id	sid	tid	token	lemma	pos
ajst-23440	1	1	academic	academic	ADJ
ajst-23440	1	2	journal	journal	NOUN
ajst-23440	1	3	of	of	ADP
ajst-23440	1	4	science	science	NOUN
ajst-23440	1	5	and	and	CCONJ
ajst-23440	1	6	technology	technology	NOUN
ajst-23440	1	7	issn	issn	NOUN
ajst-23440	1	8	:	:	PUNCT
ajst-23440	1	9	2771	2771	NUM
ajst-23440	1	10	-	-	SYM
ajst-23440	1	11	3032	3032	NUM
ajst-23440	1	12	|	|	NOUN
ajst-23440	1	13	vol	vol	NOUN
ajst-23440	1	14	.	.	PROPN
ajst-23440	2	1	11	11	NUM
ajst-23440	2	2	,	,	PUNCT
ajst-23440	2	3	no	no	INTJ
ajst-23440	2	4	.	.	NOUN
ajst-23440	2	5	3	3	NUM
ajst-23440	2	6	,	,	PUNCT
ajst-23440	2	7	2024	2024	NUM
ajst-23440	2	8	45	45	NUM
ajst-23440	2	9	research	research	NOUN
ajst-23440	2	10	on	on	ADP
ajst-23440	2	11	tool	tool	NOUN
ajst-23440	2	12	detection	detection	NOUN
ajst-23440	2	13	algorithm	algorithm	NOUN
ajst-23440	2	14	based	base	VERB
ajst-23440	2	15	on	on	ADP
ajst-23440	2	16	yolov8	yolov8	PROPN
ajst-23440	2	17	improved	improve	VERB
ajst-23440	2	18	model	model	NOUN
ajst-23440	2	19	wanpeng	wanpeng	PROPN
ajst-23440	2	20	qi	qi	PROPN
ajst-23440	2	21	tianjin	tianjin	PROPN
ajst-23440	2	22	university	university	PROPN
ajst-23440	2	23	of	of	ADP
ajst-23440	2	24	technology	technology	NOUN
ajst-23440	2	25	and	and	CCONJ
ajst-23440	2	26	education	education	NOUN
ajst-23440	2	27	,	,	PUNCT
ajst-23440	2	28	china	china	PROPN
ajst-23440	2	29	abstract	abstract	PROPN
ajst-23440	2	30	:	:	PUNCT
ajst-23440	2	31	with	with	ADP
ajst-23440	2	32	the	the	DET
ajst-23440	2	33	continuous	continuous	ADJ
ajst-23440	2	34	improvement	improvement	NOUN
ajst-23440	2	35	of	of	ADP
ajst-23440	2	36	people	people	NOUN
ajst-23440	2	37	's	's	PART
ajst-23440	2	38	living	living	NOUN
ajst-23440	2	39	standards	standard	NOUN
ajst-23440	2	40	,	,	PUNCT
ajst-23440	2	41	the	the	DET
ajst-23440	2	42	public	public	NOUN
ajst-23440	2	43	's	's	PART
ajst-23440	2	44	awareness	awareness	NOUN
ajst-23440	2	45	of	of	ADP
ajst-23440	2	46	protecting	protect	VERB
ajst-23440	2	47	the	the	DET
ajst-23440	2	48	right	right	NOUN
ajst-23440	2	49	to	to	ADP
ajst-23440	2	50	life	life	NOUN
ajst-23440	2	51	safety	safety	NOUN
ajst-23440	2	52	has	have	AUX
ajst-23440	2	53	become	become	VERB
ajst-23440	2	54	increasingly	increasingly	ADV
ajst-23440	2	55	important	important	ADJ
ajst-23440	2	56	,	,	PUNCT
ajst-23440	2	57	and	and	CCONJ
ajst-23440	2	58	higher	high	ADJ
ajst-23440	2	59	requirements	requirement	NOUN
ajst-23440	2	60	have	have	AUX
ajst-23440	2	61	been	be	AUX
ajst-23440	2	62	put	put	VERB
ajst-23440	2	63	forward	forward	ADV
ajst-23440	2	64	for	for	ADP
ajst-23440	2	65	the	the	DET
ajst-23440	2	66	prevention	prevention	NOUN
ajst-23440	2	67	of	of	ADP
ajst-23440	2	68	public	public	ADJ
ajst-23440	2	69	safety	safety	NOUN
ajst-23440	2	70	.	.	PUNCT
ajst-23440	3	1	controlled	control	VERB
ajst-23440	3	2	cutting	cutting	NOUN
ajst-23440	3	3	tools	tool	NOUN
ajst-23440	3	4	are	be	AUX
ajst-23440	3	5	a	a	DET
ajst-23440	3	6	part	part	NOUN
ajst-23440	3	7	of	of	ADP
ajst-23440	3	8	cutting	cut	VERB
ajst-23440	3	9	tools	tool	NOUN
ajst-23440	3	10	,	,	PUNCT
ajst-23440	3	11	which	which	PRON
ajst-23440	3	12	are	be	AUX
ajst-23440	3	13	controlled	control	VERB
ajst-23440	3	14	and	and	CCONJ
ajst-23440	3	15	managed	manage	VERB
ajst-23440	3	16	by	by	ADP
ajst-23440	3	17	law	law	NOUN
ajst-23440	3	18	to	to	PART
ajst-23440	3	19	prevent	prevent	VERB
ajst-23440	3	20	criminals	criminal	NOUN
ajst-23440	3	21	from	from	ADP
ajst-23440	3	22	using	use	VERB
ajst-23440	3	23	them	they	PRON
ajst-23440	3	24	as	as	ADP
ajst-23440	3	25	weapons	weapon	NOUN
ajst-23440	3	26	for	for	ADP
ajst-23440	3	27	illegal	illegal	ADJ
ajst-23440	3	28	activities	activity	NOUN
ajst-23440	3	29	.	.	PUNCT
ajst-23440	4	1	to	to	PART
ajst-23440	4	2	prevent	prevent	VERB
ajst-23440	4	3	criminals	criminal	NOUN
ajst-23440	4	4	from	from	ADP
ajst-23440	4	5	committing	commit	VERB
ajst-23440	4	6	crimes	crime	NOUN
ajst-23440	4	7	in	in	ADP
ajst-23440	4	8	places	place	NOUN
ajst-23440	4	9	with	with	ADP
ajst-23440	4	10	high	high	ADJ
ajst-23440	4	11	population	population	NOUN
ajst-23440	4	12	mobility	mobility	NOUN
ajst-23440	4	13	,	,	PUNCT
ajst-23440	4	14	such	such	ADJ
ajst-23440	4	15	as	as	ADP
ajst-23440	4	16	subways	subway	NOUN
ajst-23440	4	17	,	,	PUNCT
ajst-23440	4	18	campuses	campus	NOUN
ajst-23440	4	19	,	,	PUNCT
ajst-23440	4	20	airports	airport	NOUN
ajst-23440	4	21	,	,	PUNCT
ajst-23440	4	22	and	and	CCONJ
ajst-23440	4	23	other	other	ADJ
ajst-23440	4	24	enclosed	enclose	VERB
ajst-23440	4	25	areas	area	NOUN
ajst-23440	4	26	with	with	ADP
ajst-23440	4	27	high	high	ADJ
ajst-23440	4	28	population	population	NOUN
ajst-23440	4	29	mobility	mobility	NOUN
ajst-23440	4	30	.	.	PUNCT
ajst-23440	5	1	this	this	DET
ajst-23440	5	2	article	article	NOUN
ajst-23440	5	3	proposes	propose	VERB
ajst-23440	5	4	a	a	DET
ajst-23440	5	5	controlled	control	VERB
ajst-23440	5	6	tool	tool	NOUN
ajst-23440	5	7	detection	detection	NOUN
ajst-23440	5	8	algorithm	algorithm	NOUN
ajst-23440	5	9	based	base	VERB
ajst-23440	5	10	on	on	ADP
ajst-23440	5	11	the	the	DET
ajst-23440	5	12	improved	improved	ADJ
ajst-23440	5	13	yolov8	yolov8	NOUN
ajst-23440	5	14	model	model	NOUN
ajst-23440	5	15	to	to	PART
ajst-23440	5	16	improve	improve	VERB
ajst-23440	5	17	the	the	DET
ajst-23440	5	18	detection	detection	NOUN
ajst-23440	5	19	accuracy	accuracy	NOUN
ajst-23440	5	20	and	and	CCONJ
ajst-23440	5	21	efficiency	efficiency	NOUN
ajst-23440	5	22	of	of	ADP
ajst-23440	5	23	controlled	control	VERB
ajst-23440	5	24	tools	tool	NOUN
ajst-23440	5	25	.	.	PUNCT
ajst-23440	6	1	after	after	ADP
ajst-23440	6	2	experiments	experiment	NOUN
ajst-23440	6	3	,	,	PUNCT
ajst-23440	6	4	se	se	X
ajst-23440	6	5	attention	attention	NOUN
ajst-23440	6	6	mechanism	mechanism	NOUN
ajst-23440	6	7	was	be	AUX
ajst-23440	6	8	added	add	VERB
ajst-23440	6	9	to	to	ADP
ajst-23440	6	10	the	the	DET
ajst-23440	6	11	original	original	ADJ
ajst-23440	6	12	yolov8	yolov8	NOUN
ajst-23440	6	13	algorithm	algorithm	PROPN
ajst-23440	6	14	model	model	NOUN
ajst-23440	6	15	,	,	PUNCT
ajst-23440	6	16	and	and	CCONJ
ajst-23440	6	17	the	the	DET
ajst-23440	6	18	loss	loss	NOUN
ajst-23440	6	19	function	function	NOUN
ajst-23440	6	20	siou	siou	NOUN
ajst-23440	6	21	was	be	AUX
ajst-23440	6	22	added	add	VERB
ajst-23440	6	23	to	to	ADP
ajst-23440	6	24	the	the	DET
ajst-23440	6	25	yolov8	yolov8	NOUN
ajst-23440	6	26	algorithm	algorithm	PROPN
ajst-23440	6	27	.	.	PUNCT
ajst-23440	7	1	the	the	DET
ajst-23440	7	2	improved	improved	ADJ
ajst-23440	7	3	map	map	NOUN
ajst-23440	7	4	increased	increase	VERB
ajst-23440	7	5	from	from	ADP
ajst-23440	7	6	the	the	DET
ajst-23440	7	7	initial	initial	ADJ
ajst-23440	7	8	80.8	80.8	NUM
ajst-23440	7	9	%	%	NOUN
ajst-23440	7	10	to	to	ADP
ajst-23440	7	11	the	the	DET
ajst-23440	7	12	improved	improved	ADJ
ajst-23440	7	13	85	85	NUM
ajst-23440	7	14	%	%	NOUN
ajst-23440	7	15	,	,	PUNCT
ajst-23440	7	16	an	an	DET
ajst-23440	7	17	increase	increase	NOUN
ajst-23440	7	18	of	of	ADP
ajst-23440	7	19	4.2	4.2	NUM
ajst-23440	7	20	%	%	NOUN
ajst-23440	7	21	.	.	PUNCT
