-22592	231	42	and	and	CCONJ
brj-22592	231	43	there	there	PRON
brj-22592	231	44	are	be	VERB
brj-22592	231	45	a	a	DET
brj-22592	231	46	large	large	ADJ
brj-22592	231	47	number	number	NOUN
brj-22592	231	48	of	of	ADP
brj-22592	231	49	positive	positive	ADJ
brj-22592	231	50	samples	sample	NOUN
brj-22592	231	51	predicted	predict	VERB
brj-22592	231	52	as	as	ADP
brj-22592	231	53	positive	positive	ADJ
brj-22592	231	54	by	by	ADP
brj-22592	231	55	the	the	DET
brj-22592	231	56	network	network	NOUN
brj-22592	231	57	model	model	NOUN
brj-22592	231	58	.	.	PUNCT
brj-22592	232	1	(	(	PUNCT
brj-22592	232	2	a	a	X
brj-22592	232	3	)	)	PUNCT
brj-22592	232	4	(	(	PUNCT
brj-22592	232	5	b	b	X
brj-22592	232	6	)	)	PUNCT
brj-22592	232	7	(	(	PUNCT
brj-22592	232	8	c	c	X
brj-22592	232	9	)	)	PUNCT
brj-22592	232	10	(	(	PUNCT
brj-22592	232	11	d	d	X
brj-22592	232	12	)	)	PUNCT
brj-22592	232	13	fig	fig	NOUN
brj-22592	232	14	.	.	PUNCT
brj-22592	233	1	12	12	NUM
brj-22592	233	2	.	.	PUNCT
brj-22592	234	1	networks	network	NOUN
brj-22592	234	2	confusion	confusion	NOUN
brj-22592	234	3	matrix	matrix	NOUN
brj-22592	234	4	(	(	PUNCT
brj-22592	234	5	a	a	NOUN
brj-22592	234	6	)	)	PUNCT
brj-22592	234	7	regnet	regnet	NOUN
brj-22592	234	8	+	+	CCONJ
brj-22592	234	9	nam	nam	NOUN
brj-22592	234	10	network	network	NOUN
brj-22592	234	11	confusion	confusion	NOUN
brj-22592	234	12	matrix	matrix	NOUN
brj-22592	234	13	(	(	PUNCT
brj-22592	234	14	b	b	NOUN
brj-22592	234	15	)	)	PUNCT
brj-22592	234	16	mobilenet	mobilenet	NOUN
brj-22592	234	17	-	-	PUNCT
brj-22592	234	18	v2	v2	NOUN
brj-22592	234	19	network	network	NOUN
brj-22592	234	20	confusion	confusion	NOUN
brj-22592	234	21	matrix	matrix	NOUN
brj-22592	234	22	(	(	PUNCT
brj-22592	234	23	c	c	NOUN
brj-22592	234	24	)	)	PUNCT
brj-22592	234	25	efficientnet	efficientnet	NOUN
brj-22592	234	26	network	network	NOUN
brj-22592	234	27	confusion	confusion	NOUN
brj-22592	234	28	matrix	matrix	NOUN
brj-22592	234	29	(	(	PUNCT
brj-22592	234	30	d	d	NOUN
brj-22592	234	31	)	)	PUNCT
brj-22592	234	32	vision	vision	NOUN
brj-22592	234	33	-	-	PUNCT
brj-22592	234	34	transformer	transformer	NOUN
brj-22592	234	35	network	network	NOUN
brj-22592	234	36	confusion	confusion	NOUN
brj-22592	234	37	matrix	matrix	NOUN
brj-22592	234	38	the	the	DET
brj-22592	234	39	performances	performance	NOUN
brj-22592	234	40	of	of	ADP
brj-22592	234	41	four	four	NUM
brj-22592	234	42	regnet	regnet	NOUN
brj-22592	234	43	network	network	NOUN
brj-22592	234	44	improved	improve	VERB
brj-22592	234	45	schemes	scheme	NOUN
brj-22592	234	46	are	be	AUX
brj-22592	234	47	compared	compare	VERB
brj-22592	234	48	in	in	ADP
brj-22592	234	49	table	table	NOUN
brj-22592	234	50	2	2	NUM
brj-22592	234	51	.	.	PUNCT
brj-22592	235	1	the	the	DET
brj-22592	235	2	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	235	3	network	network	NOUN
brj-22592	235	4	exhibited	exhibit	VERB
brj-22592	235	5	the	the	DET
brj-22592	235	6	highest	high	ADJ
brj-22592	235	7	precision	precision	NOUN
brj-22592	235	8	,	,	PUNCT
brj-22592	235	9	recall	recall	NOUN
brj-22592	235	10	,	,	PUNCT
brj-22592	235	11	and	and	CCONJ
brj-22592	235	12	specificity	specificity	NOUN
brj-22592	235	13	values	value	NOUN
brj-22592	235	14	for	for	ADP
brj-22592	235	15	the	the	DET
brj-22592	235	16	three	three	NUM
brj-22592	235	17	types	type	NOUN
brj-22592	235	18	of	of	ADP
brj-22592	235	19	wood	wood	NOUN
brj-22592	235	20	defects	defect	NOUN
brj-22592	235	21	and	and	CCONJ
brj-22592	235	22	lower	low	ADJ
brj-22592	235	23	false	false	ADJ
brj-22592	235	24	detection	detection	NOUN
brj-22592	235	25	probability	probability	NOUN
brj-22592	235	26	.	.	PUNCT
brj-22592	236	1	the	the	DET
brj-22592	236	2	samples	sample	NOUN
brj-22592	236	3	of	of	ADP
brj-22592	236	4	the	the	DET
brj-22592	236	5	wood	wood	NOUN
brj-22592	236	6	defect	defect	NOUN
brj-22592	236	7	test	test	NOUN
brj-22592	236	8	were	be	AUX
brj-22592	236	9	set	set	VERB
brj-22592	236	10	into	into	ADP
brj-22592	236	11	the	the	DET
brj-22592	236	12	sample	sample	NOUN
brj-22592	236	13	library	library	NOUN
brj-22592	236	14	of	of	ADP
brj-22592	236	15	the	the	DET
brj-22592	236	16	wood	wood	NOUN
brj-22592	236	17	defect	defect	NOUN
brj-22592	236	18	and	and	CCONJ
brj-22592	236	19	the	the	DET
brj-22592	236	20	optimal	optimal	ADJ
brj-22592	236	21	weight	weight	NOUN
brj-22592	236	22	pre	pre	ADJ
brj-22592	236	23	-	-	ADJ
brj-22592	236	24	training	training	ADJ
brj-22592	236	25	model	model	NOUN
brj-22592	236	26	is	be	AUX
brj-22592	236	27	trained	train	VERB
brj-22592	236	28	on	on	ADP
brj-22592	236	29	the	the	DET
brj-22592	236	30	set	set	NOUN
brj-22592	236	31	to	to	PART
brj-22592	236	32	obtain	obtain	VERB
brj-22592	236	33	the	the	DET
brj-22592	236	34	recognition	recognition	NOUN
brj-22592	236	35	accuracy	accuracy	NOUN
brj-22592	236	36	,	,	PUNCT
brj-22592	236	37	network	network	NOUN
brj-22592	236	38	parameter	parameter	NOUN
brj-22592	236	39	quantity	quantity	NOUN
brj-22592	236	40	,	,	PUNCT
brj-22592	236	41	and	and	CCONJ
brj-22592	236	42	average	average	ADJ
brj-22592	236	43	image	image	NOUN
brj-22592	236	44	recognition	recognition	NOUN
brj-22592	236	45	time	time	NOUN
brj-22592	236	46	of	of	ADP
brj-22592	236	47	four	four	NUM
brj-22592	236	48	regnet	regnet	NOUN
brj-22592	236	49	network	network	NOUN
brj-22592	236	50	improved	improve	VERB
brj-22592	236	51	schemes	scheme	NOUN
brj-22592	236	52	.	.	PUNCT
brj-22592	237	1	a	a	DET
brj-22592	237	2	comparison	comparison	NOUN
brj-22592	237	3	of	of	ADP
brj-22592	237	4	the	the	DET
brj-22592	237	5	data	data	NOUN
brj-22592	237	6	is	be	AUX
brj-22592	237	7	presented	present	VERB
brj-22592	237	8	in	in	ADP
brj-22592	237	9	table	table	NOUN
brj-22592	237	10	3	3	NUM
brj-22592	237	11	.	.	PUNCT
brj-22592	237	12	compared	compare	VERB
brj-22592	237	13	with	with	ADP
brj-22592	237	14	other	other	ADJ
brj-22592	237	15	improved	improved	ADJ
brj-22592	237	16	schemes	scheme	NOUN
brj-22592	237	17	,	,	PUNCT
brj-22592	237	18	the	the	DET
brj-22592	237	19	parameters	parameter	NOUN
brj-22592	237	20	of	of	ADP
brj-22592	237	21	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	237	22	network	network	NOUN
brj-22592	237	23	are	be	AUX
brj-22592	237	24	peer	peer	NOUN
brj-22592	237	25	-	-	PUNCT
brj-22592	237	26	reviewed	review	VERB
brj-22592	237	27	article	article	NOUN
brj-22592	237	28	bioresources.com	bioresources.com	X
brj-22592	237	29	xie	xie	PROPN
brj-22592	237	30	&	&	CCONJ
brj-22592	237	31	ling	ling	PROPN
brj-22592	237	32	(	(	PUNCT
brj-22592	237	33	2023	2023	NUM
brj-22592	237	34	)	)	PUNCT
brj-22592	237	35	.	.	PUNCT
brj-22592	238	1	“	"	PUNCT
brj-22592	238	2	wood	wood	NOUN
brj-22592	238	3	defect	defect	NOUN
brj-22592	238	4	classification	classification	NOUN
brj-22592	238	5	,	,	PUNCT
brj-22592	238	6	”	"	PUNCT
brj-22592	238	7	bioresources	bioresource	NOUN
brj-22592	238	8	18(4	18(4	NUM
brj-22592	238	9	)	)	PUNCT
brj-22592	238	10	,	,	PUNCT
brj-22592	238	11	7663	7663	NUM
brj-22592	238	12	-	-	SYM
brj-22592	238	13	7680	7680	NUM
brj-22592	238	14	.	.	PUNCT
brj-22592	238	15	7677	7677	NUM
brj-22592	238	16	significantly	significantly	ADV
brj-22592	238	17	reduced	reduce	VERB
brj-22592	238	18	,	,	PUNCT
brj-22592	238	19	the	the	DET
brj-22592	238	20	average	average	ADJ
brj-22592	238	21	image	image	NOUN
brj-22592	238	22	recognition	recognition	NOUN
brj-22592	238	23	time	time	NOUN
brj-22592	238	24	was	be	AUX
brj-22592	238	25	significantly	significantly	ADV
brj-22592	238	26	reduced	reduce	VERB
brj-22592	238	27	,	,	PUNCT
brj-22592	238	28	and	and	CCONJ
brj-22592	238	29	the	the	DET
brj-22592	238	30	recognition	recognition	NOUN
brj-22592	238	31	accuracy	accuracy	NOUN
brj-22592	238	32	was	be	AUX
brj-22592	238	33	improved	improve	VERB
brj-22592	238	34	.	.	PUNCT
brj-22592	239	1	table	table	NOUN
brj-22592	239	2	2	2	NUM
brj-22592	239	3	.	.	PUNCT
brj-22592	239	4	performance	performance	NOUN
brj-22592	239	5	comparison	comparison	NOUN
brj-22592	239	6	of	of	ADP
brj-22592	239	7	regnet	regnet	NOUN
brj-22592	239	8	network	network	NOUN
brj-22592	239	9	,	,	PUNCT
brj-22592	239	10	regnet+se	regnet+se	ADJ
brj-22592	239	11	network	network	NOUN
brj-22592	239	12	,	,	PUNCT
brj-22592	239	13	regnet+eca	regnet+eca	NOUN
brj-22592	239	14	network	network	NOUN
brj-22592	239	15	,	,	PUNCT
brj-22592	239	16	regnet+nam	regnet+nam	PROPN
brj-22592	239	17	network	network	NOUN
brj-22592	239	18	,	,	PUNCT
brj-22592	239	19	and	and	CCONJ
brj-22592	239	20	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	239	21	network	network	NOUN
brj-22592	239	22	model	model	NOUN
brj-22592	239	23	types	type	NOUN
brj-22592	239	24	of	of	ADP
brj-22592	239	25	defects	defect	NOUN
brj-22592	239	26	precision	precision	NOUN
brj-22592	239	27	recall	recall	VERB
brj-22592	239	28	specificity	specificity	NOUN
brj-22592	239	29	regnet	regnet	NOUN
brj-22592	239	30	wormhole	wormhole	NOUN