ajst-23440	8	1	this	this	PRON
ajst-23440	8	2	provides	provide	VERB
ajst-23440	8	3	a	a	DET
ajst-23440	8	4	certain	certain	ADJ
ajst-23440	8	5	reference	reference	NOUN
ajst-23440	8	6	value	value	NOUN
ajst-23440	8	7	for	for	ADP
ajst-23440	8	8	future	future	ADJ
ajst-23440	8	9	algorithm	algorithm	NOUN
ajst-23440	8	10	models	model	NOUN
ajst-23440	8	11	for	for	ADP
ajst-23440	8	12	identifying	identify	VERB
ajst-23440	8	13	controlled	control	VERB
ajst-23440	8	14	tools	tool	NOUN
ajst-23440	8	15	.	.	PUNCT
ajst-23440	9	1	keywords	keyword	NOUN
ajst-23440	9	2	:	:	PUNCT
ajst-23440	9	3	yolov8	yolov8	NOUN
ajst-23440	9	4	;	;	PUNCT
ajst-23440	9	5	se	se	X
ajst-23440	9	6	attention	attention	NOUN
ajst-23440	9	7	mechanism	mechanism	NOUN
ajst-23440	9	8	;	;	PUNCT
ajst-23440	9	9	loss	loss	NOUN
ajst-23440	9	10	function	function	NOUN
ajst-23440	9	11	siou	siou	NOUN
ajst-23440	9	12	;	;	PUNCT
ajst-23440	9	13	map	map	NOUN
ajst-23440	9	14	.	.	PUNCT
ajst-23440	10	1	1	1	X
ajst-23440	10	2	.	.	X
ajst-23440	10	3	introduction	introduction	NOUN
ajst-23440	10	4	with	with	ADP
ajst-23440	10	5	the	the	DET
ajst-23440	10	6	increasing	increase	VERB
ajst-23440	10	7	prosperity	prosperity	NOUN
ajst-23440	10	8	of	of	ADP
ajst-23440	10	9	our	our	PRON
ajst-23440	10	10	country	country	NOUN
ajst-23440	10	11	's	's	PART
ajst-23440	10	12	national	national	ADJ
ajst-23440	10	13	strength	strength	NOUN
ajst-23440	10	14	,	,	PUNCT
ajst-23440	10	15	the	the	DET
ajst-23440	10	16	continuous	continuous	ADJ
ajst-23440	10	17	improvement	improvement	NOUN
ajst-23440	10	18	of	of	ADP
ajst-23440	10	19	economic	economic	ADJ
ajst-23440	10	20	level	level	NOUN
ajst-23440	10	21	,	,	PUNCT
ajst-23440	10	22	and	and	CCONJ
ajst-23440	10	23	the	the	DET
ajst-23440	10	24	continuous	continuous	ADJ
ajst-23440	10	25	improvement	improvement	NOUN
ajst-23440	10	26	of	of	ADP
ajst-23440	10	27	people	people	NOUN
ajst-23440	10	28	's	's	PART
ajst-23440	10	29	happiness	happiness	NOUN
ajst-23440	10	30	index	index	NOUN
ajst-23440	10	31	,	,	PUNCT
ajst-23440	10	32	the	the	DET
ajst-23440	10	33	public	public	NOUN
ajst-23440	10	34	's	's	PART
ajst-23440	10	35	awareness	awareness	NOUN
ajst-23440	10	36	of	of	ADP
ajst-23440	10	37	protecting	protect	VERB
ajst-23440	10	38	the	the	DET
ajst-23440	10	39	right	right	NOUN
ajst-23440	10	40	to	to	ADP
ajst-23440	10	41	life	life	NOUN
ajst-23440	10	42	and	and	CCONJ
ajst-23440	10	43	safety	safety	NOUN
ajst-23440	10	44	is	be	AUX
ajst-23440	10	45	also	also	ADV
ajst-23440	10	46	constantly	constantly	ADV
ajst-23440	10	47	increasing	increase	VERB
ajst-23440	10	48	.	.	PUNCT
ajst-23440	11	1	controlled	control	VERB
ajst-23440	11	2	knives	knife	NOUN
ajst-23440	11	3	play	play	VERB
ajst-23440	11	4	an	an	DET
ajst-23440	11	5	essential	essential	ADJ
ajst-23440	11	6	role	role	NOUN
ajst-23440	11	7	in	in	ADP
ajst-23440	11	8	daily	daily	ADJ
ajst-23440	11	9	life	life	NOUN
ajst-23440	11	10	,	,	PUNCT
ajst-23440	11	11	but	but	CCONJ
ajst-23440	11	12	if	if	SCONJ
ajst-23440	11	13	they	they	PRON
ajst-23440	11	14	fall	fall	VERB
ajst-23440	11	15	into	into	ADP
ajst-23440	11	16	the	the	DET
ajst-23440	11	17	hands	hand	NOUN
ajst-23440	11	18	of	of	ADP
ajst-23440	11	19	criminals	criminal	NOUN
ajst-23440	11	20	,	,	PUNCT
ajst-23440	11	21	they	they	PRON
ajst-23440	11	22	can	can	AUX
ajst-23440	11	23	pose	pose	VERB
ajst-23440	11	24	a	a	DET
ajst-23440	11	25	significant	significant	ADJ
ajst-23440	11	26	threat	threat	NOUN
ajst-23440	11	27	to	to	ADP
ajst-23440	11	28	the	the	DET
ajst-23440	11	29	safety	safety	NOUN
ajst-23440	11	30	of	of	ADP
ajst-23440	11	31	citizens	citizen	NOUN
ajst-23440	11	32	,	,	PUNCT
ajst-23440	11	33	especially	especially	ADV
ajst-23440	11	34	in	in	ADP
ajst-23440	11	35	densely	densely	ADV
ajst-23440	11	36	populated	populate	VERB
ajst-23440	11	37	areas	area	NOUN
ajst-23440	11	38	such	such	ADJ
ajst-23440	11	39	as	as	ADP
ajst-23440	11	40	subways	subway	NOUN
ajst-23440	11	41	,	,	PUNCT
ajst-23440	11	42	campuses	campus	NOUN
ajst-23440	11	43	,	,	PUNCT
ajst-23440	11	44	airports	airport	NOUN
ajst-23440	11	45	,	,	PUNCT
ajst-23440	11	46	and	and	CCONJ
ajst-23440	11	47	other	other	ADJ
ajst-23440	11	48	closed	closed	ADJ
ajst-23440	11	49	areas	area	NOUN
ajst-23440	11	50	with	with	ADP
ajst-23440	11	51	high	high	ADJ
ajst-23440	11	52	population	population	NOUN
ajst-23440	11	53	mobility	mobility	NOUN
ajst-23440	11	54	.	.	PUNCT
ajst-23440	12	1	the	the	DET
ajst-23440	12	2	monitoring	monitoring	NOUN
ajst-23440	12	3	and	and	CCONJ
ajst-23440	12	4	control	control	NOUN
ajst-23440	12	5	of	of	ADP
ajst-23440	12	6	knives	knife	NOUN
ajst-23440	12	7	should	should	AUX
ajst-23440	12	8	be	be	AUX
ajst-23440	12	9	strengthened	strengthen	VERB
ajst-23440	12	10	.	.	PUNCT
ajst-23440	13	1	this	this	PRON
ajst-23440	13	2	also	also	ADV
ajst-23440	13	3	places	place	VERB
ajst-23440	13	4	higher	high	ADJ
ajst-23440	13	5	demands	demand	NOUN
ajst-23440	13	6	on	on	ADP
ajst-23440	13	7	algorithms	algorithm	NOUN
ajst-23440	13	8	for	for	ADP
ajst-23440	13	9	identifying	identify	VERB
ajst-23440	13	10	controlled	control	VERB
ajst-23440	13	11	cutting	cutting	NOUN
ajst-23440	13	12	tools	tool	NOUN
ajst-23440	13	13	.	.	PUNCT
ajst-23440	14	1	object	object	NOUN
ajst-23440	14	2	detection	detection	NOUN
ajst-23440	14	3	is	be	AUX
ajst-23440	14	4	one	one	NUM
ajst-23440	14	5	of	of	ADP
ajst-23440	14	6	the	the	DET
ajst-23440	14	7	most	most	ADV
ajst-23440	14	8	important	important	ADJ
ajst-23440	14	9	domains	domain	NOUN
ajst-23440	14	10	in	in	ADP
ajst-23440	14	11	computer	computer	NOUN
ajst-23440	14	12	vision	vision	NOUN
ajst-23440	14	13	tasks	task	NOUN
ajst-23440	14	14	,	,	PUNCT
ajst-23440	14	15	which	which	PRON
ajst-23440	14	16	is	be	AUX
ajst-23440	14	17	an	an	DET
ajst-23440	14	18	important	important	ADJ
ajst-23440	14	19	branch	branch	NOUN
ajst-23440	14	20	of	of	ADP
ajst-23440	14	21	artificial	artificial	ADJ
ajst-23440	14	22	intelligence	intelligence	NOUN
ajst-23440	14	23	.	.	PUNCT
ajst-23440	15	1	it	it	PRON
ajst-23440	15	2	aims	aim	VERB
ajst-23440	15	3	at	at	ADP
ajst-23440	15	4	finding	find	VERB
ajst-23440	15	5	and	and	CCONJ
ajst-23440	15	6	locating	locate	VERB
ajst-23440	15	7	the	the	DET
ajst-23440	15	8	accurate	accurate	ADJ
ajst-23440	15	9	position	position	NOUN
ajst-23440	15	10	of	of	ADP
ajst-23440	15	11	objects	object	NOUN
ajst-23440	15	12	in	in	ADP
ajst-23440	15	13	given	give	VERB
ajst-23440	15	14	pictures	picture	NOUN
ajst-23440	15	15	or	or	CCONJ
ajst-23440	15	16	videos	video	NOUN
ajst-23440	15	17	.	.	PUNCT
ajst-23440	16	1	with	with	ADP
ajst-23440	16	2	the	the	DET
ajst-23440	16	3	development	development	NOUN
ajst-23440	16	4	of	of	ADP
ajst-23440	16	5	deep	deep	ADJ
ajst-23440	16	6	learning	learning	NOUN
ajst-23440	16	7	techniques	technique	NOUN
ajst-23440	16	8	,	,	PUNCT
ajst-23440	16	9	more	more	ADV
ajst-23440	16	10	powerful	powerful	ADJ
ajst-23440	16	11	and	and	CCONJ
ajst-23440	16	12	robust	robust	ADJ
ajst-23440	16	13	algorithms	algorithm	NOUN
ajst-23440	16	14	have	have	AUX
ajst-23440	16	15	emerged	emerge	VERB
ajst-23440	16	16	to	to	PART
ajst-23440	16	17	deal	deal	VERB
ajst-23440	16	18	with	with	ADP
ajst-23440	16	19	multi	multi	ADJ
ajst-23440	16	20	-	-	NOUN
ajst-23440	16	21	scale	scale	ADJ
ajst-23440	16	22	,	,	PUNCT
ajst-23440	16	23	high	high	ADJ
ajst-23440	16	24	-	-	PUNCT
ajst-23440	16	25	level	level	NOUN
ajst-23440	16	26	features	feature	NOUN
ajst-23440	16	27	to	to	PART
ajst-23440	16	28	overcome	overcome	VERB
ajst-23440	16	29	the	the	DET
ajst-23440	16	30	limitations	limitation	NOUN
ajst-23440	16	31	of	of	ADP
ajst-23440	16	32	traditional	traditional	ADJ
ajst-23440	16	33	pipeline	pipeline	NOUN
ajst-23440	16	34	of	of	ADP
ajst-23440	16	35	object	object	NOUN
ajst-23440	16	36	detectors[1].at	detectors[1].at	VERB
ajst-23440	16	37	present	present	ADJ
ajst-23440	16	38	,	,	PUNCT
ajst-23440	16	39	object	object	VERB
ajst-23440	16	40	detection	detection	NOUN
ajst-23440	16	41	algorithms	algorithm	NOUN
ajst-23440	16	42	are	be	AUX
ajst-23440	16	43	mainly	mainly	ADV
ajst-23440	16	44	divided	divide	VERB
ajst-23440	16	45	into	into	ADP
ajst-23440	16	46	two	two	NUM
ajst-23440	16	47	types	type	NOUN
ajst-23440	16	48	.	.	PUNCT
ajst-23440	17	1	one	one	NUM
ajst-23440	17	2	is	be	AUX
ajst-23440	17	3	the	the	DET
ajst-23440	17	4	twostage	twostage	NOUN
ajst-23440	17	5	algorithm	algorithm	NOUN
ajst-23440	17	6	represented	represent	VERB
ajst-23440	17	7	by	by	ADP
ajst-23440	17	8	r	r	PROPN
ajst-23440	17	9	-	-	PUNCT
ajst-23440	17	10	cnn	cnn	PROPN
ajst-23440	17	11	series	series	NOUN
ajst-23440	17	12	algorithms	algorithms	PROPN
ajst-23440	17	13	.	.	PUNCT
ajst-23440	18	1	the	the	DET
ajst-23440	18	2	other	other	ADJ
ajst-23440	18	3	one	one	NOUN
ajst-23440	18	4	is	be	AUX
ajst-23440	18	5	the	the	DET
ajst-23440	18	6	one	one	NUM
ajst-23440	18	7	-	-	PUNCT
ajst-23440	18	8	stage	stage	NOUN
ajst-23440	18	9	algorithm	algorithm	NOUN
ajst-23440	18	10	represented	represent	VERB
ajst-23440	18	11	by	by	ADP
ajst-23440	18	12	you	you	PRON
ajst-23440	18	13	only	only	ADV
ajst-23440	18	14	look	look	VERB
ajst-23440	18	15	once	once	ADV
ajst-23440	18	16	(	(	PUNCT
ajst-23440	18	17	yolo	yolo	ADJ
ajst-23440	18	18	)	)	PUNCT
ajst-23440	18	19	series	series	PROPN
ajst-23440	18	20	algorithms	algorithm	NOUN
ajst-23440	18	21	.	.	PUNCT
ajst-23440	19	1	the	the	DET
ajst-23440	19	2	two	two	NUM
ajst-23440	19	3	-	-	PUNCT
ajst-23440	19	4	stage	stage	NOUN
ajst-23440	19	5	target	target	NOUN
ajst-23440	19	6	detection	detection	NOUN
ajst-23440	19	7	algorithm	algorithm	NOUN
ajst-23440	19	8	is	be	AUX
ajst-23440	19	9	based	base	VERB
ajst-23440	19	10	on	on	ADP
ajst-23440	19	11	the	the	DET
ajst-23440	19	12	idea	idea	NOUN
ajst-23440	19	13	of	of	ADP
ajst-23440	19	14	candidate	candidate	NOUN
ajst-23440	19	15	region[2	region[2	PROPN
ajst-23440	19	16	]	]	PUNCT
ajst-23440	19	17	.	.	PUNCT
ajst-23440	20	1	the	the	DET
ajst-23440	20	2	two	two	NUM
ajst-23440	20	3	-	-	PUNCT
ajst-23440	20	4	stage	stage	NOUN
ajst-23440	20	5	object	object	NOUN
ajst-23440	20	6	detection	detection	NOUN
ajst-23440	20	7	algorithm	algorithm	NOUN
ajst-23440	20	8	has	have	VERB
ajst-23440	20	9	high	high	ADJ
ajst-23440	20	10	detection	detection	NOUN
ajst-23440	20	11	accuracy	accuracy	NOUN
ajst-23440	20	12	,	,	PUNCT
ajst-23440	20	13	but	but	CCONJ
ajst-23440	20	14	the	the	DET
ajst-23440	20	15	detection	detection	NOUN
ajst-23440	20	16	speed	speed	NOUN
ajst-23440	20	17	is	be	AUX
ajst-23440	20	18	slower	slow	ADJ
ajst-23440	20	19	compared	compare	VERB
ajst-23440	20	20	to	to	ADP
ajst-23440	20	21	the	the	DET
ajst-23440	20	22	single	single	ADJ
ajst-23440	20	23	object	object	NOUN
ajst-23440	20	24	detection	detection	NOUN
ajst-23440	20	25	algorithm	algorithm	NOUN
ajst-23440	20	26	.	.	PUNCT
ajst-23440	21	1	one	one	NUM
ajst-23440	21	2	-	-	PUNCT
ajst-23440	21	3	stage	stage	NOUN
ajst-23440	21	4	target	target	NOUN
ajst-23440	21	5	detection	detection	NOUN
ajst-23440	21	6	algorithm	algorithm	NOUN
ajst-23440	21	7	is	be	AUX
ajst-23440	21	8	based	base	VERB
ajst-23440	21	9	on	on	ADP
ajst-23440	21	10	the	the	DET
ajst-23440	21	11	global	global	ADJ
ajst-23440	21	12	regression	regression	NOUN
ajst-23440	21	13	and	and	CCONJ
ajst-23440	21	14	classification	classification	NOUN
ajst-23440	21	15	,	,	PUNCT
ajst-23440	21	16	which	which	PRON
ajst-23440	21	17	directly	directly	ADV
ajst-23440	21	18	generates	generate	VERB
ajst-23440	21	19	the	the	DET
ajst-23440	21	20	location	location	NOUN
ajst-23440	21	21	and	and	CCONJ
ajst-23440	21	22	category	category	NOUN
ajst-23440	21	23	of	of	ADP
ajst-23440	21	24	the	the	DET
ajst-23440	21	25	target	target	NOUN
ajst-23440	21	26	object	object	NOUN
ajst-23440	21	27	.	.	PUNCT
ajst-23440	22	1	compared	compare	VERB
ajst-23440	22	2	with	with	ADP
ajst-23440	22	3	two	two	NUM
ajst-23440	22	4	-	-	PUNCT
ajst-23440	22	5	stage	stage	NOUN
ajst-23440	22	6	algorithm	algorithm	NOUN
ajst-23440	22	7	,	,	PUNCT
ajst-23440	22	8	one	one	NUM
ajst-23440	22	9	-	-	PUNCT
ajst-23440	22	10	stage	stage	NOUN
ajst-23440	22	11	algorithm	algorithm	NOUN
ajst-23440	22	12	is	be	AUX
ajst-23440	22	13	more	more	ADV
ajst-23440	22	14	real	real	ADJ
ajst-23440	22	15	-	-	PUNCT
ajst-23440	22	16	time	time	NOUN
ajst-23440	22	17	,	,	PUNCT
ajst-23440	22	18	but	but	CCONJ
ajst-23440	22	19	the	the	DET
ajst-23440	22	20	detection	detection	NOUN
ajst-23440	22	21	accuracy	accuracy	NOUN
ajst-23440	22	22	is	be	AUX
ajst-23440	22	23	lost[3].due	lost[3].due	VERB
ajst-23440	22	24	to	to	ADP
ajst-23440	22	25	the	the	DET
ajst-23440	22	26	fact	fact	NOUN
ajst-23440	22	27	that	that	SCONJ
ajst-23440	22	28	identifying	identify	VERB
ajst-23440	22	29	controlled	control	VERB
ajst-23440	22	30	tools	tool	NOUN
ajst-23440	22	31	belongs	belong	VERB
ajst-23440	22	32	to	to	ADP
ajst-23440	22	33	small	small	ADJ
ajst-23440	22	34	target	target	NOUN
ajst-23440	22	35	recognition	recognition	NOUN
ajst-23440	22	36	and	and	CCONJ
ajst-23440	22	37	requires	require	VERB
ajst-23440	22	38	real	real	ADJ
ajst-23440	22	39	-	-	PUNCT
ajst-23440	22	40	time	time	NOUN
ajst-23440	22	41	detection	detection	NOUN
ajst-23440	22	42	of	of	ADP
ajst-23440	22	43	whether	whether	SCONJ
ajst-23440	22	44	pedestrians	pedestrian	NOUN
ajst-23440	22	45	are	be	AUX
ajst-23440	22	46	carrying	carry	VERB
ajst-23440	22	47	controlled	control	VERB
ajst-23440	22	48	tools	tool	NOUN
ajst-23440	22	49	,	,	PUNCT
ajst-23440	22	50	we	we	PRON
ajst-23440	22	51	need	need	VERB
ajst-23440	22	52	a	a	DET
ajst-23440	22	53	single	single	ADJ
ajst-23440	22	54	target	target	NOUN
ajst-23440	22	55	detection	detection	NOUN
ajst-23440	22	56	algorithm	algorithm	NOUN
ajst-23440	22	57	with	with	ADP
ajst-23440	22	58	faster	fast	ADJ
ajst-23440	22	59	detection	detection	NOUN
ajst-23440	22	60	speed	speed	NOUN
ajst-23440	22	61	to	to	PART
ajst-23440	22	62	undertake	undertake	VERB
ajst-23440	22	63	the	the	DET
ajst-23440	22	64	detection	detection	NOUN
ajst-23440	22	65	task	task	NOUN
ajst-23440	22	66	in	in	ADP
ajst-23440	22	67	the	the	DET
ajst-23440	22	68	algorithm	algorithm	NOUN
ajst-23440	22	69	for	for	ADP
ajst-23440	22	70	identifying	identify	VERB
ajst-23440	22	71	controlled	control	VERB
ajst-23440	22	72	tools	tool	NOUN