brj-22592	239	31	0.933	0.933	NUM
brj-22592	239	32	0.944	0.944	NUM
brj-22592	239	33	0.968	0.968	NUM
brj-22592	239	34	slip	slip	NOUN
brj-22592	239	35	knot	knot	NOUN
brj-22592	239	36	0.966	0.966	NUM
brj-22592	239	37	0.969	0.969	NUM
brj-22592	239	38	0.983	0.983	NUM
brj-22592	239	39	dead	dead	ADJ
brj-22592	239	40	knot	knot	NOUN
brj-22592	239	41	0.946	0.946	NUM
brj-22592	239	42	0.933	0.933	NUM
brj-22592	239	43	0.973	0.973	NUM
brj-22592	239	44	regnet+se	regnet+se	ADJ
brj-22592	239	45	wormhole	wormhole	NOUN
brj-22592	239	46	0.952	0.952	NUM
brj-22592	239	47	0.935	0.935	NUM
brj-22592	239	48	0.978	0.978	NUM
brj-22592	239	49	slip	slip	NOUN
brj-22592	239	50	knot	knot	NOUN
brj-22592	239	51	0.961	0.961	NUM
brj-22592	239	52	0.969	0.969	NUM
brj-22592	239	53	0.98	0.98	NUM
brj-22592	239	54	dead	dead	ADJ
brj-22592	239	55	knot	knot	NOUN
brj-22592	239	56	0.939	0.939	NUM
brj-22592	239	57	0.947	0.947	NUM
brj-22592	239	58	0.968	0.968	NUM
brj-22592	239	59	regnet+eca	regnet+eca	NOUN
brj-22592	239	60	wormhole	wormhole	NOUN
brj-22592	239	61	0.895	0.895	NUM
brj-22592	239	62	0.930	0.930	NUM
brj-22592	239	63	0.948	0.948	NUM
brj-22592	239	64	slip	slip	NOUN
brj-22592	239	65	knot	knot	NOUN
brj-22592	239	66	0.967	0.967	NUM
brj-22592	239	67	0.913	0.913	NUM
brj-22592	239	68	0.984	0.984	NUM
brj-22592	239	69	dead	dead	ADJ
brj-22592	239	70	knot	knot	NOUN
brj-22592	239	71	0.901	0.901	NUM
brj-22592	239	72	0.919	0.919	NUM
brj-22592	239	73	0.948	0.948	NUM
brj-22592	239	74	regnet+nam	regnet+nam	PROPN
brj-22592	239	75	wormhole	wormhole	NOUN
brj-22592	239	76	0.942	0.942	NUM
brj-22592	239	77	0.959	0.959	NUM
brj-22592	239	78	0.972	0.972	NUM
brj-22592	239	79	slip	slip	NOUN
brj-22592	239	80	knot	knot	NOUN
brj-22592	239	81	0.963	0.963	NUM
brj-22592	239	82	0.952	0.952	NUM
brj-22592	239	83	0.981	0.981	NUM
brj-22592	239	84	dead	dead	ADJ
brj-22592	239	85	knot	knot	NOUN
brj-22592	239	86	0.935	0.935	NUM
brj-22592	239	87	0.93	0.93	NUM
brj-22592	239	88	0.967	0.967	NUM
brj-22592	239	89	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	239	90	wormhole	wormhole	NOUN
brj-22592	239	91	0.954	0.954	NUM
brj-22592	239	92	0.965	0.965	NUM
brj-22592	239	93	0.978	0.978	NUM
brj-22592	239	94	slip	slip	NOUN
brj-22592	239	95	knot	knot	NOUN
brj-22592	239	96	0.986	0.986	NUM
brj-22592	239	97	0.969	0.969	NUM
brj-22592	239	98	0.993	0.993	NUM
brj-22592	239	99	dead	dead	ADJ
brj-22592	239	100	knot	knot	NOUN
brj-22592	239	101	0.958	0.958	NUM
brj-22592	239	102	0.963	0.963	NUM
brj-22592	239	103	0.979	0.979	NUM
brj-22592	239	104	table	table	NOUN
brj-22592	239	105	3	3	NUM
brj-22592	239	106	.	.	NOUN
brj-22592	239	107	comparison	comparison	NOUN
brj-22592	239	108	of	of	ADP
brj-22592	239	109	accuracy	accuracy	NOUN
brj-22592	239	110	,	,	PUNCT
brj-22592	239	111	network	network	NOUN
brj-22592	239	112	parameters	parameter	NOUN
brj-22592	239	113	,	,	PUNCT
brj-22592	239	114	and	and	CCONJ
brj-22592	239	115	average	average	ADJ
brj-22592	239	116	image	image	NOUN
brj-22592	239	117	recognition	recognition	NOUN
brj-22592	239	118	time	time	NOUN
brj-22592	239	119	of	of	ADP
brj-22592	239	120	regnet	regnet	NOUN
brj-22592	239	121	network	network	NOUN
brj-22592	239	122	,	,	PUNCT
brj-22592	239	123	regnet+se	regnet+se	ADJ
brj-22592	239	124	network	network	NOUN
brj-22592	239	125	,	,	PUNCT
brj-22592	239	126	regnet+eca	regnet+eca	NOUN
brj-22592	239	127	network	network	NOUN
brj-22592	239	128	,	,	PUNCT
brj-22592	239	129	regnet+nam	regnet+nam	PROPN
brj-22592	239	130	network	network	NOUN
brj-22592	239	131	,	,	PUNCT
brj-22592	239	132	and	and	CCONJ
brj-22592	239	133	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	239	134	network	network	NOUN
brj-22592	239	135	on	on	ADP
brj-22592	239	136	the	the	DET
brj-22592	239	137	test	test	NOUN
brj-22592	239	138	set	set	VERB
brj-22592	239	139	model	model	NOUN
brj-22592	239	140	accuracy	accuracy	NOUN
brj-22592	239	141	parameter	parameter	NOUN
brj-22592	239	142	infer	infer	VERB
brj-22592	239	143	regnet	regnet	NOUN
brj-22592	239	144	94.88	94.88	NUM
brj-22592	239	145	%	%	NOUN
brj-22592	239	146	2.32	2.32	NUM
brj-22592	239	147	m	m	NUM
brj-22592	239	148	0.046s	0.046	NOUN
brj-22592	239	149	regnet+se	regnet+se	NOUN
brj-22592	239	150	95.07	95.07	NUM
brj-22592	239	151	%	%	NOUN
brj-22592	239	152	2.8	2.8	NUM
brj-22592	239	153	m	m	NOUN
brj-22592	239	154	0.047s	0.047	NOUN
brj-22592	239	155	regnet+eca	regnet+eca	NOUN
brj-22592	239	156	92.03	92.03	NUM
brj-22592	239	157	%	%	NOUN
brj-22592	239	158	2.32	2.32	NUM
brj-22592	239	159	m	m	NOUN
brj-22592	239	160	0.043s	0.043s	NOUN
brj-22592	239	161	regnet+nam	regnet+nam	PROPN
brj-22592	239	162	94.69	94.69	NUM
brj-22592	239	163	%	%	NOUN
brj-22592	239	164	2.32	2.32	NUM
brj-22592	239	165	m	m	NOUN
brj-22592	239	166	0.036s	0.036s	NOUN
brj-22592	239	167	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	239	168	96.58	96.58	NUM
brj-22592	239	169	%	%	NOUN
brj-22592	239	170	1.86	1.86	NUM
brj-22592	239	171	m	m	NOUN
brj-22592	239	172	0.034s	0.034s	ADJ
brj-22592	239	173	table	table	NOUN
brj-22592	239	174	4	4	NUM
brj-22592	239	175	.	.	PUNCT
brj-22592	240	1	performance	performance	NOUN
brj-22592	240	2	comparison	comparison	NOUN
brj-22592	240	3	of	of	ADP
brj-22592	240	4	mobilenet	mobilenet	NOUN
brj-22592	240	5	-	-	PUNCT
brj-22592	240	6	v2	v2	NOUN
brj-22592	240	7	network	network	NOUN
brj-22592	240	8	,	,	PUNCT
brj-22592	240	9	efficientnet	efficientnet	NOUN
brj-22592	240	10	network	network	NOUN
brj-22592	240	11	,	,	PUNCT
brj-22592	240	12	vision	vision	NOUN
brj-22592	240	13	-	-	PUNCT
brj-22592	240	14	transformer	transformer	NOUN
brj-22592	240	15	network	network	NOUN
brj-22592	240	16	and	and	CCONJ
brj-22592	240	17	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	240	18	network	network	NOUN
brj-22592	240	19	model	model	NOUN
brj-22592	240	20	types	type	NOUN
brj-22592	240	21	of	of	ADP
brj-22592	240	22	defects	defect	NOUN
brj-22592	240	23	precision	precision	NOUN
brj-22592	240	24	recall	recall	NOUN
brj-22592	240	25	specificity	specificity	NOUN
brj-22592	240	26	mobilenet	mobilenet	NOUN
brj-22592	240	27	-	-	PUNCT
brj-22592	240	28	v2	v2	NOUN
brj-22592	240	29	wormhole	wormhole	NOUN
brj-22592	240	30	0.89	0.89	NUM
brj-22592	240	31	0.953	0.953	NUM
brj-22592	240	32	0.944	0.944	NUM
brj-22592	240	33	slip	slip	NOUN
brj-22592	240	34	knot	knot	NOUN
brj-22592	240	35	0.952	0.952	NUM
brj-22592	240	36	0.95	0.95	NUM
brj-22592	240	37	0.976	0.976	NUM
brj-22592	240	38	dead	dead	ADJ
brj-22592	240	39	knot	knot	NOUN
brj-22592	240	40	0.937	0.937	NUM
brj-22592	240	41	0.876	0.876	NUM
brj-22592	240	42	0.97	0.97	NUM
brj-22592	240	43	efficientnet	efficientnet	ADJ
brj-22592	240	44	wormhole	wormhole	NOUN
brj-22592	240	45	0.895	0.895	NUM
brj-22592	240	46	0.977	0.977	NUM
brj-22592	240	47	0.945	0.945	NUM
brj-22592	240	48	slip	slip	NOUN
brj-22592	240	49	knot	knot	NOUN
brj-22592	240	50	0.997	0.997	NUM
brj-22592	240	51	0.964	0.964	NUM
brj-22592	240	52	0.999	0.999	NUM
brj-22592	240	53	dead	dead	ADJ
brj-22592	240	54	knot	knot	NOUN
brj-22592	240	55	0.95	0.95	NUM
brj-22592	240	56	0.899	0.899	NUM
brj-22592	240	57	0.976	0.976	NUM
brj-22592	240	58	vision	vision	NOUN
brj-22592	240	59	-	-	PUNCT
brj-22592	240	60	transformer	transformer	NOUN
brj-22592	240	61	wormhole	wormhole	NOUN
brj-22592	240	62	0.944	0.944	NUM
brj-22592	240	63	0.935	0.935	NUM
brj-22592	240	64	0.973	0.973	NUM
brj-22592	240	65	slip	slip	NOUN
brj-22592	240	66	knot	knot	NOUN
brj-22592	240	67	0.977	0.977	NUM
brj-22592	240	68	0.969	0.969	NUM
brj-22592	240	69	0.989	0.989	NUM
brj-22592	240	70	dead	dead	ADJ
brj-22592	240	71	knot	knot	NOUN
brj-22592	240	72	0.934	0.934	NUM
brj-22592	240	73	0.949	0.949	NUM
brj-22592	240	74	0.966	0.966	NUM
brj-22592	240	75	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	240	76	wormhole	wormhole	NOUN
brj-22592	240	77	0.954	0.954	NUM
brj-22592	240	78	0.965	0.965	NUM
brj-22592	240	79	0.978	0.978	NUM
brj-22592	240	80	slip	slip	NOUN
brj-22592	240	81	knot	knot	NOUN
brj-22592	240	82	0.986	0.986	NUM
brj-22592	240	83	0.969	0.969	NUM
brj-22592	240	84	0.993	0.993	NUM
brj-22592	240	85	dead	dead	ADJ
brj-22592	240	86	knot	knot	NOUN
brj-22592	240	87	0.958	0.958	NUM
brj-22592	240	88	0.963	0.963	NUM
brj-22592	240	89	0.979	0.979	NUM
brj-22592	240	90	peer	peer	NOUN
brj-22592	240	91	-	-	PUNCT
brj-22592	240	92	reviewed	review	VERB
brj-22592	240	93	article	article	NOUN
brj-22592	240	94	bioresources.com	bioresources.com	X
brj-22592	240	95	xie	xie	PROPN
brj-22592	240	96	&	&	CCONJ
brj-22592	240	97	ling	ling	PROPN
brj-22592	240	98	(	(	PUNCT
brj-22592	240	99	2023	2023	NUM
brj-22592	240	100	)	)	PUNCT
brj-22592	240	101	.	.	PUNCT
brj-22592	241	1	“	"	PUNCT
brj-22592	241	2	wood	wood	NOUN
brj-22592	241	3	defect	defect	NOUN
brj-22592	241	4	classification	classification	NOUN
brj-22592	241	5	,	,	PUNCT
brj-22592	241	6	”	"	PUNCT
brj-22592	241	7	bioresources	bioresource	NOUN