ajst-23440	22	73	.	.	PUNCT
ajst-23440	23	1	due	due	ADP
ajst-23440	23	2	to	to	ADP
ajst-23440	23	3	the	the	DET
ajst-23440	23	4	small	small	ADJ
ajst-23440	23	5	target	target	NOUN
ajst-23440	23	6	size	size	NOUN
ajst-23440	23	7	of	of	ADP
ajst-23440	23	8	controlled	control	VERB
ajst-23440	23	9	cutting	cutting	NOUN
ajst-23440	23	10	tools	tool	NOUN
ajst-23440	23	11	and	and	CCONJ
ajst-23440	23	12	the	the	DET
ajst-23440	23	13	complexity	complexity	NOUN
ajst-23440	23	14	and	and	CCONJ
ajst-23440	23	15	diversity	diversity	NOUN
ajst-23440	23	16	of	of	ADP
ajst-23440	23	17	detection	detection	NOUN
ajst-23440	23	18	scenarios	scenario	NOUN
ajst-23440	23	19	,	,	PUNCT
ajst-23440	23	20	it	it	PRON
ajst-23440	23	21	is	be	AUX
ajst-23440	23	22	necessary	necessary	ADJ
ajst-23440	23	23	to	to	PART
ajst-23440	23	24	design	design	VERB
ajst-23440	23	25	efficient	efficient	ADJ
ajst-23440	23	26	,	,	PUNCT
ajst-23440	23	27	accurate	accurate	ADJ
ajst-23440	23	28	,	,	PUNCT
ajst-23440	23	29	and	and	CCONJ
ajst-23440	23	30	diverse	diverse	ADJ
ajst-23440	23	31	algorithms	algorithm	NOUN
ajst-23440	23	32	with	with	ADP
ajst-23440	23	33	diverse	diverse	ADJ
ajst-23440	23	34	application	application	NOUN
ajst-23440	23	35	scenarios	scenario	NOUN
ajst-23440	23	36	.	.	PUNCT
ajst-23440	24	1	therefore	therefore	ADV
ajst-23440	24	2	,	,	PUNCT
ajst-23440	24	3	this	this	DET
ajst-23440	24	4	article	article	NOUN
ajst-23440	24	5	will	will	AUX
ajst-23440	24	6	make	make	VERB
ajst-23440	24	7	certain	certain	ADJ
ajst-23440	24	8	improvements	improvement	NOUN
ajst-23440	24	9	to	to	ADP
ajst-23440	24	10	the	the	DET
ajst-23440	24	11	yolov8	yolov8	NOUN
ajst-23440	24	12	algorithm	algorithm	PROPN
ajst-23440	24	13	model	model	NOUN
ajst-23440	24	14	to	to	PART
ajst-23440	24	15	make	make	VERB
ajst-23440	24	16	the	the	DET
ajst-23440	24	17	improved	improved	ADJ
ajst-23440	24	18	model	model	NOUN
ajst-23440	24	19	more	more	ADV
ajst-23440	24	20	efficient	efficient	ADJ
ajst-23440	24	21	in	in	ADP
ajst-23440	24	22	recognition	recognition	NOUN
ajst-23440	24	23	.	.	PUNCT
ajst-23440	25	1	2	2	X
ajst-23440	25	2	.	.	X
ajst-23440	25	3	yolov8	yolov8	NOUN
ajst-23440	25	4	object	object	PROPN
ajst-23440	25	5	detection	detection	NOUN
ajst-23440	25	6	algorithm	algorithm	NOUN
ajst-23440	25	7	2.1	2.1	NUM
ajst-23440	25	8	.	.	PUNCT
ajst-23440	26	1	yolov8	yolov8	NOUN
ajst-23440	26	2	algorithm	algorithm	PROPN
ajst-23440	26	3	model	model	PROPN
ajst-23440	26	4	yolo	yolo	PROPN
ajst-23440	26	5	has	have	AUX
ajst-23440	26	6	become	become	VERB
ajst-23440	26	7	a	a	DET
ajst-23440	26	8	leading	lead	VERB
ajst-23440	26	9	object	object	NOUN
ajst-23440	26	10	detector	detector	NOUN
ajst-23440	26	11	due	due	ADP
ajst-23440	26	12	to	to	ADP
ajst-23440	26	13	its	its	PRON
ajst-23440	26	14	excellent	excellent	ADJ
ajst-23440	26	15	speed	speed	NOUN
ajst-23440	26	16	accuracy	accuracy	NOUN
ajst-23440	26	17	balance	balance	NOUN
ajst-23440	26	18	,	,	PUNCT
ajst-23440	26	19	and	and	CCONJ
ajst-23440	26	20	yolov8	yolov8	NOUN
ajst-23440	26	21	was	be	AUX
ajst-23440	26	22	opened	open	VERB
ajst-23440	26	23	by	by	ADP
ajst-23440	26	24	ultratics	ultratic	NOUN
ajst-23440	26	25	in	in	ADP
ajst-23440	26	26	january	january	PROPN
ajst-23440	26	27	2023	2023	NUM
ajst-23440	26	28	,	,	PUNCT
ajst-23440	26	29	providing	provide	VERB
ajst-23440	26	30	state	state	NOUN
ajst-23440	26	31	-	-	PUNCT
ajst-23440	26	32	of	of	ADP
ajst-23440	26	33	-	-	PUNCT
ajst-23440	26	34	the	the	DET
ajst-23440	26	35	-	-	PUNCT
ajst-23440	26	36	art	art	NOUN
ajst-23440	26	37	object	object	NOUN
ajst-23440	26	38	detection	detection	NOUN
ajst-23440	26	39	performance[4].in	performance[4].in	VERB
ajst-23440	26	40	the	the	DET
ajst-23440	26	41	yolov8	yolov8	NOUN
ajst-23440	26	42	algorithm	algorithm	PROPN
ajst-23440	26	43	,	,	PUNCT
ajst-23440	26	44	there	there	PRON
ajst-23440	26	45	are	be	VERB
ajst-23440	26	46	five	five	NUM
ajst-23440	26	47	different	different	ADJ
ajst-23440	26	48	models	model	NOUN
ajst-23440	26	49	:	:	PUNCT
ajst-23440	26	50	yolov8n	yolov8n	NOUN
ajst-23440	26	51	,	,	PUNCT
ajst-23440	26	52	yolov8s	yolov8	NOUN
ajst-23440	26	53	,	,	PUNCT
ajst-23440	26	54	yolov8	yolov8	PROPN
ajst-23440	26	55	m	m	PROPN
ajst-23440	26	56	,	,	PUNCT
ajst-23440	26	57	yolov8l	yolov8l	NOUN
ajst-23440	26	58	,	,	PUNCT
ajst-23440	26	59	and	and	CCONJ
ajst-23440	26	60	yolov8x	yolov8x	PROPN
ajst-23440	26	61	.	.	PUNCT
ajst-23440	27	1	the	the	DET
ajst-23440	27	2	depth	depth	NOUN
ajst-23440	27	3	and	and	CCONJ
ajst-23440	27	4	width	width	NOUN
ajst-23440	27	5	of	of	ADP
ajst-23440	27	6	these	these	DET
ajst-23440	27	7	five	five	NUM
ajst-23440	27	8	models	model	NOUN
ajst-23440	27	9	are	be	AUX
ajst-23440	27	10	constantly	constantly	ADV
ajst-23440	27	11	increasing	increase	VERB
ajst-23440	27	12	.	.	PUNCT
ajst-23440	28	1	among	among	ADP
ajst-23440	28	2	the	the	DET
ajst-23440	28	3	five	five	NUM
ajst-23440	28	4	models	model	NOUN
ajst-23440	28	5	,	,	PUNCT
ajst-23440	28	6	yolov8n	yolov8n	PROPN
ajst-23440	28	7	has	have	VERB
ajst-23440	28	8	the	the	DET
ajst-23440	28	9	smallest	small	ADJ
ajst-23440	28	10	depth	depth	NOUN
ajst-23440	28	11	and	and	CCONJ
ajst-23440	28	12	width	width	ADJ
ajst-23440	28	13	,	,	PUNCT
ajst-23440	28	14	because	because	SCONJ
ajst-23440	28	15	as	as	SCONJ
ajst-23440	28	16	the	the	DET
ajst-23440	28	17	depth	depth	NOUN
ajst-23440	28	18	and	and	CCONJ
ajst-23440	28	19	width	width	NOUN
ajst-23440	28	20	of	of	ADP
ajst-23440	28	21	the	the	DET
ajst-23440	28	22	models	model	NOUN
ajst-23440	28	23	increase	increase	VERB
ajst-23440	28	24	,	,	PUNCT
ajst-23440	28	25	although	although	SCONJ
ajst-23440	28	26	the	the	DET
ajst-23440	28	27	detection	detection	NOUN
ajst-23440	28	28	accuracy	accuracy	NOUN
ajst-23440	28	29	increases	increase	NOUN
ajst-23440	28	30	,	,	PUNCT
ajst-23440	28	31	the	the	DET
ajst-23440	28	32	detection	detection	NOUN
ajst-23440	28	33	speed	speed	NOUN
ajst-23440	28	34	decreases	decrease	VERB
ajst-23440	28	35	.	.	PUNCT
ajst-23440	29	1	as	as	SCONJ
ajst-23440	29	2	the	the	DET
ajst-23440	29	3	target	target	NOUN
ajst-23440	29	4	of	of	ADP
ajst-23440	29	5	detection	detection	NOUN
ajst-23440	29	6	is	be	AUX
ajst-23440	29	7	controlled	control	VERB
ajst-23440	29	8	cutting	cutting	NOUN
ajst-23440	29	9	tools	tool	NOUN
ajst-23440	29	10	and	and	CCONJ
ajst-23440	29	11	belongs	belong	VERB
ajst-23440	29	12	to	to	ADP
ajst-23440	29	13	small	small	ADJ
ajst-23440	29	14	target	target	NOUN
ajst-23440	29	15	detection	detection	NOUN
ajst-23440	29	16	,	,	PUNCT
ajst-23440	29	17	improving	improve	VERB
ajst-23440	29	18	detection	detection	NOUN
ajst-23440	29	19	speed	speed	NOUN
ajst-23440	29	20	can	can	AUX
ajst-23440	29	21	not	not	PART
ajst-23440	29	22	be	be	AUX
ajst-23440	29	23	ignored	ignore	VERB
ajst-23440	29	24	in	in	ADP
ajst-23440	29	25	improving	improve	VERB
ajst-23440	29	26	the	the	DET
ajst-23440	29	27	algorithm	algorithm	NOUN
ajst-23440	29	28	model	model	NOUN
ajst-23440	29	29	.	.	PUNCT
ajst-23440	30	1	therefore	therefore	ADV
ajst-23440	30	2	,	,	PUNCT
ajst-23440	30	3	this	this	DET