brj-22592	241	8	18(4	18(4	NUM
brj-22592	241	9	)	)	PUNCT
brj-22592	241	10	,	,	PUNCT
brj-22592	241	11	7663	7663	NUM
brj-22592	241	12	-	-	SYM
brj-22592	241	13	7680	7680	NUM
brj-22592	241	14	.	.	PUNCT
brj-22592	241	15	7678	7678	NUM
brj-22592	241	16	to	to	PART
brj-22592	241	17	verify	verify	VERB
brj-22592	241	18	the	the	DET
brj-22592	241	19	performance	performance	NOUN
brj-22592	241	20	of	of	ADP
brj-22592	241	21	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	241	22	network	network	NOUN
brj-22592	241	23	model	model	NOUN
brj-22592	241	24	,	,	PUNCT
brj-22592	241	25	this	this	DET
brj-22592	241	26	study	study	NOUN
brj-22592	241	27	introduced	introduce	VERB
brj-22592	241	28	the	the	DET
brj-22592	241	29	comparative	comparative	ADJ
brj-22592	241	30	analysis	analysis	NOUN
brj-22592	241	31	of	of	ADP
brj-22592	241	32	mobilenet	mobilenet	NOUN
brj-22592	241	33	-	-	PUNCT
brj-22592	241	34	v2	v2	NOUN
brj-22592	241	35	,	,	PUNCT
brj-22592	241	36	efficientnet	efficientnet	NOUN
brj-22592	241	37	,	,	PUNCT
brj-22592	241	38	and	and	CCONJ
brj-22592	241	39	visiontransformer	visiontransformer	NOUN
brj-22592	241	40	networks	network	NOUN
brj-22592	241	41	,	,	PUNCT
brj-22592	241	42	as	as	SCONJ
brj-22592	241	43	shown	show	VERB
brj-22592	241	44	in	in	ADP
brj-22592	241	45	table	table	NOUN
brj-22592	241	46	4	4	NUM
brj-22592	241	47	.	.	PUNCT
brj-22592	242	1	according	accord	VERB
brj-22592	242	2	to	to	ADP
brj-22592	242	3	table	table	NOUN
brj-22592	242	4	4	4	NUM
brj-22592	242	5	,	,	PUNCT
brj-22592	242	6	the	the	DET
brj-22592	242	7	recall	recall	NOUN
brj-22592	242	8	rate	rate	NOUN
brj-22592	242	9	,	,	PUNCT
brj-22592	242	10	specificity	specificity	NOUN
brj-22592	242	11	,	,	PUNCT
brj-22592	242	12	and	and	CCONJ
brj-22592	242	13	false	false	ADJ
brj-22592	242	14	detection	detection	NOUN
brj-22592	242	15	probability	probability	NOUN
brj-22592	242	16	of	of	ADP
brj-22592	242	17	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	242	18	network	network	NOUN
brj-22592	242	19	were	be	AUX
brj-22592	242	20	still	still	ADV
brj-22592	242	21	higher	high	ADJ
brj-22592	242	22	than	than	ADP
brj-22592	242	23	those	those	PRON
brj-22592	242	24	of	of	ADP
brj-22592	242	25	mobilenet	mobilenet	NOUN
brj-22592	242	26	-	-	PUNCT
brj-22592	242	27	v2	v2	NOUN
brj-22592	242	28	,	,	PUNCT
brj-22592	242	29	efficientnet	efficientnet	NOUN
brj-22592	242	30	,	,	PUNCT
brj-22592	242	31	and	and	CCONJ
brj-22592	242	32	vision	vision	NOUN
brj-22592	242	33	-	-	PUNCT
brj-22592	242	34	transformer	transformer	NOUN
brj-22592	242	35	networks	network	NOUN
brj-22592	242	36	.	.	PUNCT
brj-22592	243	1	the	the	DET
brj-22592	243	2	recognition	recognition	NOUN
brj-22592	243	3	accuracy	accuracy	NOUN
brj-22592	243	4	,	,	PUNCT
brj-22592	243	5	network	network	NOUN
brj-22592	243	6	parameter	parameter	NOUN
brj-22592	243	7	quantity	quantity	NOUN
brj-22592	243	8	,	,	PUNCT
brj-22592	243	9	and	and	CCONJ
brj-22592	243	10	average	average	ADJ
brj-22592	243	11	image	image	NOUN
brj-22592	243	12	recognition	recognition	NOUN
brj-22592	243	13	time	time	NOUN
brj-22592	243	14	of	of	ADP
brj-22592	243	15	the	the	DET
brj-22592	243	16	mobilenet	mobilenet	NOUN
brj-22592	243	17	-	-	PUNCT
brj-22592	243	18	v2	v2	NOUN
brj-22592	243	19	,	,	PUNCT
brj-22592	243	20	efficientnet	efficientnet	NOUN
brj-22592	243	21	,	,	PUNCT
brj-22592	243	22	and	and	CCONJ
brj-22592	243	23	visiontransformer	visiontransformer	NOUN
brj-22592	243	24	networks	network	NOUN
brj-22592	243	25	are	be	AUX
brj-22592	243	26	presented	present	VERB
brj-22592	243	27	in	in	ADP
brj-22592	243	28	table	table	NOUN
brj-22592	243	29	5	5	NUM
brj-22592	243	30	.	.	PUNCT
brj-22592	243	31	table	table	NOUN
brj-22592	243	32	5	5	NUM
brj-22592	243	33	.	.	PUNCT
brj-22592	243	34	comparison	comparison	NOUN
brj-22592	243	35	of	of	ADP
brj-22592	243	36	accuracy	accuracy	NOUN
brj-22592	243	37	,	,	PUNCT
brj-22592	243	38	network	network	NOUN
brj-22592	243	39	parameters	parameter	NOUN
brj-22592	243	40	,	,	PUNCT
brj-22592	243	41	and	and	CCONJ
brj-22592	243	42	average	average	ADJ
brj-22592	243	43	image	image	NOUN
brj-22592	243	44	recognition	recognition	NOUN
brj-22592	243	45	time	time	NOUN
brj-22592	243	46	of	of	ADP
brj-22592	243	47	mobilenet	mobilenet	NOUN
brj-22592	243	48	-	-	PUNCT
brj-22592	243	49	v2	v2	NOUN
brj-22592	243	50	network	network	NOUN
brj-22592	243	51	,	,	PUNCT
brj-22592	243	52	efficientnet	efficientnet	NOUN
brj-22592	243	53	network	network	NOUN
brj-22592	243	54	,	,	PUNCT
brj-22592	243	55	visiontransformer	visiontransformer	NOUN
brj-22592	243	56	network	network	NOUN
brj-22592	243	57	and	and	CCONJ
brj-22592	243	58	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	243	59	network	network	NOUN
brj-22592	243	60	on	on	ADP
brj-22592	243	61	the	the	DET
brj-22592	243	62	test	test	NOUN
brj-22592	243	63	set	set	VERB
brj-22592	243	64	model	model	NOUN
brj-22592	243	65	accuracy	accuracy	NOUN
brj-22592	243	66	parameter	parameter	NOUN
brj-22592	243	67	infer	infer	VERB
brj-22592	243	68	mobilenet	mobilenet	NOUN
brj-22592	243	69	-	-	PUNCT
brj-22592	243	70	v2	v2	NOUN
brj-22592	243	71	92.60	92.60	NUM
brj-22592	243	72	%	%	NOUN
brj-22592	243	73	4.01	4.01	NUM
brj-22592	243	74	m	m	NOUN
brj-22592	243	75	0.070s	0.070	NOUN
brj-22592	243	76	efficientnet	efficientnet	NOUN
brj-22592	243	77	94.59	94.59	NUM
brj-22592	243	78	%	%	NOUN
brj-22592	243	79	2.23	2.23	NUM
brj-22592	243	80	m	m	NOUN
brj-22592	243	81	0.081s	0.081s	NOUN
brj-22592	243	82	vision	vision	NOUN
brj-22592	243	83	-	-	PUNCT
brj-22592	243	84	transformer	transformer	NOUN
brj-22592	243	85	95.16	95.16	NUM
brj-22592	243	86	%	%	NOUN
brj-22592	243	87	85.80	85.80	NUM
brj-22592	243	88	m	m	NOUN
brj-22592	243	89	0.081s	0.081s	NOUN
brj-22592	243	90	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	243	91	96.58	96.58	NUM
brj-22592	243	92	%	%	NOUN
brj-22592	243	93	1.86	1.86	NUM
brj-22592	243	94	m	m	NOUN
brj-22592	243	95	0.034s	0.034s	NOUN
brj-22592	243	96	compared	compare	VERB
brj-22592	243	97	with	with	ADP
brj-22592	243	98	the	the	DET
brj-22592	243	99	other	other	ADJ
brj-22592	243	100	two	two	NUM
brj-22592	243	101	networks	network	NOUN
brj-22592	243	102	,	,	PUNCT
brj-22592	243	103	the	the	DET
brj-22592	243	104	accuracy	accuracy	NOUN
brj-22592	243	105	of	of	ADP
brj-22592	243	106	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	243	107	network	network	NOUN
brj-22592	243	108	was	be	AUX
brj-22592	243	109	as	as	ADV
brj-22592	243	110	high	high	ADJ
brj-22592	243	111	as	as	ADP
brj-22592	243	112	96.58	96.58	NUM
brj-22592	243	113	%	%	NOUN
brj-22592	243	114	,	,	PUNCT
brj-22592	243	115	the	the	DET
brj-22592	243	116	amount	amount	NOUN
brj-22592	243	117	of	of	ADP
brj-22592	243	118	network	network	NOUN
brj-22592	243	119	parameters	parameter	NOUN
brj-22592	243	120	was	be	AUX
brj-22592	243	121	only	only	ADV
brj-22592	243	122	1.86	1.86	NUM
brj-22592	243	123	m	m	NOUN
brj-22592	243	124	,	,	PUNCT
brj-22592	243	125	and	and	CCONJ
brj-22592	243	126	the	the	DET
brj-22592	243	127	average	average	ADJ
brj-22592	243	128	image	image	NOUN
brj-22592	243	129	recognition	recognition	NOUN
brj-22592	243	130	time	time	NOUN
brj-22592	243	131	was	be	AUX
brj-22592	243	132	0.034	0.034	NUM
brj-22592	243	133	s.	s.	PROPN
brj-22592	243	134	while	while	SCONJ
brj-22592	243	135	reducing	reduce	VERB
brj-22592	243	136	the	the	DET
brj-22592	243	137	model	model	NOUN
brj-22592	243	138	weight	weight	NOUN
brj-22592	243	139	,	,	PUNCT
brj-22592	243	140	the	the	DET
brj-22592	243	141	detection	detection	NOUN
brj-22592	243	142	speed	speed	NOUN
brj-22592	243	143	and	and	CCONJ
brj-22592	243	144	accuracy	accuracy	NOUN
brj-22592	243	145	are	be	AUX
brj-22592	243	146	improved	improve	VERB
brj-22592	243	147	.	.	PUNCT
brj-22592	244	1	from	from	ADP
brj-22592	244	2	table	table	NOUN
brj-22592	244	3	5	5	NUM
brj-22592	244	4	,	,	PUNCT
brj-22592	244	5	it	it	PRON
brj-22592	244	6	can	can	AUX
brj-22592	244	7	be	be	AUX
brj-22592	244	8	seen	see	VERB
brj-22592	244	9	that	that	SCONJ
brj-22592	244	10	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	244	11	network	network	NOUN
brj-22592	244	12	achieved	achieve	VERB
brj-22592	244	13	higher	high	ADJ
brj-22592	244	14	accuracy	accuracy	NOUN
brj-22592	244	15	and	and	CCONJ
brj-22592	244	16	could	could	AUX
brj-22592	244	17	accurately	accurately	ADV
brj-22592	244	18	classify	classify	VERB
brj-22592	244	19	plate	plate	NOUN
brj-22592	244	20	defect	defect	NOUN
brj-22592	244	21	models	model	NOUN
brj-22592	244	22	.	.	PUNCT
brj-22592	245	1	at	at	ADP
brj-22592	245	2	the	the	DET
brj-22592	245	3	same	same	ADJ
brj-22592	245	4	time	time	NOUN
brj-22592	245	5	,	,	PUNCT
brj-22592	245	6	the	the	DET
brj-22592	245	7	classification	classification	NOUN
brj-22592	245	8	time	time	NOUN
brj-22592	245	9	was	be	AUX
brj-22592	245	10	relatively	relatively	ADV