ajst-23440	30	4	article	article	NOUN
ajst-23440	30	5	will	will	AUX
ajst-23440	30	6	adopt	adopt	VERB
ajst-23440	30	7	the	the	DET
ajst-23440	30	8	yolov8n	yolov8n	NOUN
ajst-23440	30	9	model	model	NOUN
ajst-23440	30	10	and	and	CCONJ
ajst-23440	30	11	improve	improve	VERB
ajst-23440	30	12	the	the	DET
ajst-23440	30	13	algorithm	algorithm	NOUN
ajst-23440	30	14	based	base	VERB
ajst-23440	30	15	on	on	ADP
ajst-23440	30	16	it	it	PRON
ajst-23440	30	17	to	to	PART
ajst-23440	30	18	improve	improve	VERB
ajst-23440	30	19	the	the	DET
ajst-23440	30	20	recognition	recognition	NOUN
ajst-23440	30	21	efficiency	efficiency	NOUN
ajst-23440	30	22	of	of	ADP
ajst-23440	30	23	the	the	DET
ajst-23440	30	24	model	model	NOUN
ajst-23440	30	25	.	.	PUNCT
ajst-23440	31	1	2.2	2.2	NUM
ajst-23440	31	2	.	.	PUNCT
ajst-23440	31	3	overview	overview	NOUN
ajst-23440	31	4	of	of	ADP
ajst-23440	31	5	yolov8	yolov8	NOUN
ajst-23440	31	6	algorithm	algorithm	PROPN
ajst-23440	31	7	structure	structure	NOUN
ajst-23440	31	8	:	:	PUNCT
ajst-23440	31	9	the	the	DET
ajst-23440	31	10	yolov8	yolov8	NOUN
ajst-23440	31	11	algorithm	algorithm	PROPN
ajst-23440	31	12	provides	provide	VERB
ajst-23440	31	13	a	a	DET
ajst-23440	31	14	brand	brand	NOUN
ajst-23440	31	15	new	new	ADJ
ajst-23440	31	16	sota	sota	PROPN
ajst-23440	31	17	model	model	NOUN
ajst-23440	31	18	,	,	PUNCT
ajst-23440	31	19	which	which	PRON
ajst-23440	31	20	also	also	ADV
ajst-23440	31	21	offers	offer	VERB
ajst-23440	31	22	models	model	NOUN
ajst-23440	31	23	of	of	ADP
ajst-23440	31	24	different	different	ADJ
ajst-23440	31	25	sizes	size	NOUN
ajst-23440	31	26	based	base	VERB
ajst-23440	31	27	on	on	ADP
ajst-23440	31	28	scaling	scale	VERB
ajst-23440	31	29	coefficients	coefficient	NOUN
ajst-23440	31	30	to	to	PART
ajst-23440	31	31	meet	meet	VERB
ajst-23440	31	32	different	different	ADJ
ajst-23440	31	33	scene	scene	NOUN
ajst-23440	31	34	requirements	requirement	NOUN
ajst-23440	31	35	,	,	PUNCT
ajst-23440	31	36	while	while	SCONJ
ajst-23440	31	37	supporting	support	VERB
ajst-23440	31	38	image	image	NOUN
ajst-23440	31	39	classification	classification	NOUN
ajst-23440	31	40	,	,	PUNCT
ajst-23440	31	41	object	object	NOUN
ajst-23440	31	42	detection	detection	NOUN
ajst-23440	31	43	,	,	PUNCT
ajst-23440	31	44	instance	instance	NOUN
ajst-23440	31	45	segmentation	segmentation	NOUN
ajst-23440	31	46	,	,	PUNCT
ajst-23440	31	47	and	and	CCONJ
ajst-23440	31	48	pose	pose	VERB
ajst-23440	31	49	detection	detection	NOUN
ajst-23440	31	50	tasks	task	NOUN
ajst-23440	31	51	.	.	PUNCT
ajst-23440	32	1	the	the	DET
ajst-23440	32	2	yolov8	yolov8	NOUN
ajst-23440	32	3	algorithm	algorithm	PROPN
ajst-23440	32	4	is	be	AUX
ajst-23440	32	5	divided	divide	VERB
ajst-23440	32	6	into	into	ADP
ajst-23440	32	7	three	three	NUM
ajst-23440	32	8	parts	part	NOUN
ajst-23440	32	9	:	:	PUNCT
ajst-23440	32	10	backbone	backbone	NOUN
ajst-23440	32	11	,	,	PUNCT
ajst-23440	32	12	neck	neck	NOUN
ajst-23440	32	13	,	,	PUNCT
ajst-23440	32	14	and	and	CCONJ
ajst-23440	32	15	head	head	NOUN
ajst-23440	32	16	.	.	PUNCT
ajst-23440	33	1	backbone	backbone	NOUN
ajst-23440	33	2	network	network	NOUN
ajst-23440	33	3	usually	usually	ADV
ajst-23440	33	4	refers	refer	VERB
ajst-23440	33	5	to	to	ADP
ajst-23440	33	6	the	the	DET
ajst-23440	33	7	main	main	ADJ
ajst-23440	33	8	part	part	NOUN
ajst-23440	33	9	of	of	ADP
ajst-23440	33	10	the	the	DET
ajst-23440	33	11	entire	entire	ADJ
ajst-23440	33	12	deep	deep	ADJ
ajst-23440	33	13	neural	neural	ADJ
ajst-23440	33	14	network	network	NOUN
ajst-23440	33	15	,	,	PUNCT
ajst-23440	33	16	responsible	responsible	ADJ
ajst-23440	33	17	for	for	ADP
ajst-23440	33	18	extracting	extract	VERB
ajst-23440	33	19	features	feature	NOUN
ajst-23440	33	20	from	from	ADP
ajst-23440	33	21	input	input	NOUN
ajst-23440	33	22	data	datum	NOUN
ajst-23440	33	23	.	.	PUNCT
ajst-23440	34	1	backbone	backbone	NOUN
ajst-23440	34	2	networks	network	NOUN
ajst-23440	34	3	typically	typically	ADV
ajst-23440	34	4	consist	consist	VERB
ajst-23440	34	5	of	of	ADP
ajst-23440	34	6	multiple	multiple	ADJ
ajst-23440	34	7	convolutional	convolutional	ADJ
ajst-23440	34	8	layers	layer	NOUN
ajst-23440	34	9	or	or	CCONJ
ajst-23440	34	10	other	other	ADJ
ajst-23440	34	11	46	46	NUM
ajst-23440	34	12	feature	feature	NOUN
ajst-23440	34	13	extraction	extraction	NOUN
ajst-23440	34	14	layers	layer	NOUN
ajst-23440	34	15	,	,	PUNCT
ajst-23440	34	16	used	use	VERB
ajst-23440	34	17	to	to	PART
ajst-23440	34	18	gradually	gradually	ADV
ajst-23440	34	19	extract	extract	VERB
ajst-23440	34	20	advanced	advanced	ADJ
ajst-23440	34	21	features	feature	NOUN
ajst-23440	34	22	from	from	ADP
ajst-23440	34	23	input	input	NOUN
ajst-23440	34	24	data	datum	NOUN
ajst-23440	34	25	.	.	PUNCT
ajst-23440	35	1	neck	neck	NOUN
ajst-23440	35	2	:	:	PUNCT
ajst-23440	35	3	responsible	responsible	ADJ
ajst-23440	35	4	for	for	ADP
ajst-23440	35	5	further	further	ADJ
ajst-23440	35	6	processing	processing	NOUN
ajst-23440	35	7	and	and	CCONJ
ajst-23440	35	8	integrating	integrate	VERB
ajst-23440	35	9	the	the	DET
ajst-23440	35	10	features	feature	NOUN
ajst-23440	35	11	extracted	extract	VERB
ajst-23440	35	12	from	from	ADP
ajst-23440	35	13	the	the	DET
ajst-23440	35	14	backbone	backbone	NOUN
ajst-23440	35	15	network	network	NOUN
ajst-23440	35	16	.	.	PUNCT
ajst-23440	36	1	head	head	NOUN
ajst-23440	36	2	is	be	AUX
ajst-23440	36	3	responsible	responsible	ADJ
ajst-23440	36	4	for	for	ADP
ajst-23440	36	5	performing	perform	VERB
ajst-23440	36	6	specific	specific	ADJ
ajst-23440	36	7	tasks	task	NOUN
ajst-23440	36	8	such	such	ADJ
ajst-23440	36	9	as	as	ADP
ajst-23440	36	10	classification	classification	NOUN
ajst-23440	36	11	,	,	PUNCT
ajst-23440	36	12	object	object	NOUN
ajst-23440	36	13	detection	detection	NOUN
ajst-23440	36	14	,	,	PUNCT
ajst-23440	36	15	or	or	CCONJ
ajst-23440	36	16	semantic	semantic	ADJ
ajst-23440	36	17	segmentation	segmentation	NOUN
ajst-23440	36	18	.	.	PUNCT
ajst-23440	37	1	in	in	ADP
ajst-23440	37	2	yolov8	yolov8	PROPN
ajst-23440	37	3	,	,	PUNCT
ajst-23440	37	4	the	the	DET
ajst-23440	37	5	c3	c3	PROPN
ajst-23440	37	6	structure	structure	NOUN
ajst-23440	37	7	in	in	ADP
ajst-23440	37	8	yolov5	yolov5	NOUN
ajst-23440	37	9	was	be	AUX
ajst-23440	37	10	replaced	replace	VERB
ajst-23440	37	11	with	with	ADP
ajst-23440	37	12	a	a	DET
ajst-23440	37	13	c2f	c2f	NOUN
ajst-23440	37	14	structure	structure	NOUN
ajst-23440	37	15	with	with	ADP
ajst-23440	37	16	richer	rich	ADJ
ajst-23440	37	17	gradient	gradient	ADJ
ajst-23440	37	18	flow	flow	NOUN
ajst-23440	37	19	in	in	ADP
ajst-23440	37	20	the	the	DET
ajst-23440	37	21	backbone	backbone	NOUN
ajst-23440	37	22	network	network	NOUN
ajst-23440	37	23	and	and	CCONJ
ajst-23440	37	24	neck	neck	NOUN
ajst-23440	37	25	section	section	NOUN
ajst-23440	37	26	,	,	PUNCT
ajst-23440	37	27	which	which	PRON
ajst-23440	37	28	improved	improve	VERB
ajst-23440	37	29	the	the	DET
ajst-23440	37	30	performance	performance	NOUN
ajst-23440	37	31	of	of	ADP
ajst-23440	37	32	the	the	DET
ajst-23440	37	33	model	model	NOUN
ajst-23440	37	34	.	.	PUNCT
ajst-23440	38	1	the	the	DET
ajst-23440	38	2	head	head	NOUN
ajst-23440	38	3	section	section	NOUN
ajst-23440	38	4	has	have	AUX
ajst-23440	38	5	changed	change	VERB
ajst-23440	38	6	from	from	ADP
ajst-23440	38	7	the	the	DET
ajst-23440	38	8	original	original	ADJ
ajst-23440	38	9	coupling	coupling	NOUN
ajst-23440	38	10	head	head	NOUN
ajst-23440	38	11	to	to	ADP
ajst-23440	38	12	the	the	DET
ajst-23440	38	13	understanding	understanding	NOUN
ajst-23440	38	14	coupling	couple	VERB
ajst-23440	38	15	head	head	NOUN
ajst-23440	38	16	,	,	PUNCT
ajst-23440	38	17	and	and	CCONJ
ajst-23440	38	18	has	have	AUX
ajst-23440	38	19	changed	change	VERB
ajst-23440	38	20	from	from	ADP
ajst-23440	38	21	anchor	anchor	NOUN
ajst-23440	38	22	based	base	VERB
ajst-23440	38	23	in	in	ADP
ajst-23440	38	24	yolov5	yolov5	NOUN
ajst-23440	38	25	to	to	PART
ajst-23440	38	26	anchor	anchor	VERB
ajst-23440	38	27	free	free	ADJ
ajst-23440	38	28	.	.	PUNCT
ajst-23440	39	1	in	in	ADP
ajst-23440	39	2	terms	term	NOUN
ajst-23440	39	3	of	of	ADP
ajst-23440	39	4	computation	computation	NOUN
ajst-23440	39	5	,	,	PUNCT
ajst-23440	39	6	the	the	DET
ajst-23440	39	7	task	task	NOUN
ajst-23440	39	8	alignedassigner	alignedassigner	VERB
ajst-23440	39	9	positive	positive	ADJ
ajst-23440	39	10	sample	sample	NOUN
ajst-23440	39	11	allocation	allocation	NOUN
ajst-23440	39	12	strategy	strategy	NOUN
ajst-23440	39	13	has	have	AUX
ajst-23440	39	14	been	be	AUX
ajst-23440	39	15	adopted	adopt	VERB
ajst-23440	39	16	,	,	PUNCT
ajst-23440	39	17	and	and	CCONJ
ajst-23440	39	18	distribution	distribution	NOUN
ajst-23440	39	19	focal	focal	ADJ
ajst-23440	39	20	loss	loss	NOUN
ajst-23440	39	21	has	have	AUX
ajst-23440	39	22	been	be	AUX
ajst-23440	39	23	introduced	introduce	VERB
ajst-23440	39	24	the	the	DET
ajst-23440	39	25	network	network	NOUN
ajst-23440	39	26	structure	structure	NOUN
ajst-23440	39	27	of	of	ADP
ajst-23440	39	28	yolov8	yolov8	NOUN
ajst-23440	39	29	is	be	AUX
ajst-23440	39	30	shown	show	VERB
ajst-23440	39	31	in	in	ADP
ajst-23440	39	32	figure	figure	NOUN
ajst-23440	39	33	1	1	NUM
ajst-23440	39	34	:	:	PUNCT
ajst-23440	39	35	input	input	PROPN
ajst-23440	39	36	conv	conv	PROPN
ajst-23440	39	37	conv	conv	PROPN
ajst-23440	39	38	c2f	c2f	PROPN
ajst-23440	39	39	conv	conv	PROPN
ajst-23440	39	40	c2f	c2f	PROPN
ajst-23440	39	41	conv	conv	PROPN
ajst-23440	39	42	c2f	c2f	PROPN
ajst-23440	39	43	conv	conv	PROPN
ajst-23440	39	44	c2f	c2f	PROPN
ajst-23440	39	45	sppf	sppf	PROPN
ajst-23440	39	46	c2f	c2f	PROPN
ajst-23440	39	47	concat	concat	PROPN
ajst-23440	39	48	upsamlpe	upsamlpe	PROPN
ajst-23440	39	49	c2f	c2f	PROPN
ajst-23440	39	50	concat	concat	PROPN
ajst-23440	39	51	upsamlpe	upsamlpe	NOUN
ajst-23440	39	52	conv	conv	PROPN
ajst-23440	39	53	concat	concat	PROPN
ajst-23440	39	54	c2f	c2f	PROPN
ajst-23440	39	55	conv	conv	PROPN
ajst-23440	39	56	concat	concat	PROPN
ajst-23440	39	57	c2f	c2f	PROPN
ajst-23440	39	58	detect	detect	PROPN
ajst-23440	39	59	detect	detect	NOUN
ajst-23440	39	60	detect	detect	NOUN
ajst-23440	39	61	figure	figure	NOUN
ajst-23440	39	62	1	1	NUM
ajst-23440	39	63	.	.	PUNCT
ajst-23440	40	1	the	the	DET
ajst-23440	40	2	network	network	NOUN
ajst-23440	40	3	structure	structure	NOUN
ajst-23440	40	4	of	of	ADP
ajst-23440	40	5	yolov8	yolov8	PROPN
ajst-23440	40	6	3	3	NUM
ajst-23440	40	7	.	.	PUNCT
ajst-23440	41	1	network	network	NOUN
ajst-23440	41	2	improvement	improvement	NOUN
ajst-23440	41	3	of	of	ADP
ajst-23440	41	4	yolov8	yolov8	PROPN
ajst-23440	41	5	3.1	3.1	NUM
ajst-23440	41	6	.	.	PUNCT
ajst-23440	42	1	seattention	seattention	NOUN
ajst-23440	42	2	attention	attention	NOUN
ajst-23440	42	3	mechanism	mechanism	NOUN
ajst-23440	42	4	the	the	DET
ajst-23440	42	5	innovation	innovation	NOUN
ajst-23440	42	6	of	of	ADP
ajst-23440	42	7	seattention	seattention	NOUN
ajst-23440	42	8	lies	lie	VERB
ajst-23440	42	9	in	in	ADP
ajst-23440	42	10	its	its	PRON
ajst-23440	42	11	focus	focus	NOUN
ajst-23440	42	12	on	on	ADP
ajst-23440	42	13	the	the	DET
ajst-23440	42	14	relationships	relationship	NOUN
ajst-23440	42	15	between	between	ADP
ajst-23440	42	16	channels	channel	NOUN
ajst-23440	42	17	,	,	PUNCT
ajst-23440	42	18	hoping	hope	VERB
ajst-23440	42	19	that	that	SCONJ
ajst-23440	42	20	the	the	DET
ajst-23440	42	21	model	model	NOUN
ajst-23440	42	22	can	can	AUX
ajst-23440	42	23	automatically	automatically	ADV
ajst-23440	42	24	learn	learn	VERB
ajst-23440	42	25	the	the	DET
ajst-23440	42	26	importance	importance	NOUN
ajst-23440	42	27	of	of	ADP
ajst-23440	42	28	different	different	ADJ
ajst-23440	42	29	channel	channel	NOUN
ajst-23440	42	30	features	feature	NOUN
ajst-23440	42	31	,	,	PUNCT
ajst-23440	42	32	and	and	CCONJ
ajst-23440	42	33	focus	focus	VERB
ajst-23440	42	34	more	more	ADV
ajst-23440	42	35	on	on	ADP
ajst-23440	42	36	the	the	DET
ajst-23440	42	37	channel	channel	NOUN
ajst-23440	42	38	features	feature	VERB
ajst-23440	42	39	with	with	ADP
ajst-23440	42	40	the	the	DET
ajst-23440	42	41	highest	high	ADJ
ajst-23440	42	42	amount	amount	NOUN
ajst-23440	42	43	of	of	ADP
ajst-23440	42	44	information	information	NOUN
ajst-23440	42	45	,	,	PUNCT
ajst-23440	42	46	while	while	SCONJ
ajst-23440	42	47	suppressing	suppress	VERB
ajst-23440	42	48	those	those	DET
ajst-23440	42	49	unimportant	unimportant	ADJ
ajst-23440	42	50	channel	channel	NOUN
ajst-23440	42	51	features	feature	NOUN
ajst-23440	42	52	.	.	PUNCT
ajst-23440	43	1	seattention	seattention	NOUN
ajst-23440	43	2	mainly	mainly	ADV
ajst-23440	43	3	consists	consist	VERB
ajst-23440	43	4	of	of	ADP
ajst-23440	43	5	two	two	NUM
ajst-23440	43	6	parts	part	NOUN
ajst-23440	43	7	:	:	PUNCT
ajst-23440	43	8	squeeze	squeeze	NOUN
ajst-23440	43	9	and	and	CCONJ
ajst-23440	43	10	excitation	excitation	NOUN
ajst-23440	43	11	[	[	X
ajst-23440	43	12	5	5	NUM
ajst-23440	43	13	]	]	PUNCT
ajst-23440	43	14	.	.	PUNCT
ajst-23440	44	1	firstly	firstly	ADV
ajst-23440	44	2	,	,	PUNCT
ajst-23440	44	3	the	the	DET
ajst-23440	44	4	se	se	PROPN
ajst-23440	44	5	attention	attention	NOUN
ajst-23440	44	6	module	module	NOUN
ajst-23440	44	7	aggregates	aggregate	NOUN
ajst-23440	44	8	feature	feature	NOUN
ajst-23440	44	9	maps	map	NOUN
ajst-23440	44	10	across	across	ADP
ajst-23440	44	11	spatial	spatial	ADJ
ajst-23440	44	12	dimensions	dimension	NOUN
ajst-23440	44	13	through	through	ADP
ajst-23440	44	14	global	global	ADJ
ajst-23440	44	15	pooling	pooling	NOUN
ajst-23440	44	16	,	,	PUNCT
ajst-23440	44	17	i.e.	i.e.	X
ajst-23440	44	18	h	h	NOUN
ajst-23440	44	19	*	*	PUNCT
ajst-23440	44	20	w	w	PROPN
ajst-23440	44	21	*	*	PUNCT
ajst-23440	44	22	c	c	X
ajst-23440	44	23	→	→	SYM
ajst-23440	44	24	1	1	NUM
ajst-23440	44	25	*	*	SYM
ajst-23440	44	26	1	1	NUM
ajst-23440	44	27	*	*	SYM
ajst-23440	44	28	c	c	NOUN
ajst-23440	44	29	,	,	PUNCT
ajst-23440	44	30	and	and	CCONJ
ajst-23440	44	31	converts	convert	VERB
ajst-23440	44	32	them	they	PRON
ajst-23440	44	33	into	into	ADP
ajst-23440	44	34	feature	feature	NOUN
ajst-23440	44	35	maps	map	NOUN
ajst-23440	44	36	of	of	ADP
ajst-23440	44	37	size	size	NOUN
ajst-23440	44	38	1	1	NUM
ajst-23440	44	39	*	*	SYM
ajst-23440	44	40	1	1	NUM
ajst-23440	44	41	*	*	SYM