brj-22592	245	11	short	short	ADJ
brj-22592	245	12	.	.	PUNCT
brj-22592	246	1	wood	wood	NOUN
brj-22592	246	2	defect	defect	NOUN
brj-22592	246	3	detection	detection	NOUN
brj-22592	246	4	is	be	AUX
brj-22592	246	5	often	often	ADV
brj-22592	246	6	used	use	VERB
brj-22592	246	7	for	for	ADP
brj-22592	246	8	industrial	industrial	ADJ
brj-22592	246	9	optimizing	optimizing	NOUN
brj-22592	246	10	saw	see	VERB
brj-22592	246	11	for	for	ADP
brj-22592	246	12	online	online	ADJ
brj-22592	246	13	sorting	sorting	NOUN
brj-22592	246	14	of	of	ADP
brj-22592	246	15	woods	wood	NOUN
brj-22592	246	16	.	.	PUNCT
brj-22592	247	1	modern	modern	ADJ
brj-22592	247	2	industrial	industrial	ADJ
brj-22592	247	3	production	production	NOUN
brj-22592	247	4	lines	line	NOUN
brj-22592	247	5	have	have	VERB
brj-22592	247	6	track	track	NOUN
brj-22592	247	7	feed	feed	NOUN
brj-22592	247	8	speeds	speed	NOUN
brj-22592	247	9	of	of	ADP
brj-22592	247	10	up	up	ADP
brj-22592	247	11	to	to	PART
brj-22592	247	12	120m	120m	NUM
brj-22592	247	13	/	/	SYM
brj-22592	247	14	min	min	NOUN
brj-22592	247	15	,	,	PUNCT
brj-22592	247	16	and	and	CCONJ
brj-22592	247	17	it	it	PRON
brj-22592	247	18	is	be	AUX
brj-22592	247	19	very	very	ADV
brj-22592	247	20	important	important	ADJ
brj-22592	247	21	to	to	PART
brj-22592	247	22	improve	improve	VERB
brj-22592	247	23	the	the	DET
brj-22592	247	24	speed	speed	NOUN
brj-22592	247	25	even	even	ADV
brj-22592	247	26	at	at	ADP
brj-22592	247	27	millisecond	millisecond	NOUN
brj-22592	247	28	level	level	NOUN
brj-22592	247	29	.	.	PUNCT
brj-22592	248	1	meanwhile	meanwhile	ADV
brj-22592	248	2	,	,	PUNCT
brj-22592	248	3	the	the	DET
brj-22592	248	4	params	param	NOUN
brj-22592	248	5	of	of	ADP
brj-22592	248	6	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	248	7	network	network	NOUN
brj-22592	248	8	are	be	AUX
brj-22592	248	9	minimal	minimal	ADJ
brj-22592	248	10	,	,	PUNCT
brj-22592	248	11	which	which	PRON
brj-22592	248	12	can	can	AUX
brj-22592	248	13	alleviate	alleviate	VERB
brj-22592	248	14	computational	computational	ADJ
brj-22592	248	15	pressure	pressure	NOUN
brj-22592	248	16	for	for	ADP
brj-22592	248	17	computational	computational	ADJ
brj-22592	248	18	power	power	NOUN
brj-22592	248	19	.	.	PUNCT
brj-22592	249	1	conclusions	conclusion	NOUN
brj-22592	249	2	1	1	X
brj-22592	249	3	.	.	PUNCT
brj-22592	250	1	the	the	DET
brj-22592	250	2	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	250	3	network	network	NOUN
brj-22592	250	4	model	model	NOUN
brj-22592	250	5	can	can	AUX
brj-22592	250	6	distinguish	distinguish	VERB
brj-22592	250	7	the	the	DET
brj-22592	250	8	types	type	NOUN
brj-22592	250	9	of	of	ADP
brj-22592	250	10	wood	wood	NOUN
brj-22592	250	11	defects	defect	NOUN
brj-22592	250	12	,	,	PUNCT
brj-22592	250	13	and	and	CCONJ
brj-22592	250	14	the	the	DET
brj-22592	250	15	classification	classification	NOUN
brj-22592	250	16	accuracy	accuracy	NOUN
brj-22592	250	17	is	be	AUX
brj-22592	250	18	high	high	ADJ
brj-22592	250	19	.	.	PUNCT
brj-22592	251	1	2	2	X
brj-22592	251	2	.	.	PUNCT
brj-22592	251	3	to	to	PART
brj-22592	251	4	verify	verify	VERB
brj-22592	251	5	the	the	DET
brj-22592	251	6	performance	performance	NOUN
brj-22592	251	7	of	of	ADP
brj-22592	251	8	the	the	DET
brj-22592	251	9	model	model	NOUN
brj-22592	251	10	,	,	PUNCT
brj-22592	251	11	mobilenet	mobilenet	NOUN
brj-22592	251	12	-	-	PUNCT
brj-22592	251	13	v2	v2	NOUN
brj-22592	251	14	,	,	PUNCT
brj-22592	251	15	efficientnet	efficientnet	NOUN
brj-22592	251	16	,	,	PUNCT
brj-22592	251	17	and	and	CCONJ
brj-22592	251	18	visiontransformer	visiontransformer	NOUN
brj-22592	251	19	networks	network	NOUN
brj-22592	251	20	are	be	AUX
brj-22592	251	21	introduced	introduce	VERB
brj-22592	251	22	for	for	ADP
brj-22592	251	23	comparison	comparison	NOUN
brj-22592	251	24	.	.	PUNCT
brj-22592	252	1	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	252	2	model	model	NOUN
brj-22592	252	3	is	be	AUX
brj-22592	252	4	still	still	ADV
brj-22592	252	5	better	well	ADJ
brj-22592	252	6	than	than	ADP
brj-22592	252	7	the	the	DET
brj-22592	252	8	comparison	comparison	NOUN
brj-22592	252	9	networks	network	NOUN
brj-22592	252	10	with	with	ADP
brj-22592	252	11	less	less	ADJ
brj-22592	252	12	parameters	parameter	NOUN
brj-22592	252	13	.	.	PUNCT
brj-22592	253	1	3	3	X
brj-22592	253	2	.	.	X
brj-22592	253	3	the	the	DET
brj-22592	253	4	ghostregnet+nam	ghostregnet+nam	PROPN
brj-22592	253	5	network	network	NOUN
brj-22592	253	6	model	model	NOUN
brj-22592	253	7	in	in	ADP
brj-22592	253	8	distinguishing	distinguish	VERB
brj-22592	253	9	the	the	DET
brj-22592	253	10	types	type	NOUN
brj-22592	253	11	of	of	ADP
brj-22592	253	12	sample	sample	NOUN
brj-22592	253	13	defects	defect	NOUN
brj-22592	253	14	with	with	ADP
brj-22592	253	15	high	high	ADJ
brj-22592	253	16	accuracy	accuracy	NOUN
brj-22592	253	17	and	and	CCONJ
brj-22592	253	18	fewer	few	ADJ
brj-22592	253	19	network	network	NOUN
brj-22592	253	20	parameters	parameter	NOUN
brj-22592	253	21	is	be	AUX
brj-22592	253	22	determined	determine	VERB
brj-22592	253	23	.	.	PUNCT
brj-22592	254	1	this	this	PRON
brj-22592	254	2	can	can	AUX
brj-22592	254	3	effectively	effectively	ADV
brj-22592	254	4	reduce	reduce	VERB
brj-22592	254	5	the	the	DET
brj-22592	254	6	detection	detection	NOUN
brj-22592	254	7	time	time	NOUN
brj-22592	254	8	and	and	CCONJ
brj-22592	254	9	reduce	reduce	VERB
brj-22592	254	10	the	the	DET
brj-22592	254	11	hardware	hardware	NOUN
brj-22592	254	12	requirements	requirement	NOUN
brj-22592	254	13	of	of	ADP
brj-22592	254	14	the	the	DET
brj-22592	254	15	algorithm	algorithm	NOUN
brj-22592	254	16	,	,	PUNCT
brj-22592	254	17	which	which	PRON
brj-22592	254	18	is	be	AUX
brj-22592	254	19	robust	robust	ADJ
brj-22592	254	20	and	and	CCONJ
brj-22592	254	21	practical	practical	ADJ
brj-22592	254	22	.	.	PUNCT
brj-22592	255	1	4	4	X
brj-22592	255	2	.	.	X
brj-22592	255	3	a	a	DET
brj-22592	255	4	training	training	NOUN
brj-22592	255	5	overfitting	overfitte	VERB
brj-22592	255	6	phenomenon	phenomenon	NOUN
brj-22592	255	7	was	be	AUX
brj-22592	255	8	encountered	encounter	VERB
brj-22592	255	9	.	.	PUNCT
brj-22592	256	1	by	by	ADP
brj-22592	256	2	adding	add	VERB
brj-22592	256	3	a	a	DET
brj-22592	256	4	regularization	regularization	NOUN
brj-22592	256	5	term	term	NOUN
brj-22592	256	6	in	in	ADP
brj-22592	256	7	the	the	DET
brj-22592	256	8	process	process	NOUN
brj-22592	256	9	of	of	ADP
brj-22592	256	10	calculating	calculate	VERB
brj-22592	256	11	the	the	DET
brj-22592	256	12	loss	loss	NOUN
brj-22592	256	13	function	function	NOUN
brj-22592	256	14	of	of	ADP
brj-22592	256	15	the	the	DET
brj-22592	256	16	attention	attention	NOUN
brj-22592	256	17	mechanism	mechanism	NOUN
brj-22592	256	18	nam	nam	NOUN
brj-22592	256	19	,	,	PUNCT
brj-22592	256	20	this	this	DET
brj-22592	256	21	problem	problem	NOUN
brj-22592	256	22	can	can	AUX
brj-22592	256	23	be	be	AUX
brj-22592	256	24	effectively	effectively	ADV
brj-22592	256	25	solved	solve	VERB
brj-22592	256	26	.	.	PUNCT
brj-22592	257	1	however	however	ADV
brj-22592	257	2	,	,	PUNCT
brj-22592	257	3	during	during	ADP
brj-22592	257	4	the	the	DET
brj-22592	257	5	learning	learning	NOUN
brj-22592	257	6	process	process	NOUN
brj-22592	257	7	of	of	ADP
brj-22592	257	8	the	the	DET
brj-22592	257	9	training	training	NOUN
brj-22592	257	10	set	set	NOUN
brj-22592	257	11	,	,	PUNCT
brj-22592	257	12	the	the	DET
brj-22592	257	13	loss	loss	NOUN
brj-22592	257	14	function	function	NOUN
brj-22592	257	15	fluctuated	fluctuate	VERB
brj-22592	257	16	slightly	slightly	ADV
brj-22592	257	17	.	.	PUNCT
brj-22592	258	1	peer	peer	NOUN
brj-22592	258	2	-	-	PUNCT
brj-22592	258	3	reviewed	review	VERB
brj-22592	258	4	article	article	NOUN
brj-22592	258	5	bioresources.com	bioresources.com	X
brj-22592	258	6	xie	xie	PROPN
brj-22592	258	7	&	&	CCONJ
brj-22592	258	8	ling	ling	PROPN
brj-22592	258	9	(	(	PUNCT
brj-22592	258	10	2023	2023	NUM
brj-22592	258	11	)	)	PUNCT
brj-22592	258	12	.	.	PUNCT
brj-22592	259	1	“	"	PUNCT
brj-22592	259	2	wood	wood	NOUN
brj-22592	259	3	defect	defect	NOUN
brj-22592	259	4	classification	classification	NOUN
brj-22592	259	5	,	,	PUNCT
brj-22592	259	6	”	"	PUNCT
brj-22592	259	7	bioresources	bioresource	NOUN
brj-22592	259	8	18(4	18(4	NUM
brj-22592	259	9	)	)	PUNCT
brj-22592	259	10	,	,	PUNCT
brj-22592	259	11	7663	7663	NUM
brj-22592	259	12	-	-	SYM
brj-22592	259	13	7680	7680	NUM
brj-22592	259	14	.	.	PUNCT
brj-22592	260	1	7679	7679	NUM
brj-22592	261	1	acknowledgements	acknowledgement	NOUN
brj-22592	261	2	this	this	DET
brj-22592	261	3	study	study	NOUN
brj-22592	261	4	is	be	AUX
brj-22592	261	5	funded	fund	VERB
brj-22592	261	6	by	by	ADP
brj-22592	261	7	a	a	DET
brj-22592	261	8	program	program	NOUN
brj-22592	261	9	of	of	ADP
brj-22592	261	10	heilongjiang	heilongjiang	PROPN
brj-22592	261	11	province	province	PROPN
brj-22592	261	12	postdoctoral	postdoctoral	PROPN
brj-22592	261	13	exit	exit	NOUN
brj-22592	261	14	research	research	NOUN
brj-22592	261	15	initiation	initiation	NOUN
brj-22592	261	16	fund	fund	NOUN
brj-22592	261	17	funded	fund	VERB
brj-22592	261	18	project	project	NOUN
brj-22592	261	19	(	(	PUNCT
brj-22592	261	20	2022/415858	2022/415858	NUM
brj-22592	261	21	)	)	PUNCT
brj-22592	261	22	and	and	CCONJ
brj-22592	261	23	a	a	DET
brj-22592	261	24	program	program	NOUN
brj-22592	261	25	of	of	ADP
brj-22592	261	26	the	the	DET
brj-22592	261	27	natural	natural	ADJ
brj-22592	261	28	science	science	NOUN
brj-22592	261	29	foundation	foundation	NOUN
brj-22592	261	30	of	of	ADP
brj-22592	261	31	heilongjiang	heilongjiang	PROPN
brj-22592	261	32	province	province	PROPN
brj-22592	261	33	(	(	PUNCT
brj-22592	261	34	lh2020c034	lh2020c034	NOUN
brj-22592	261	35	)	)	PUNCT
brj-22592	261	36	.	.	PUNCT
brj-22592	262	1	conflicts	conflict	NOUN
brj-22592	262	2	of	of	ADP
brj-22592	262	3	interest	interest	NOUN
brj-22592	262	4	the	the	DET
brj-22592	262	5	author	author	NOUN
brj-22592	262	6	declares	declare	VERB
brj-22592	262	7	that	that	SCONJ
brj-22592	262	8	there	there	PRON
brj-22592	262	9	is	be	VERB
brj-22592	262	10	no	no	DET
brj-22592	262	11	competing	compete	VERB
brj-22592	262	12	interest	interest	NOUN
brj-22592	262	13	.	.	PUNCT
brj-22592	263	1	references	reference	NOUN
brj-22592	263	2	cited	cite	VERB
brj-22592	263	3	diao	diao	PROPN
brj-22592	263	4	,	,	PUNCT
brj-22592	263	5	z.	z.	PROPN
brj-22592	263	6	d.	d.	PROPN
brj-22592	263	7	,	,	PUNCT
brj-22592	263	8	zhang	zhang	PROPN
brj-22592	263	9	,	,	PUNCT
brj-22592	263	10	d.	d.	PROPN
brj-22592	263	11	y.	y.	PROPN
brj-22592	263	12	,	,	PUNCT
brj-22592	263	13	and	and	CCONJ
brj-22592	263	14	cao	cao	PROPN
brj-22592	263	15	,	,	PUNCT
brj-22592	263	16	j.	j.	PROPN
brj-22592	263	17	(	(	PUNCT
brj-22592	263	18	2012	2012	NUM
brj-22592	263	19	)	)	PUNCT
brj-22592	263	20	.	.	PUNCT
brj-22592	264	1	“	"	PUNCT
brj-22592	264	2	research	research	NOUN
brj-22592	264	3	on	on	ADP
brj-22592	264	4	wood	wood	NOUN
brj-22592	264	5	sorting	sort	VERB
brj-22592	264	6	image	image	NOUN
brj-22592	264	7	segmentation	segmentation	NOUN
brj-22592	264	8	based	base	VERB
brj-22592	264	9	on	on	ADP
brj-22592	264	10	genetic	genetic	ADJ
brj-22592	264	11	algorithms	algorithm	NOUN
brj-22592	264	12	and	and	CCONJ
brj-22592	264	13	mathematical	mathematical	ADJ
brj-22592	264	14	morphology	morphology	NOUN
brj-22592	264	15	,	,	PUNCT
brj-22592	264	16	”	"	PUNCT
brj-22592	264	17	automation	automation	NOUN
brj-22592	264	18	technology	technology	NOUN
brj-22592	264	19	and	and	CCONJ
brj-22592	264	20	applications	application	NOUN
brj-22592	264	21	31(01	31(01	NUM
brj-22592	264	22	)	)	PUNCT
brj-22592	264	23	,	,	PUNCT
brj-22592	264	24	59	59	NUM
brj-22592	264	25	-	-	SYM
brj-22592	264	26	62	62	NUM
brj-22592	264	27	.	.	PUNCT
brj-22592	265	1	doi	doi	NOUN
brj-22592	265	2	:	:	PUNCT
brj-22592	265	3	10.3969	10.3969	NUM
brj-22592	265	4	/	/	SYM
brj-22592	265	5	j.issn.10037241.2012.01.016	j.issn.10037241.2012.01.016	PROPN
brj-22592	265	6	.	.	PUNCT
brj-22592	266	1	gong	gong	PROPN
brj-22592	266	2	,	,	PUNCT
brj-22592	266	3	x.	x.	PROPN
brj-22592	266	4	y.	y.	PROPN
brj-22592	266	5	,	,	PUNCT
brj-22592	266	6	chang	chang	PROPN
brj-22592	266	7	,	,	PUNCT
brj-22592	266	8	s.	s.	PROPN
brj-22592	266	9	y.	y.	PROPN
brj-22592	266	10	,	,	PUNCT
brj-22592	266	11	jiang	jiang	PROPN
brj-22592	266	12	,	,	PUNCT
brj-22592	266	13	y.f	y.f	PROPN
brj-22592	266	14	.	.	PROPN
brj-22592	266	15	,	,	PUNCT
brj-22592	266	16	and	and	CCONJ
brj-22592	266	17	wang	wang	PROPN
brj-22592	266	18	,	,	PUNCT
brj-22592	266	19	z.	z.	PROPN
brj-22592	266	20	y.	y.	PROPN
brj-22592	266	21	(	(	PUNCT
brj-22592	266	22	2019	2019	NUM
brj-22592	266	23	)	)	PUNCT
brj-22592	266	24	.	.	PUNCT
brj-22592	267	1	“	"	PUNCT
brj-22592	267	2	autogan	autogan	ADJ
brj-22592	267	3	:	:	PUNCT
brj-22592	267	4	neural	neural	ADJ
brj-22592	267	5	architecture	architecture	NOUN
brj-22592	267	6	search	search	NOUN
brj-22592	267	7	for	for	ADP
brj-22592	267	8	generative	generative	ADJ
brj-22592	267	9	adversarial	adversarial	ADJ
brj-22592	267	10	networks	network	NOUN
brj-22592	267	11	,	,	PUNCT
brj-22592	267	12	”	"	PUNCT
brj-22592	267	13	doi	doi	NOUN
brj-22592	267	14	:	:	PUNCT
brj-22592	267	15	10.48550	10.48550	NUM
brj-22592	267	16	/	/	SYM
brj-22592	267	17	arxiv.1908.03835	arxiv.1908.03835	NOUN
brj-22592	267	18	guo	guo	PROPN
brj-22592	267	19	,	,	PUNCT
brj-22592	267	20	k.	k.	PROPN
brj-22592	267	21	,	,	PUNCT
brj-22592	267	22	huang	huang	PROPN
brj-22592	267	23	,	,	PUNCT
brj-22592	267	24	y.	y.	PROPN
brj-22592	267	25	,	,	PUNCT
brj-22592	267	26	yang	yang	PROPN
brj-22592	267	27	,	,	PUNCT
brj-22592	267	28	n.	n.	PROPN
brj-22592	267	29	,	,	PUNCT
brj-22592	267	30	tan	tan	PROPN
brj-22592	267	31	,	,	PUNCT
brj-22592	267	32	g.	g.	PROPN
brj-22592	267	33	,	,	PUNCT
brj-22592	267	34	dong	dong	PROPN
brj-22592	267	35	,	,	PUNCT
brj-22592	267	36	q.	q.	PROPN
brj-22592	267	37	q.	q.	PROPN
brj-22592	267	38	,	,	PUNCT
brj-22592	267	39	and	and	CCONJ
brj-22592	267	40	cheng	cheng	PROPN
brj-22592	267	41	,	,	PUNCT
brj-22592	267	42	y.	y.	PROPN
brj-22592	267	43	z.	z.	PROPN
brj-22592	267	44	(	(	PUNCT
brj-22592	267	45	2020	2020	NUM
brj-22592	267	46	)	)	PUNCT
brj-22592	267	47	.	.	PUNCT
brj-22592	268	1	“	"	PUNCT
brj-22592	268	2	multichannel	multichannel	ADJ
brj-22592	268	3	wood	wood	NOUN
brj-22592	268	4	defect	defect	NOUN
brj-22592	268	5	image	image	NOUN
brj-22592	268	6	segmentation	segmentation	NOUN
brj-22592	268	7	algorithm	algorithm	NOUN
brj-22592	268	8	based	base	VERB
brj-22592	268	9	on	on	ADP
brj-22592	268	10	tvcv	tvcv	NOUN
brj-22592	268	11	model	model	NOUN
brj-22592	268	12	,	,	PUNCT
brj-22592	268	13	”	"	PUNCT
brj-22592	268	14	forestry	forestry	NOUN
brj-22592	268	15	machinery	machinery	NOUN
brj-22592	268	16	and	and	CCONJ
brj-22592	268	17	woodworking	woodworke	VERB
brj-22592	268	18	equipment	equipment	NOUN
brj-22592	268	19	48	48	NUM
brj-22592	268	20	,	,	PUNCT
brj-22592	268	21	22	22	NUM
brj-22592	268	22	-	-	SYM
brj-22592	268	23	26	26	NUM
brj-22592	268	24	.	.	PUNCT
brj-22592	269	1	doi	doi	NOUN
brj-22592	269	2	:	:	PUNCT
brj-22592	269	3	10.13279	10.13279	NUM
brj-22592	269	4	/	/	SYM
brj-22592	269	5	j.cnki.fmwe.2020.0100	j.cnki.fmwe.2020.0100	PROPN
brj-22592	269	6	han	han	PROPN
brj-22592	269	7	,	,	PUNCT
brj-22592	269	8	k.	k.	PROPN
brj-22592	269	9	,	,	PUNCT
brj-22592	269	10	wang	wang	PROPN
brj-22592	269	11	,	,	PUNCT
brj-22592	269	12	y.	y.	PROPN
brj-22592	269	13	,	,	PUNCT
brj-22592	269	14	tian	tian	PROPN
brj-22592	269	15	,	,	PUNCT
brj-22592	269	16	q.	q.	PROPN
brj-22592	269	17	,	,	PUNCT
brj-22592	269	18	guo	guo	PROPN
brj-22592	269	19	,	,	PUNCT
brj-22592	269	20	j.	j.	PROPN
brj-22592	269	21	,	,	PUNCT
brj-22592	269	22	xu	xu	PROPN
brj-22592	269	23	,	,	PUNCT
brj-22592	269	24	c.	c.	PROPN
brj-22592	269	25	,	,	PUNCT
brj-22592	269	26	and	and	CCONJ
brj-22592	269	27	xu	xu	PROPN
brj-22592	269	28	,	,	PUNCT
brj-22592	269	29	c.	c.	PROPN
brj-22592	269	30	(	(	PUNCT
brj-22592	269	31	2020	2020	NUM
brj-22592	269	32	)	)	PUNCT
brj-22592	269	33	.	.	PUNCT
brj-22592	270	1	“	"	PUNCT
brj-22592	270	2	ghostnet	ghostnet	NOUN
brj-22592	270	3	:	:	PUNCT
brj-22592	270	4	more	more	ADJ
brj-22592	270	5	features	feature	NOUN
brj-22592	270	6	from	from	ADP
brj-22592	270	7	cheap	cheap	ADJ
brj-22592	270	8	operations	operation	NOUN
brj-22592	270	9	,	,	PUNCT
brj-22592	270	10	”	"	PUNCT
brj-22592	270	11	proceedings	proceeding	NOUN
brj-22592	270	12	of	of	ADP
brj-22592	270	13	the	the	DET
brj-22592	270	14	ieee	ieee	NOUN
brj-22592	270	15	computer	computer	NOUN
brj-22592	270	16	society	society	PROPN
brj-22592	270	17	conference	conference	NOUN
brj-22592	270	18	on	on	ADP
brj-22592	270	19	computer	computer	NOUN
brj-22592	270	20	vision	vision	NOUN
brj-22592	270	21	and	and	CCONJ
brj-22592	270	22	pattern	pattern	NOUN
brj-22592	270	23	recognition	recognition	NOUN
brj-22592	270	24	.	.	PUNCT
brj-22592	271	1	doi	doi	NOUN
brj-22592	271	2	:	:	PUNCT
brj-22592	271	3	10.1109	10.1109	NUM
brj-22592	271	4	/	/	SYM
brj-22592	271	5	cvpr42600.2020.00165	cvpr42600.2020.00165	PROPN
brj-22592	271	6	hu	hu	PROPN
brj-22592	271	7	,	,	PUNCT
brj-22592	271	8	y.	y.	PROPN
brj-22592	271	9	f.	f.	PROPN
brj-22592	271	10	,	,	PUNCT
brj-22592	271	11	belkhir	belkhir	ADJ
brj-22592	271	12	,	,	PUNCT
brj-22592	271	13	n.	n.	NOUN
brj-22592	271	14	,	,	PUNCT
brj-22592	271	15	angulo	angulo	PROPN
brj-22592	271	16	,	,	PUNCT
brj-22592	271	17	j.	j.	PROPN
brj-22592	271	18	,	,	PUNCT
brj-22592	271	19	yao	yao	PROPN
brj-22592	271	20	,	,	PUNCT
brj-22592	271	21	a.	a.	NOUN
brj-22592	271	22	,	,	PUNCT
brj-22592	271	23	and	and	CCONJ
brj-22592	271	24	franchi	franchi	PROPN
brj-22592	271	25	,	,	PUNCT
brj-22592	271	26	g.	g.	PROPN
brj-22592	271	27	(	(	PUNCT
brj-22592	271	28	2022	2022	NUM
brj-22592	271	29	)	)	PUNCT
brj-22592	271	30	.	.	PUNCT