ajst-23440	44	42	c	c	X
ajst-23440	44	43	,	,	PUNCT
ajst-23440	44	44	through	through	ADP
ajst-23440	44	45	the	the	DET
ajst-23440	44	46	squeeze	squeeze	NOUN
ajst-23440	44	47	compression	compression	NOUN
ajst-23440	44	48	operation	operation	NOUN
ajst-23440	44	49	;	;	PUNCT
ajst-23440	44	50	the	the	DET
ajst-23440	44	51	incentive	incentive	NOUN
ajst-23440	44	52	process	process	NOUN
ajst-23440	44	53	involves	involve	VERB
ajst-23440	44	54	nonlinear	nonlinear	ADJ
ajst-23440	44	55	changes	change	NOUN
ajst-23440	44	56	to	to	ADP
ajst-23440	44	57	the	the	DET
ajst-23440	44	58	compression	compression	NOUN
ajst-23440	44	59	results	result	NOUN
ajst-23440	44	60	,	,	PUNCT
ajst-23440	44	61	combining	combine	VERB
ajst-23440	44	62	the	the	DET
ajst-23440	44	63	learned	learn	VERB
ajst-23440	44	64	channel	channel	NOUN
ajst-23440	44	65	attention	attention	NOUN
ajst-23440	44	66	information	information	NOUN
ajst-23440	44	67	with	with	ADP
ajst-23440	44	68	feature	feature	NOUN
ajst-23440	44	69	maps	map	NOUN
ajst-23440	44	70	to	to	PART
ajst-23440	44	71	reduce	reduce	VERB
ajst-23440	44	72	the	the	DET
ajst-23440	44	73	complexity	complexity	NOUN
ajst-23440	44	74	of	of	ADP
ajst-23440	44	75	the	the	DET
ajst-23440	44	76	model	model	NOUN
ajst-23440	44	77	and	and	CCONJ
ajst-23440	44	78	improve	improve	VERB
ajst-23440	44	79	its	its	PRON
ajst-23440	44	80	generalization	generalization	NOUN
ajst-23440	44	81	ability	ability	NOUN
ajst-23440	44	82	.	.	PUNCT
ajst-23440	45	1	the	the	DET
ajst-23440	45	2	structural	structural	ADJ
ajst-23440	45	3	diagram	diagram	NOUN
ajst-23440	45	4	of	of	ADP
ajst-23440	45	5	seattention	seattention	NOUN
ajst-23440	45	6	is	be	AUX
ajst-23440	45	7	shown	show	VERB
ajst-23440	45	8	in	in	ADP
ajst-23440	45	9	figure	figure	NOUN
ajst-23440	45	10	2	2	NUM
ajst-23440	45	11	,	,	PUNCT
ajst-23440	45	12	place	place	VERB
ajst-23440	45	13	the	the	DET
ajst-23440	45	14	seattention	seattention	NOUN
ajst-23440	45	15	attention	attention	NOUN
ajst-23440	45	16	mechanism	mechanism	NOUN
ajst-23440	45	17	module	module	NOUN
ajst-23440	45	18	in	in	ADP
ajst-23440	45	19	the	the	DET
ajst-23440	45	20	backbone	backbone	NOUN
ajst-23440	45	21	,	,	PUNCT
ajst-23440	45	22	as	as	SCONJ
ajst-23440	45	23	shown	show	VERB
ajst-23440	45	24	in	in	ADP
ajst-23440	45	25	figure	figure	NOUN
ajst-23440	45	26	3	3	NUM
ajst-23440	45	27	:	:	SYM
ajst-23440	45	28	3.2	3.2	NUM
ajst-23440	45	29	.	.	PUNCT
ajst-23440	46	1	loss	loss	NOUN
ajst-23440	46	2	function	function	NOUN
ajst-23440	46	3	siou	siou	NOUN
ajst-23440	46	4	:	:	PUNCT
ajst-23440	46	5	object	object	NOUN
ajst-23440	46	6	detection	detection	NOUN
ajst-23440	46	7	is	be	AUX
ajst-23440	46	8	one	one	NUM
ajst-23440	46	9	of	of	ADP
ajst-23440	46	10	the	the	DET
ajst-23440	46	11	core	core	NOUN
ajst-23440	46	12	issues	issue	NOUN
ajst-23440	46	13	in	in	ADP
ajst-23440	46	14	computer	computer	NOUN
ajst-23440	46	15	vision	vision	NOUN
ajst-23440	46	16	tasks	task	NOUN
ajst-23440	46	17	,	,	PUNCT
ajst-23440	46	18	and	and	CCONJ
ajst-23440	46	19	its	its	PRON
ajst-23440	46	20	effectiveness	effectiveness	NOUN
ajst-23440	46	21	largely	largely	ADV
ajst-23440	46	22	depends	depend	VERB
ajst-23440	46	23	on	on	ADP
ajst-23440	46	24	the	the	DET
ajst-23440	46	25	definition	definition	NOUN
ajst-23440	46	26	of	of	ADP
ajst-23440	46	27	the	the	DET
ajst-23440	46	28	loss	loss	NOUN
ajst-23440	46	29	function	function	NOUN
ajst-23440	46	30	.	.	PUNCT
ajst-23440	47	1	the	the	DET
ajst-23440	47	2	traditional	traditional	ADJ
ajst-23440	47	3	object	object	NOUN
ajst-23440	47	4	detection	detection	NOUN
ajst-23440	47	5	loss	loss	NOUN
ajst-23440	47	6	function	function	NOUN
ajst-23440	47	7	relies	rely	VERB
ajst-23440	47	8	on	on	ADP
ajst-23440	47	9	the	the	DET
ajst-23440	47	10	aggregation	aggregation	NOUN
ajst-23440	47	11	of	of	ADP
ajst-23440	47	12	bounding	bound	VERB
ajst-23440	47	13	box	box	NOUN
ajst-23440	47	14	regression	regression	NOUN
ajst-23440	47	15	indicators	indicator	NOUN
ajst-23440	47	16	,	,	PUNCT
ajst-23440	47	17	such	such	ADJ
ajst-23440	47	18	as	as	ADP
ajst-23440	47	19	the	the	DET
ajst-23440	47	20	distance	distance	NOUN
ajst-23440	47	21	,	,	PUNCT
ajst-23440	47	22	overlapping	overlap	VERB
ajst-23440	47	23	area	area	NOUN
ajst-23440	47	24	,	,	PUNCT
ajst-23440	47	25	and	and	CCONJ
ajst-23440	47	26	aspect	aspect	VERB
ajst-23440	47	27	ratio	ratio	NOUN
ajst-23440	47	28	between	between	ADP
ajst-23440	47	29	the	the	DET
ajst-23440	47	30	predicted	predict	VERB
ajst-23440	47	31	box	box	NOUN
ajst-23440	47	32	and	and	CCONJ
ajst-23440	47	33	the	the	DET
ajst-23440	47	34	true	true	ADJ
ajst-23440	47	35	box	box	PROPN
ajst-23440	47	36	(	(	PUNCT
ajst-23440	47	37	i.e.	i.e.	X
ajst-23440	47	38	giou	giou	NOUN
ajst-23440	47	39	,	,	PUNCT
ajst-23440	47	40	ciou	ciou	NOUN
ajst-23440	47	41	,	,	PUNCT
ajst-23440	47	42	iciou	iciou	NOUN
ajst-23440	47	43	,	,	PUNCT
ajst-23440	47	44	etc	etc	X
ajst-23440	47	45	.	.	X
ajst-23440	47	46	)	)	PUNCT
ajst-23440	47	47	.	.	PUNCT
ajst-23440	48	1	in	in	ADP
ajst-23440	48	2	this	this	DET
ajst-23440	48	3	article	article	NOUN
ajst-23440	48	4	,	,	PUNCT
ajst-23440	48	5	a	a	DET
ajst-23440	48	6	new	new	ADJ
ajst-23440	48	7	loss	loss	NOUN
ajst-23440	48	8	function	function	NOUN
ajst-23440	48	9	siou	siou	NOUN
ajst-23440	48	10	is	be	AUX
ajst-23440	48	11	proposed	propose	VERB
ajst-23440	48	12	,	,	PUNCT
ajst-23440	48	13	which	which	PRON
ajst-23440	48	14	redefines	redefine	VERB
ajst-23440	48	15	the	the	DET
ajst-23440	48	16	penalty	penalty	NOUN
ajst-23440	48	17	metric	metric	ADJ
ajst-23440	48	18	taking	taking	NOUN
ajst-23440	48	19	into	into	ADP
ajst-23440	48	20	account	account	NOUN
ajst-23440	48	21	the	the	DET
ajst-23440	48	22	vector	vector	NOUN
ajst-23440	48	23	angles	angle	NOUN
ajst-23440	48	24	between	between	ADP
ajst-23440	48	25	the	the	DET
ajst-23440	48	26	required	require	VERB
ajst-23440	48	27	regressions	regression	NOUN
ajst-23440	48	28	.	.	PUNCT
ajst-23440	49	1	siou	siou	NOUN
ajst-23440	49	2	consists	consist	VERB
ajst-23440	49	3	of	of	ADP
ajst-23440	49	4	four	four	NUM
ajst-23440	49	5	parts	part	NOUN
ajst-23440	49	6	,	,	PUNCT
ajst-23440	49	7	namely	namely	ADV
ajst-23440	49	8	angle	angle	NOUN
ajst-23440	49	9	loss	loss	NOUN
ajst-23440	49	10	,	,	PUNCT
ajst-23440	49	11	distance	distance	NOUN
ajst-23440	49	12	loss	loss	NOUN
ajst-23440	49	13	,	,	PUNCT
ajst-23440	49	14	shape	shape	NOUN
ajst-23440	49	15	loss	loss	NOUN
ajst-23440	49	16	,	,	PUNCT
ajst-23440	49	17	and	and	CCONJ
ajst-23440	49	18	iou	iou	VERB
ajst-23440	49	19	loss	loss	NOUN
ajst-23440	49	20	.	.	PUNCT
ajst-23440	50	1	siou	siou	NOUN
ajst-23440	50	2	can	can	AUX
ajst-23440	50	3	effectively	effectively	ADV
ajst-23440	50	4	reduce	reduce	VERB
ajst-23440	50	5	the	the	DET
ajst-23440	50	6	impact	impact	NOUN
ajst-23440	50	7	of	of	ADP
ajst-23440	50	8	errors	error	NOUN
ajst-23440	50	9	on	on	ADP
ajst-23440	50	10	the	the	DET
ajst-23440	50	11	loss	loss	NOUN
ajst-23440	50	12	function	function	NOUN
ajst-23440	50	13	and	and	CCONJ
ajst-23440	50	14	improve	improve	VERB
ajst-23440	50	15	network	network	NOUN