brj-22592	272	1	“	"	PUNCT
brj-22592	272	2	learning	learn	VERB
brj-22592	272	3	deep	deep	ADJ
brj-22592	272	4	morphological	morphological	ADJ
brj-22592	272	5	networks	network	NOUN
brj-22592	272	6	with	with	ADP
brj-22592	272	7	neural	neural	ADJ
brj-22592	272	8	architecture	architecture	NOUN
brj-22592	272	9	search	search	NOUN
brj-22592	272	10	,	,	PUNCT
brj-22592	272	11	”	"	PUNCT
brj-22592	272	12	pattern	pattern	NOUN
brj-22592	272	13	recognition	recognition	NOUN
brj-22592	272	14	131	131	NUM
brj-22592	272	15	.	.	PUNCT
brj-22592	273	1	doi	doi	NOUN
brj-22592	273	2	:	:	PUNCT
brj-22592	273	3	10.48550	10.48550	NUM
brj-22592	273	4	/	/	SYM
brj-22592	273	5	arxiv.2106.07714	arxiv.2106.07714	ADP
brj-22592	273	6	jin	jin	NOUN
brj-22592	273	7	,	,	PUNCT
brj-22592	273	8	h.	h.	PROPN
brj-22592	273	9	f.	f.	PROPN
brj-22592	273	10	,	,	PUNCT
brj-22592	273	11	song	song	PROPN
brj-22592	273	12	,	,	PUNCT
brj-22592	273	13	q.	q.	PROPN
brj-22592	273	14	q.	q.	PROPN
brj-22592	273	15	,	,	PUNCT
brj-22592	273	16	and	and	CCONJ
brj-22592	273	17	hu	hu	PROPN
brj-22592	273	18	,	,	PUNCT
brj-22592	273	19	x.	x.	NOUN
brj-22592	273	20	(	(	PUNCT
brj-22592	273	21	2019	2019	NUM
brj-22592	273	22	)	)	PUNCT
brj-22592	273	23	.	.	PUNCT
brj-22592	274	1	“	"	PUNCT
brj-22592	274	2	auto	auto	NOUN
brj-22592	274	3	-	-	PUNCT
brj-22592	274	4	keras	keras	PROPN
brj-22592	274	5	:	:	PUNCT
brj-22592	274	6	an	an	DET
brj-22592	274	7	efficient	efficient	ADJ
brj-22592	274	8	neural	neural	ADJ
brj-22592	274	9	architecture	architecture	NOUN
brj-22592	274	10	search	search	NOUN
brj-22592	274	11	system	system	NOUN
brj-22592	274	12	,	,	PUNCT
brj-22592	274	13	”	"	PUNCT
brj-22592	274	14	knowledge	knowledge	NOUN
brj-22592	274	15	discovery	discovery	PROPN
brj-22592	274	16	&	&	CCONJ
brj-22592	274	17	data	data	PROPN
brj-22592	274	18	mining	mining	NOUN
brj-22592	274	19	.	.	PUNCT
brj-22592	275	1	doi	doi	NOUN
brj-22592	275	2	:	:	PUNCT
brj-22592	275	3	10.48550	10.48550	NUM
brj-22592	275	4	/	/	SYM
brj-22592	275	5	arxiv.1806.10282	arxiv.1806.10282	PROPN
brj-22592	275	6	li	li	PROPN
brj-22592	275	7	,	,	PUNCT
brj-22592	275	8	g.	g.	PROPN
brj-22592	275	9	,	,	PUNCT
brj-22592	275	10	qian	qian	PROPN
brj-22592	275	11	,	,	PUNCT
brj-22592	275	12	g.	g.	PROPN
brj-22592	275	13	,	,	PUNCT
brj-22592	275	14	delgadillo	delgadillo	PROPN
brj-22592	275	15	,	,	PUNCT
brj-22592	275	16	i.	i.	PROPN
brj-22592	275	17	c.	c.	PROPN
brj-22592	275	18	,	,	PUNCT
brj-22592	275	19	müller	müller	PROPN
brj-22592	275	20	,	,	PUNCT
brj-22592	275	21	m.	m.	NOUN
brj-22592	275	22	,	,	PUNCT
brj-22592	275	23	thabet	thabet	ADJ
brj-22592	275	24	,	,	PUNCT
brj-22592	275	25	a.	a.	NOUN
brj-22592	275	26	,	,	PUNCT
brj-22592	275	27	and	and	CCONJ
brj-22592	275	28	ghanem	ghanem	PROPN
brj-22592	275	29	,	,	PUNCT
brj-22592	275	30	b.	b.	PROPN
brj-22592	275	31	(	(	PUNCT
brj-22592	275	32	2020	2020	NUM
brj-22592	275	33	)	)	PUNCT
brj-22592	275	34	.	.	PUNCT
brj-22592	276	1	“	"	PUNCT
brj-22592	276	2	sgas	sga	NOUN
brj-22592	276	3	:	:	PUNCT
brj-22592	276	4	sequential	sequential	ADJ
brj-22592	276	5	greedy	greedy	ADJ
brj-22592	276	6	architecture	architecture	NOUN
brj-22592	276	7	search	search	NOUN
brj-22592	276	8	,	,	PUNCT
brj-22592	276	9	”	"	PUNCT
brj-22592	276	10	proceedings	proceeding	NOUN
brj-22592	276	11	of	of	ADP
brj-22592	276	12	the	the	DET
brj-22592	276	13	ieee	ieee	NOUN
brj-22592	276	14	computer	computer	NOUN
brj-22592	276	15	society	society	PROPN
brj-22592	276	16	conference	conference	NOUN
brj-22592	276	17	on	on	ADP
brj-22592	276	18	computer	computer	NOUN
brj-22592	276	19	vision	vision	NOUN
brj-22592	276	20	and	and	CCONJ
brj-22592	276	21	pattern	pattern	NOUN
brj-22592	276	22	recognition	recognition	NOUN
brj-22592	276	23	.	.	PUNCT
brj-22592	277	1	doi	doi	NOUN
brj-22592	277	2	:	:	PUNCT
brj-22592	277	3	10.1109	10.1109	NUM
brj-22592	277	4	/	/	SYM
brj-22592	277	5	cvpr42600.2020.00169	cvpr42600.2020.00169	PROPN
brj-22592	277	6	ling	ling	PROPN
brj-22592	277	7	,	,	PUNCT
brj-22592	277	8	j.	j.	PROPN
brj-22592	277	9	x.	x.	PROPN
brj-22592	277	10	,	,	PUNCT
brj-22592	277	11	and	and	CCONJ
brj-22592	277	12	xie	xie	PROPN
brj-22592	277	13	,	,	PUNCT
brj-22592	277	14	y.	y.	PROPN
brj-22592	277	15	h.	h.	PROPN
brj-22592	277	16	(	(	PUNCT
brj-22592	277	17	2022	2022	NUM
brj-22592	277	18	)	)	PUNCT
brj-22592	277	19	.	.	PUNCT
brj-22592	278	1	“	"	PUNCT
brj-22592	278	2	research	research	NOUN
brj-22592	278	3	on	on	ADP
brj-22592	278	4	wood	wood	NOUN
brj-22592	278	5	defects	defect	NOUN
brj-22592	278	6	classification	classification	NOUN
brj-22592	278	7	based	base	VERB
brj-22592	278	8	on	on	ADP
brj-22592	278	9	deep	deep	ADJ
brj-22592	278	10	learning	learning	NOUN
brj-22592	278	11	,	,	PUNCT
brj-22592	278	12	”	"	PUNCT
brj-22592	278	13	wood	wood	NOUN
brj-22592	278	14	research	research	NOUN
brj-22592	278	15	67	67	NUM
brj-22592	278	16	.	.	PUNCT
brj-22592	279	1	doi	doi	NOUN
brj-22592	279	2	:	:	PUNCT
brj-22592	279	3	10.37763	10.37763	NUM
brj-22592	279	4	/	/	SYM
brj-22592	279	5	wr.1336	wr.1336	VERB
brj-22592	279	6	-	-	PUNCT
brj-22592	279	7	4561/67.1.147156	4561/67.1.147156	PROPN
brj-22592	279	8	liu	liu	PROPN
brj-22592	279	9	,	,	PUNCT
brj-22592	279	10	c.	c.	PROPN
brj-22592	279	11	r.	r.	PROPN
brj-22592	279	12	(	(	PUNCT
brj-22592	279	13	2008	2008	NUM
brj-22592	279	14	)	)	PUNCT
brj-22592	279	15	.	.	PUNCT
brj-22592	280	1	“	"	PUNCT
brj-22592	280	2	development	development	NOUN
brj-22592	280	3	status	status	NOUN
brj-22592	280	4	and	and	CCONJ
brj-22592	280	5	prospects	prospect	NOUN
brj-22592	280	6	of	of	ADP
brj-22592	280	7	china	china	PROPN
brj-22592	280	8	’s	’s	PART
brj-22592	280	9	wood	wood	NOUN
brj-22592	280	10	industry	industry	NOUN
brj-22592	280	11	,	,	PUNCT
brj-22592	280	12	”	"	PUNCT
brj-22592	280	13	shaanxi	shaanxi	PROPN
brj-22592	280	14	forestry	forestry	PROPN
brj-22592	280	15	141(03	141(03	NUM
brj-22592	280	16	)	)	PUNCT
brj-22592	280	17	,	,	PUNCT
brj-22592	280	18	15	15	NUM
brj-22592	280	19	-	-	SYM
brj-22592	280	20	16	16	NUM
brj-22592	280	21	.	.	PUNCT
brj-22592	281	1	doi	doi	NOUN
brj-22592	281	2	:	:	PUNCT
brj-22592	281	3	cnki	cnki	ADJ
brj-22592	281	4	:	:	PUNCT
brj-22592	281	5	sun	sun	NOUN
brj-22592	281	6	:	:	PUNCT
brj-22592	281	7	sxni.0.2008	sxni.0.2008	NOUN
brj-22592	281	8	-	-	PUNCT
brj-22592	281	9	03	03	NUM
brj-22592	281	10	-	-	PUNCT
brj-22592	281	11	012	012	NUM
brj-22592	281	12	.	.	PUNCT
brj-22592	282	1	liu	liu	PROPN
brj-22592	282	2	,	,	PUNCT
brj-22592	282	3	y.	y.	PROPN
brj-22592	282	4	,	,	PUNCT
brj-22592	282	5	shao	shao	PROPN
brj-22592	282	6	,	,	PUNCT
brj-22592	282	7	z.	z.	PROPN
brj-22592	282	8	,	,	PUNCT
brj-22592	282	9	and	and	CCONJ
brj-22592	282	10	teng	teng	PROPN
brj-22592	282	11	,	,	PUNCT
brj-22592	282	12	y.	y.	PROPN
brj-22592	282	13	(	(	PUNCT
brj-22592	282	14	2021	2021	NUM
brj-22592	282	15	)	)	PUNCT
brj-22592	282	16	.	.	PUNCT
brj-22592	283	1	“	"	PUNCT
brj-22592	283	2	nam	nam	NOUN
brj-22592	283	3	:	:	PUNCT
brj-22592	283	4	normalization	normalization	NOUN
brj-22592	283	5	-	-	PUNCT
brj-22592	283	6	based	base	VERB
brj-22592	283	7	attention	attention	NOUN
brj-22592	283	8	module	module	NOUN
brj-22592	283	9	,	,	PUNCT
brj-22592	283	10	”	"	PUNCT
brj-22592	283	11	doi	doi	NOUN
brj-22592	283	12	:	:	PUNCT
brj-22592	283	13	10.48550	10.48550	NUM
brj-22592	283	14	/	/	SYM
brj-22592	283	15	arxiv.2111.12419	arxiv.2111.12419	NUM
brj-22592	283	16	luo	luo	PROPN
brj-22592	283	17	,	,	PUNCT
brj-22592	283	18	w.	w.	PROPN
brj-22592	283	19	,	,	PUNCT
brj-22592	283	20	and	and	CCONJ
brj-22592	283	21	sun	sun	NOUN
brj-22592	283	22	,	,	PUNCT
brj-22592	283	23	l.	l.	PROPN
brj-22592	283	24	p.	p.	PROPN
brj-22592	283	25	(	(	PUNCT
brj-22592	283	26	2019	2019	NUM
brj-22592	283	27	)	)	PUNCT
brj-22592	283	28	.	.	PUNCT
brj-22592	284	1	“	"	PUNCT
brj-22592	284	2	support	support	NOUN
brj-22592	284	3	vector	vector	NOUN
brj-22592	284	4	machine	machine	NOUN
brj-22592	284	5	learning	learn	VERB
brj-22592	284	6	classification	classification	NOUN
brj-22592	284	7	of	of	ADP
brj-22592	284	8	wood	wood	NOUN
brj-22592	284	9	defects	defect	NOUN
brj-22592	284	10	using	use	VERB
brj-22592	284	11	local	local	ADJ
brj-22592	284	12	binary	binary	ADJ
brj-22592	284	13	patterns	pattern	NOUN
brj-22592	284	14	and	and	CCONJ
brj-22592	284	15	directional	directional	ADJ
brj-22592	284	16	gradient	gradient	NOUN
brj-22592	284	17	histogram	histogram	NOUN
brj-22592	284	18	fusion	fusion	NOUN
brj-22592	284	19	features	feature	NOUN
brj-22592	284	20	,	,	PUNCT
brj-22592	284	21	”	"	PUNCT
brj-22592	284	22	journal	journal	NOUN
brj-22592	284	23	of	of	ADP
brj-22592	284	24	northeast	northeast	ADJ
brj-22592	284	25	forestry	forestry	PROPN
brj-22592	284	26	university	university	PROPN
brj-22592	284	27	47(06	47(06	PROPN
brj-22592	284	28	)	)	PUNCT
brj-22592	284	29	,	,	PUNCT
brj-22592	284	30	70	70	NUM
brj-22592	284	31	-	-	SYM
brj-22592	284	32	73	73	NUM
brj-22592	284	33	.	.	PUNCT
brj-22592	285	1	doi	doi	NOUN
brj-22592	285	2	:	:	PUNCT
brj-22592	285	3	10.13759	10.13759	NUM
brj-22592	285	4	/	/	SYM
brj-22592	285	5	j.cnki.dlxb.2019.06.014	j.cnki.dlxb.2019.06.014	PROPN
brj-22592	285	6	peer	peer	NOUN
brj-22592	285	7	-	-	PUNCT
brj-22592	285	8	reviewed	review	VERB
brj-22592	285	9	article	article	NOUN
brj-22592	285	10	bioresources.com	bioresources.com	X
brj-22592	285	11	xie	xie	PROPN
brj-22592	285	12	&	&	CCONJ
brj-22592	285	13	ling	ling	PROPN
brj-22592	285	14	(	(	PUNCT
brj-22592	285	15	2023	2023	NUM
brj-22592	285	16	)	)	PUNCT
brj-22592	285	17	.	.	PUNCT
brj-22592	286	1	“	"	PUNCT
brj-22592	286	2	wood	wood	NOUN
brj-22592	286	3	defect	defect	NOUN
brj-22592	286	4	classification	classification	NOUN
brj-22592	286	5	,	,	PUNCT
brj-22592	286	6	”	"	PUNCT
brj-22592	286	7	bioresources	bioresource	NOUN
brj-22592	286	8	18(4	18(4	NUM
brj-22592	286	9	)	)	PUNCT
brj-22592	286	10	,	,	PUNCT
brj-22592	286	11	7663	7663	NUM
brj-22592	286	12	-	-	SYM
brj-22592	286	13	7680	7680	NUM
brj-22592	286	14	.	.	PUNCT
brj-22592	287	1	7680	7680	NUM
brj-22592	287	2	radosavovic	radosavovic	ADJ
brj-22592	287	3	,	,	PUNCT
brj-22592	287	4	i.	i.	PROPN
brj-22592	287	5	,	,	PUNCT
brj-22592	287	6	kosaraju	kosaraju	PROPN
brj-22592	287	7	,	,	PUNCT
brj-22592	287	8	r.	r.	PROPN
brj-22592	287	9	p.	p.	PROPN
brj-22592	287	10	,	,	PUNCT
brj-22592	287	11	girshick	girshick	PROPN
brj-22592	287	12	,	,	PUNCT
brj-22592	287	13	r.	r.	PROPN
brj-22592	287	14	,	,	PUNCT
brj-22592	287	15	he	he	PRON
brj-22592	287	16	,	,	PUNCT
brj-22592	287	17	k.	k.	PROPN
brj-22592	287	18	m.	m.	PROPN
brj-22592	287	19	,	,	PUNCT
brj-22592	287	20	and	and	CCONJ
brj-22592	287	21	dollár	dollár	NOUN
brj-22592	287	22	,	,	PUNCT
brj-22592	287	23	p.	p.	NOUN
brj-22592	287	24	(	(	PUNCT
brj-22592	287	25	2020	2020	NUM
brj-22592	287	26	)	)	PUNCT
brj-22592	287	27	.	.	PUNCT
brj-22592	288	1	“	"	PUNCT
brj-22592	288	2	designing	design	VERB
brj-22592	288	3	network	network	NOUN
brj-22592	288	4	design	design	NOUN
brj-22592	288	5	spaces	space	NOUN
brj-22592	288	6	,	,	PUNCT
brj-22592	288	7	”	"	PUNCT
brj-22592	288	8	proceedings	proceeding	NOUN
brj-22592	288	9	of	of	ADP
brj-22592	288	10	the	the	DET
brj-22592	288	11	ieee	ieee	NOUN
brj-22592	288	12	computer	computer	NOUN
brj-22592	288	13	society	society	PROPN
brj-22592	288	14	conference	conference	NOUN
brj-22592	288	15	on	on	ADP
brj-22592	288	16	computer	computer	NOUN
brj-22592	288	17	vision	vision	NOUN
brj-22592	288	18	and	and	CCONJ
brj-22592	288	19	pattern	pattern	NOUN
brj-22592	288	20	recognition	recognition	NOUN
brj-22592	288	21	.	.	PUNCT
brj-22592	289	1	doi	doi	NOUN
brj-22592	289	2	:	:	PUNCT
brj-22592	289	3	10.1109	10.1109	NUM
brj-22592	289	4	/	/	SYM
brj-22592	289	5	cvpr42600.2020.01044	cvpr42600.2020.01044	PROPN
brj-22592	289	6	riana	riana	PROPN
brj-22592	289	7	,	,	PUNCT
brj-22592	289	8	d.	d.	PROPN
brj-22592	289	9	,	,	PUNCT
brj-22592	289	10	rahayu	rahayu	PROPN
brj-22592	289	11	,	,	PUNCT
brj-22592	289	12	s.	s.	PROPN
brj-22592	289	13	,	,	PUNCT
brj-22592	289	14	hasan	hasan	PROPN
brj-22592	289	15	,	,	PUNCT
brj-22592	289	16	m.	m.	NOUN
brj-22592	289	17	,	,	PUNCT
brj-22592	289	18	and	and	CCONJ
brj-22592	289	19	anton	anton	NOUN
brj-22592	289	20	(	(	PUNCT
brj-22592	289	21	2021	2021	NUM
brj-22592	289	22	)	)	PUNCT
brj-22592	289	23	.	.	PUNCT
brj-22592	290	1	“	"	PUNCT
brj-22592	290	2	comparison	comparison	NOUN
brj-22592	290	3	of	of	ADP
brj-22592	290	4	segmentation	segmentation	NOUN
brj-22592	290	5	and	and	CCONJ
brj-22592	290	6	identification	identification	NOUN
brj-22592	290	7	of	of	ADP
brj-22592	290	8	swietenia	swietenia	NOUN
brj-22592	290	9	mahagoni	mahagoni	VERB
brj-22592	290	10	wood	wood	NOUN
brj-22592	290	11	defects	defect	NOUN
brj-22592	290	12	with	with	ADP
brj-22592	290	13	augmentation	augmentation	NOUN
brj-22592	290	14	images	image	NOUN
brj-22592	290	15	,	,	PUNCT
brj-22592	290	16	”	"	PUNCT
brj-22592	290	17	heliyou	heliyou	PROPN
brj-22592	290	18	7(6	7(6	NUM
brj-22592	290	19	)	)	PUNCT
brj-22592	290	20	.	.	PUNCT
brj-22592	291	1	doi	doi	NOUN
brj-22592	291	2	:	:	PUNCT
brj-22592	291	3	10.1016	10.1016	NUM
brj-22592	291	4	/	/	SYM
brj-22592	291	5	j.heliyon.2021.e07417	j.heliyon.2021.e07417	NUM
brj-22592	291	6	shi	shi	PROPN
brj-22592	291	7	,	,	PUNCT
brj-22592	291	8	g.y	g.y	PROPN
brj-22592	291	9	.	.	PROPN
brj-22592	291	10	,	,	PUNCT
brj-22592	291	11	cao	cao	PROPN
brj-22592	291	12	,	,	PUNCT
brj-22592	291	13	j.	j.	PROPN
brj-22592	291	14	,	,	PUNCT
brj-22592	291	15	and	and	CCONJ
brj-22592	291	16	zhang	zhang	PROPN
brj-22592	291	17	,	,	PUNCT
brj-22592	291	18	y.	y.	PROPN
brj-22592	291	19	z.	z.	PROPN
brj-22592	291	20	(	(	PUNCT
brj-22592	291	21	2020	2020	NUM
brj-22592	291	22	)	)	PUNCT
brj-22592	291	23	.	.	PUNCT
brj-22592	292	1	“	"	PUNCT
brj-22592	292	2	research	research	NOUN
brj-22592	292	3	on	on	ADP
brj-22592	292	4	near	near	ADP
brj-22592	292	5	infrared	infrared	ADJ
brj-22592	292	6	recognition	recognition	NOUN
brj-22592	292	7	method	method	NOUN
brj-22592	292	8	of	of	ADP
brj-22592	292	9	wood	wood	NOUN
brj-22592	292	10	defects	defect	NOUN
brj-22592	292	11	based	base	VERB
brj-22592	292	12	on	on	ADP
brj-22592	292	13	transfer	transfer	NOUN
brj-22592	292	14	learning	learning	NOUN
brj-22592	292	15	,	,	PUNCT
brj-22592	292	16	”	"	PUNCT
brj-22592	292	17	journal	journal	NOUN
brj-22592	292	18	of	of	ADP
brj-22592	292	19	electrical	electrical	ADJ
brj-22592	292	20	machinery	machinery	NOUN
brj-22592	292	21	and	and	CCONJ
brj-22592	292	22	control	control	NOUN
brj-22592	292	23	24	24	NUM
brj-22592	292	24	.	.	PUNCT
brj-22592	293	1	doi	doi	NOUN
brj-22592	293	2	:	:	PUNCT
brj-22592	293	3	10.15938	10.15938	NUM
brj-22592	293	4	/	/	SYM
brj-22592	293	5	j.emc.2020.10.018	j.emc.2020.10.018	PROPN
brj-22592	293	6	sun	sun	NOUN
brj-22592	293	7	,	,	PUNCT
brj-22592	293	8	y.n	y.n	PROPN
brj-22592	293	9	.	.	PROPN
brj-22592	293	10	,	,	PUNCT
brj-22592	293	11	xue	xue	PROPN
brj-22592	293	12	,	,	PUNCT
brj-22592	293	13	b.	b.	PROPN
brj-22592	293	14	,	,	PUNCT
brj-22592	293	15	zhang	zhang	PROPN
brj-22592	293	16	,	,	PUNCT
brj-22592	293	17	m.	m.	PROPN
brj-22592	293	18	j.	j.	PROPN
brj-22592	293	19	,	,	PUNCT
brj-22592	293	20	and	and	CCONJ
brj-22592	293	21	yen	yen	PROPN
brj-22592	293	22	gary	gary	PROPN
brj-22592	293	23	,	,	PUNCT
brj-22592	293	24	g.	g.	PROPN
brj-22592	293	25	(	(	PUNCT
brj-22592	293	26	2020	2020	NUM
brj-22592	293	27	)	)	PUNCT
brj-22592	293	28	.	.	PUNCT
brj-22592	294	1	“	"	PUNCT
brj-22592	294	2	evolving	evolve	VERB
brj-22592	294	3	deep	deep	ADJ
brj-22592	294	4	convolutional	convolutional	ADJ
brj-22592	294	5	neural	neural	ADJ
brj-22592	294	6	networks	network	NOUN
brj-22592	294	7	for	for	ADP
brj-22592	294	8	image	image	NOUN
brj-22592	294	9	classification	classification	NOUN
brj-22592	294	10	,	,	PUNCT
brj-22592	294	11	”	"	PUNCT
brj-22592	294	12	ieee	ieee	NOUN
brj-22592	294	13	transactions	transaction	NOUN
brj-22592	294	14	on	on	ADP
brj-22592	294	15	evolutionary	evolutionary	ADJ
brj-22592	294	16	computation	computation	NOUN
brj-22592	295	1	24	24	NUM
brj-22592	295	2	.	.	PUNCT
brj-22592	295	3	doi	doi	NOUN
brj-22592	295	4	:	:	PUNCT
brj-22592	295	5	10.1109	10.1109	NUM
brj-22592	295	6	/	/	SYM
brj-22592	295	7	tevc.2019.2916183	tevc.2019.2916183	NUM
brj-22592	295	8	wang	wang	PROPN
brj-22592	295	9	,	,	PUNCT
brj-22592	295	10	n.	n.	PROPN
brj-22592	295	11	,	,	PUNCT
brj-22592	295	12	gao	gao	PROPN
brj-22592	295	13	,	,	PUNCT
brj-22592	295	14	y.	y.	PROPN
brj-22592	295	15	,	,	PUNCT
brj-22592	295	16	chen	chen	PROPN
brj-22592	295	17	,	,	PUNCT
brj-22592	295	18	h.	h.	PROPN
brj-22592	295	19	,	,	PUNCT
brj-22592	295	20	wang	wang	PROPN
brj-22592	295	21	,	,	PUNCT
brj-22592	295	22	p.	p.	PROPN
brj-22592	295	23	,	,	PUNCT
brj-22592	295	24	tian	tian	PROPN
brj-22592	295	25	,	,	PUNCT
brj-22592	295	26	z.	z.	PROPN
brj-22592	295	27	,	,	PUNCT
brj-22592	295	28	shen	shen	PROPN
brj-22592	295	29	,	,	PUNCT
brj-22592	295	30	c.	c.	PROPN
brj-22592	295	31	,	,	PUNCT
brj-22592	295	32	and	and	CCONJ
brj-22592	295	33	zhang	zhang	PROPN
brj-22592	295	34	,	,	PUNCT
brj-22592	295	35	y.	y.	PROPN
brj-22592	295	36	(	(	PUNCT
brj-22592	295	37	2020	2020	NUM
brj-22592	295	38	)	)	PUNCT
brj-22592	295	39	.	.	PUNCT
brj-22592	296	1	“	"	PUNCT
brj-22592	296	2	nasfcos	nasfcos	ADJ
brj-22592	296	3	:	:	PUNCT
brj-22592	296	4	fast	fast	ADJ
brj-22592	296	5	neural	neural	ADJ
brj-22592	296	6	architecture	architecture	NOUN
brj-22592	296	7	search	search	NOUN
brj-22592	296	8	for	for	ADP
brj-22592	296	9	object	object	NOUN
brj-22592	296	10	detection	detection	NOUN
brj-22592	296	11	,	,	PUNCT
brj-22592	296	12	”	"	PUNCT
brj-22592	296	13	ieee	ieee	NOUN
brj-22592	296	14	/	/	SYM
brj-22592	296	15	cvf	cvf	NOUN
brj-22592	296	16	conference	conference	NOUN
brj-22592	296	17	on	on	ADP
brj-22592	296	18	computer	computer	NOUN
brj-22592	296	19	vision	vision	NOUN
brj-22592	296	20	and	and	CCONJ
brj-22592	296	21	pattern	pattern	NOUN
brj-22592	296	22	recognition	recognition	NOUN
brj-22592	296	23	(	(	PUNCT
brj-22592	296	24	cvpr	cvpr	NOUN
brj-22592	296	25	)	)	PUNCT
brj-22592	296	26	.	.	PUNCT
brj-22592	297	1	ieee.doi:10.1109	ieee.doi:10.1109	VERB
brj-22592	297	2	/	/	SYM
brj-22592	297	3	cvpr42600.2020.01196	cvpr42600.2020.01196	PROPN
brj-22592	297	4	xie	xie	PROPN
brj-22592	297	5	,	,	PUNCT
brj-22592	297	6	y.	y.	PROPN
brj-22592	297	7	h.	h.	PROPN
brj-22592	298	1	(	(	PUNCT
brj-22592	298	2	2014	2014	NUM
brj-22592	298	3	)	)	PUNCT
brj-22592	298	4	.	.	PUNCT
brj-22592	299	1	“	"	PUNCT
brj-22592	299	2	research	research	NOUN
brj-22592	299	3	on	on	ADP
brj-22592	299	4	the	the	DET
brj-22592	299	5	application	application	NOUN
brj-22592	299	6	of	of	ADP
brj-22592	299	7	digital	digital	ADJ
brj-22592	299	8	image	image	NOUN
brj-22592	299	9	processing	processing	NOUN
brj-22592	299	10	technology	technology	NOUN
brj-22592	299	11	in	in	ADP
brj-22592	299	12	wood	wood	NOUN
brj-22592	299	13	surface	surface	NOUN
brj-22592	299	14	defect	defect	NOUN
brj-22592	299	15	detection	detection	NOUN
brj-22592	299	16	,	,	PUNCT
brj-22592	299	17	”	"	PUNCT
brj-22592	299	18	northeast	northeast	ADJ
brj-22592	299	19	forestry	forestry	PROPN
brj-22592	299	20	university	university	PROPN
brj-22592	299	21	yan	yan	PROPN
brj-22592	299	22	,	,	PUNCT
brj-22592	299	23	f.	f.	PROPN
brj-22592	299	24	,	,	PUNCT
brj-22592	299	25	zhang	zhang	PROPN
brj-22592	299	26	,	,	PUNCT
brj-22592	299	27	j.	j.	PROPN
brj-22592	299	28	h.	h.	PROPN
brj-22592	299	29	,	,	PUNCT
brj-22592	299	30	yao	yao	PROPN
brj-22592	299	31	,	,	PUNCT
brj-22592	299	32	y.c	y.c	PROPN
brj-22592	299	33	.	.	PROPN
brj-22592	299	34	,	,	PUNCT
brj-22592	299	35	and	and	CCONJ
brj-22592	299	36	liu	liu	PROPN
brj-22592	299	37	,	,	PUNCT
brj-22592	299	38	j.	j.	PROPN
brj-22592	299	39	(	(	PUNCT
brj-22592	299	40	2022	2022	NUM
brj-22592	299	41	)	)	PUNCT
brj-22592	299	42	.	.	PUNCT
brj-22592	300	1	“	"	PUNCT
brj-22592	300	2	improved	improve	VERB
brj-22592	300	3	unet	unet	NOUN
brj-22592	300	4	wood	wood	NOUN
brj-22592	300	5	defect	defect	NOUN
brj-22592	300	6	image	image	NOUN
brj-22592	300	7	segmentation	segmentation	NOUN
brj-22592	300	8	method	method	NOUN
brj-22592	300	9	,	,	PUNCT
brj-22592	300	10	”	"	PUNCT
brj-22592	300	11	forestry	forestry	NOUN
brj-22592	300	12	machinery	machinery	NOUN
brj-22592	300	13	and	and	CCONJ
brj-22592	300	14	woodworking	woodworke	VERB
brj-22592	300	15	equipment	equipment	NOUN
brj-22592	300	16	50	50	NUM
brj-22592	300	17	,	,	PUNCT
brj-22592	300	18	41	41	NUM
brj-22592	300	19	-	-	SYM
brj-22592	300	20	45	45	NUM
brj-22592	300	21	.	.	PUNCT
brj-22592	301	1	doi	doi	NOUN
brj-22592	301	2	:	:	PUNCT
brj-22592	301	3	10.13279	10.13279	NUM
brj-22592	301	4	/	/	SYM
brj-22592	301	5	j.cnki.fmwe.2022.0005	j.cnki.fmwe.2022.0005	PROPN
brj-22592	301	6	.	.	PUNCT
brj-22592	302	1	zhang	zhang	PROPN
brj-22592	302	2	,	,	PUNCT
brj-22592	302	3	s.	s.	PROPN
brj-22592	302	4	,	,	PUNCT
brj-22592	302	5	wang	wang	PROPN
brj-22592	302	6	,	,	PUNCT
brj-22592	302	7	y.	y.	PROPN
brj-22592	302	8	b.	b.	PROPN
brj-22592	302	9	,	,	PUNCT
brj-22592	302	10	yang	yang	PROPN
brj-22592	302	11	,	,	PUNCT
brj-22592	302	12	t.	t.	PROPN
brj-22592	302	13	,	,	PUNCT
brj-22592	302	14	and	and	CCONJ
brj-22592	302	15	li	li	PROPN
brj-22592	302	16	,	,	PUNCT
brj-22592	302	17	m.	m.	NOUN
brj-22592	302	18	(	(	PUNCT
brj-22592	302	19	2021	2021	NUM
brj-22592	302	20	)	)	PUNCT
brj-22592	302	21	.	.	PUNCT
brj-22592	303	1	“	"	PUNCT
brj-22592	303	2	research	research	NOUN
brj-22592	303	3	on	on	ADP
brj-22592	303	4	identification	identification	NOUN
brj-22592	303	5	method	method	NOUN
brj-22592	303	6	of	of	ADP
brj-22592	303	7	typical	typical	ADJ
brj-22592	303	8	defects	defect	NOUN
brj-22592	303	9	on	on	ADP
brj-22592	303	10	wood	wood	NOUN
brj-22592	303	11	surface	surface	NOUN
brj-22592	303	12	based	base	VERB
brj-22592	303	13	on	on	ADP
brj-22592	303	14	improved	improved	ADJ
brj-22592	303	15	lenet-5	lenet-5	PROPN
brj-22592	303	16	model	model	NOUN
brj-22592	303	17	,	,	PUNCT
brj-22592	303	18	”	"	PUNCT
brj-22592	303	19	wood	wood	NOUN
brj-22592	303	20	science	science	NOUN
brj-22592	303	21	and	and	CCONJ
brj-22592	303	22	technology	technology	NOUN
brj-22592	303	23	35	35	NUM
brj-22592	303	24	,	,	PUNCT
brj-22592	303	25	31	31	NUM
brj-22592	303	26	-	-	SYM
brj-22592	303	27	37	37	NUM
brj-22592	303	28	.	.	PUNCT
brj-22592	304	1	doi	doi	NOUN
brj-22592	304	2	:	:	PUNCT
brj-22592	304	3	10.12326	10.12326	NUM
brj-22592	304	4	/	/	SYM
brj-22592	304	5	j.2096	j.2096	NOUN
brj-22592	304	6	-	-	PUNCT
brj-22592	304	7	9694.2021024	9694.2021024	NOUN
brj-22592	304	8	zhang	zhang	PROPN
brj-22592	304	9	,	,	PUNCT
brj-22592	304	10	k.	k.	PROPN
brj-22592	304	11	,	,	PUNCT
brj-22592	304	12	feng	feng	PROPN
brj-22592	304	13	,	,	PUNCT
brj-22592	304	14	x.	x.	PROPN
brj-22592	304	15	h.	h.	PROPN
brj-22592	304	16	,	,	PUNCT
brj-22592	304	17	and	and	CCONJ
brj-22592	304	18	guo	guo	PROPN
brj-22592	304	19	,	,	PUNCT
brj-22592	304	20	y.r	y.r	PROPN
brj-22592	304	21	.	.	PROPN
brj-22592	304	22	(	(	PUNCT
brj-22592	304	23	2021	2021	NUM
brj-22592	304	24	)	)	PUNCT
brj-22592	304	25	.	.	PUNCT
brj-22592	305	1	“	"	PUNCT
brj-22592	305	2	overview	overview	NOUN
brj-22592	305	3	of	of	ADP
brj-22592	305	4	deep	deep	ADJ
brj-22592	305	5	convolutional	convolutional	ADJ
brj-22592	305	6	neural	neural	ADJ
brj-22592	305	7	network	network	NOUN
brj-22592	305	8	models	model	NOUN
brj-22592	305	9	for	for	ADP
brj-22592	305	10	image	image	NOUN
brj-22592	305	11	classification	classification	NOUN
brj-22592	305	12	,	,	PUNCT
brj-22592	305	13	”	"	PUNCT
brj-22592	305	14	chinese	chinese	ADJ
brj-22592	305	15	journal	journal	NOUN
brj-22592	305	16	of	of	ADP
brj-22592	305	17	image	image	NOUN
brj-22592	305	18	graphics	graphic	NOUN
brj-22592	305	19	26	26	NUM
brj-22592	305	20	(	(	PUNCT
brj-22592	305	21	10	10	NUM
brj-22592	305	22	):	):	PUNCT
brj-22592	305	23	2305	2305	NUM
brj-22592	305	24	-	-	SYM
brj-22592	305	25	2325	2325	NUM
brj-22592	305	26	.	.	PUNCT
brj-22592	306	1	doi	doi	NOUN
brj-22592	306	2	:	:	PUNCT
brj-22592	306	3	10.11834	10.11834	NUM
brj-22592	306	4	/	/	SYM
brj-22592	306	5	jig.200302	jig.200302	PROPN
brj-22592	306	6	.	.	PROPN
brj-22592	306	7	zoph	zoph	PROPN
brj-22592	306	8	,	,	PUNCT
brj-22592	306	9	b.	b.	PROPN
brj-22592	306	10	,	,	PUNCT
brj-22592	306	11	and	and	CCONJ
brj-22592	306	12	le	le	X
brj-22592	306	13	,	,	PUNCT
brj-22592	306	14	q.	q.	PROPN
brj-22592	306	15	v.	v.	PROPN
brj-22592	306	16	(	(	PUNCT
brj-22592	306	17	2016	2016	NUM
brj-22592	306	18	)	)	PUNCT
brj-22592	306	19	.	.	PUNCT
brj-22592	307	1	“	"	PUNCT
brj-22592	307	2	neural	neural	ADJ
brj-22592	307	3	architecture	architecture	NOUN
brj-22592	307	4	search	search	NOUN
brj-22592	307	5	with	with	ADP
brj-22592	307	6	reinforcement	reinforcement	NOUN
brj-22592	307	7	learning	learning	NOUN
brj-22592	307	8	,	,	PUNCT
brj-22592	307	9	”	"	PUNCT
brj-22592	307	10	corr	corr	NOUN
brj-22592	307	11	.	.	PUNCT
brj-22592	308	1	doi	doi	NOUN
brj-22592	308	2	:	:	PUNCT
brj-22592	308	3	10.48550	10.48550	NUM
brj-22592	308	4	/	/	SYM
brj-22592	308	5	arxiv:1611.01578	arxiv:1611.01578	NOUN
brj-22592	308	6	article	article	NOUN
brj-22592	308	7	submitted	submit	VERB
brj-22592	308	8	:	:	PUNCT
brj-22592	308	9	march	march	PROPN
brj-22592	308	10	30	30	NUM
brj-22592	308	11	,	,	PUNCT
brj-22592	308	12	2023	2023	NUM
brj-22592	308	13	;	;	PUNCT
brj-22592	308	14	peer	peer	NOUN
brj-22592	308	15	review	review	NOUN
brj-22592	308	16	completed	complete	VERB
brj-22592	308	17	:	:	PUNCT
brj-22592	308	18	september	september	PROPN
brj-22592	308	19	2	2	NUM
brj-22592	308	20	,	,	PUNCT
brj-22592	308	21	2023	2023	NUM
brj-22592	308	22	;	;	PUNCT
brj-22592	308	23	revised	revise	VERB
brj-22592	308	24	version	version	NOUN
brj-22592	308	25	received	receive	VERB
brj-22592	308	26	:	:	PUNCT
brj-22592	308	27	september	september	PROPN
brj-22592	308	28	10	10	NUM
brj-22592	308	29	,	,	PUNCT
brj-22592	308	30	2023	2023	NUM
brj-22592	308	31	;	;	PUNCT
brj-22592	308	32	accepted	accept	VERB
brj-22592	308	33	:	:	PUNCT
brj-22592	308	34	september	september	PROPN
brj-22592	308	35	22	22	NUM
brj-22592	308	36	,	,	PUNCT
brj-22592	308	37	2023	2023	NUM
brj-22592	308	38	;	;	PUNCT
brj-22592	308	39	published	publish	VERB
brj-22592	308	40	:	:	PUNCT
brj-22592	308	41	september	september	PROPN
brj-22592	308	42	27	27	NUM
brj-22592	308	43	,	,	PUNCT
brj-22592	308	44	2023	2023	NUM
brj-22592	308	45	.	.	PUNCT
brj-22592	309	1	doi	doi	NOUN
brj-22592	309	2	:	:	PUNCT
brj-22592	309	3	10.15376	10.15376	NUM
brj-22592	309	4	/	/	SYM
brj-22592	309	5	biores.18.4.7663	biores.18.4.7663	PROPN
brj-22592	309	6	-	-	PUNCT
brj-22592	309	7	7680	7680	NUM
brj-22592	309	8	https://schlr.cnki.net/detail/doi/garj2021_2/xqspe46cf5ee31458fe534acc04ca216ad0f	https://schlr.cnki.net/detail/doi/garj2021_2/xqspe46cf5ee31458fe534acc04ca216ad0f	NOUN