ajst-23440	50	16	robustness	robustness	NOUN
ajst-23440	50	17	.	.	PUNCT
ajst-23440	51	1	when	when	SCONJ
ajst-23440	51	2	detecting	detect	VERB
ajst-23440	51	3	small	small	ADJ
ajst-23440	51	4	targets	target	NOUN
ajst-23440	51	5	,	,	PUNCT
ajst-23440	51	6	the	the	DET
ajst-23440	51	7	estimation	estimation	NOUN
ajst-23440	51	8	method	method	NOUN
ajst-23440	51	9	used	use	VERB
ajst-23440	51	10	by	by	ADP
ajst-23440	51	11	siou	siou	NOUN
ajst-23440	51	12	is	be	AUX
ajst-23440	51	13	weighted	weight	VERB
ajst-23440	51	14	.	.	PUNCT
ajst-23440	52	1	the	the	DET
ajst-23440	52	2	siou	siou	NOUN
ajst-23440	52	3	loss	loss	NOUN
ajst-23440	52	4	function	function	NOUN
ajst-23440	52	5	is	be	AUX
ajst-23440	52	6	used	use	VERB
ajst-23440	52	7	to	to	PART
ajst-23440	52	8	improve	improve	VERB
ajst-23440	52	9	the	the	DET
ajst-23440	52	10	convergence	convergence	NOUN
ajst-23440	52	11	speed	speed	NOUN
ajst-23440	52	12	and	and	CCONJ
ajst-23440	52	13	detection	detection	NOUN
ajst-23440	52	14	accuracy	accuracy	NOUN
ajst-23440	52	15	of	of	ADP
ajst-23440	52	16	the	the	DET
ajst-23440	52	17	model	model	NOUN
ajst-23440	53	1	[	[	X
ajst-23440	53	2	6	6	NUM
ajst-23440	53	3	]	]	PUNCT
ajst-23440	53	4	.	.	PUNCT
ajst-23440	54	1	applying	apply	VERB
ajst-23440	54	2	the	the	DET
ajst-23440	54	3	siou	siou	NOUN
ajst-23440	54	4	loss	loss	NOUN
ajst-23440	54	5	function	function	NOUN
ajst-23440	54	6	to	to	ADP
ajst-23440	54	7	the	the	DET
ajst-23440	54	8	algorithm	algorithm	NOUN
ajst-23440	54	9	model	model	NOUN
ajst-23440	54	10	for	for	ADP
ajst-23440	54	11	detecting	detect	VERB
ajst-23440	54	12	controlled	control	VERB
ajst-23440	54	13	tools	tool	NOUN
ajst-23440	54	14	can	can	AUX
ajst-23440	54	15	also	also	ADV
ajst-23440	54	16	improve	improve	VERB
ajst-23440	54	17	the	the	DET
ajst-23440	54	18	recognition	recognition	NOUN
ajst-23440	54	19	efficiency	efficiency	NOUN
ajst-23440	54	20	of	of	ADP
ajst-23440	54	21	the	the	DET
ajst-23440	54	22	model	model	NOUN
ajst-23440	54	23	.	.	PUNCT
ajst-23440	55	1	the	the	DET
ajst-23440	55	2	siou	siou	NOUN
ajst-23440	55	3	loss	loss	NOUN
ajst-23440	55	4	function	function	NOUN
ajst-23440	55	5	formula	formula	NOUN
ajst-23440	55	6	is	be	AUX
ajst-23440	55	7	shown	show	VERB
ajst-23440	55	8	in	in	ADP
ajst-23440	55	9	(	(	PUNCT
ajst-23440	55	10	1	1	NUM
ajst-23440	55	11	)	)	PUNCT
ajst-23440	55	12	,	,	PUNCT
ajst-23440	55	13	(	(	PUNCT
ajst-23440	55	14	2	2	NUM
ajst-23440	55	15	)	)	PUNCT
ajst-23440	55	16	,	,	PUNCT
ajst-23440	55	17	(	(	PUNCT
ajst-23440	55	18	3	3	NUM
ajst-23440	55	19	)	)	PUNCT
ajst-23440	55	20	,	,	PUNCT
ajst-23440	55	21	and	and	CCONJ
ajst-23440	55	22	(	(	PUNCT
ajst-23440	55	23	4	4	NUM
ajst-23440	55	24	):	):	PUNCT
ajst-23440	55	25	hh	hh	PROPN
ajst-23440	55	26	'	'	PART
ajst-23440	55	27	x	x	SYM
ajst-23440	55	28	'	'	PUNCT
ajst-23440	55	29	x	x	PUNCT
ajst-23440	55	30	c	c	X
ajst-23440	55	31	'	'	PUNCT
ajst-23440	55	32	c	c	NOUN
ajst-23440	55	33	w	w	NOUN
ajst-23440	55	34	'	'	PUNCT
ajst-23440	55	35	w	w	PROPN
ajst-23440	55	36	ftr	ftr	PROPN
ajst-23440	55	37	fscale(ꞏ,ꞏ	fscale(ꞏ,ꞏ	PROPN
ajst-23440	55	38	)	)	PUNCT
ajst-23440	55	39	h	h	NOUN
ajst-23440	55	40	c	c	NOUN
ajst-23440	55	41	w	w	PROPN
ajst-23440	55	42	fex(ꞏ,w	fex(ꞏ,w	PROPN
ajst-23440	55	43	)	)	PUNCT
ajst-23440	55	44	fsq(ꞏ	fsq(ꞏ	PROPN
ajst-23440	55	45	)	)	PUNCT
ajst-23440	55	46	1×1×c	1×1×c	NUM
ajst-23440	55	47	1×1×c	1×1×c	NUM
ajst-23440	55	48	x	x	NOUN
ajst-23440	55	49	figure	figure	NOUN
ajst-23440	55	50	2	2	NUM
ajst-23440	55	51	.	.	PUNCT
ajst-23440	56	1	the	the	DET
ajst-23440	56	2	structural	structural	ADJ
ajst-23440	56	3	diagram	diagram	NOUN
ajst-23440	56	4	of	of	ADP
ajst-23440	56	5	seattention	seattention	PROPN
ajst-23440	56	6	conv	conv	PROPN
ajst-23440	56	7	conv	conv	PROPN
ajst-23440	56	8	c2f	c2f	PROPN
ajst-23440	56	9	conv	conv	PROPN
ajst-23440	56	10	c2f	c2f	PROPN
ajst-23440	56	11	convc2fconvc2fsppf	convc2fconvc2fsppf	PROPN
ajst-23440	56	12	input	input	NOUN
ajst-23440	56	13	seattention	seattention	NOUN
ajst-23440	56	14	figure	figure	NOUN
ajst-23440	56	15	3	3	NUM
ajst-23440	56	16	.	.	NOUN
ajst-23440	56	17	place	place	VERB
ajst-23440	56	18	the	the	DET
ajst-23440	56	19	seattention	seattention	NOUN
ajst-23440	56	20	attention	attention	NOUN
ajst-23440	56	21	mechanism	mechanism	NOUN
ajst-23440	56	22	module	module	NOUN
ajst-23440	56	23	in	in	ADP
ajst-23440	56	24	the	the	DET
ajst-23440	56	25	backbone	backbone	NOUN
ajst-23440	56	26	the	the	DET
ajst-23440	56	27	calculation	calculation	NOUN
ajst-23440	56	28	formula	formula	NOUN
ajst-23440	56	29	for	for	ADP
ajst-23440	56	30	angle	angle	NOUN
ajst-23440	56	31	cost	cost	NOUN
ajst-23440	56	32	is	be	AUX
ajst-23440	56	33	:	:	PUNCT
ajst-23440	56	34	1	1	NUM
ajst-23440	56	35	2	2	NUM
ajst-23440	56	36	²	²	NUM
ajst-23440	56	37	(	(	PUNCT
ajst-23440	56	38	1	1	X
ajst-23440	56	39	)	)	PUNCT
ajst-23440	56	40	the	the	DET
ajst-23440	56	41	definition	definition	NOUN
ajst-23440	56	42	of	of	ADP
ajst-23440	56	43	distance	distance	NOUN
ajst-23440	56	44	cost	cost	NOUN
ajst-23440	56	45	is	be	AUX
ajst-23440	56	46	:	:	PUNCT
ajst-23440	56	47	∑	∑	PROPN
ajst-23440	56	48	1	1	NUM
ajst-23440	56	49	,	,	PUNCT
ajst-23440	56	50	(	(	PUNCT
ajst-23440	56	51	2	2	X
ajst-23440	56	52	)	)	PUNCT
ajst-23440	56	53	the	the	DET
ajst-23440	56	54	definition	definition	NOUN
ajst-23440	56	55	of	of	ADP
ajst-23440	56	56	shape	shape	NOUN
ajst-23440	56	57	cost	cost	NOUN
ajst-23440	56	58	is	be	AUX
ajst-23440	56	59	:	:	PUNCT
ajst-23440	56	60	∑	∑	ADV
ajst-23440	56	61	,	,	PUNCT
ajst-23440	56	62	1	1	NUM
ajst-23440	56	63	(	(	PUNCT
ajst-23440	56	64	3	3	NUM
ajst-23440	56	65	)	)	PUNCT
ajst-23440	56	66	the	the	DET
ajst-23440	56	67	definition	definition	NOUN
ajst-23440	56	68	of	of	ADP
ajst-23440	56	69	the	the	DET
ajst-23440	56	70	siou	siou	NOUN
ajst-23440	56	71	loss	loss	NOUN
ajst-23440	56	72	function	function	NOUN
ajst-23440	56	73	is	be	AUX
ajst-23440	56	74	:	:	PUNCT
ajst-23440	56	75	47	47	NUM
ajst-23440	56	76	_	_	NOUN
ajst-23440	56	77	1	1	NUM
ajst-23440	56	78	(	(	PUNCT
ajst-23440	56	79	4	4	NUM
ajst-23440	56	80	)	)	PUNCT
ajst-23440	56	81	δ	δ	PROPN
ajst-23440	56	82	:	:	PUNCT
ajst-23440	56	83	distance	distance	NOUN
ajst-23440	56	84	loss	loss	NOUN
ajst-23440	56	85	,	,	PUNCT
ajst-23440	56	86	ω	ω	NUM
ajst-23440	56	87	:	:	PUNCT
ajst-23440	56	88	shape	shape	NOUN
ajst-23440	56	89	loss	loss	NOUN
ajst-23440	56	90	,	,	PUNCT
ajst-23440	56	91	θ	θ	NOUN
ajst-23440	56	92	:	:	PUNCT
ajst-23440	56	93	the	the	DET
ajst-23440	56	94	degree	degree	NOUN
ajst-23440	56	95	of	of	ADP
ajst-23440	56	96	attention	attention	NOUN
ajst-23440	56	97	paid	pay	VERB
ajst-23440	56	98	to	to	PART
ajst-23440	56	99	shape	shape	VERB
ajst-23440	56	100	loss	loss	NOUN
ajst-23440	56	101	4	4	NUM
ajst-23440	56	102	.	.	PUNCT
ajst-23440	57	1	experimental	experimental	ADJ
ajst-23440	57	2	results	result	NOUN
ajst-23440	57	3	and	and	CCONJ
ajst-23440	57	4	analysis	analysis	NOUN
ajst-23440	57	5	4.1	4.1	NUM
ajst-23440	57	6	.	.	PUNCT
ajst-23440	58	1	dataset	dataset	NOUN
ajst-23440	58	2	processing	processing	NOUN
ajst-23440	58	3	:	:	PUNCT
ajst-23440	58	4	the	the	DET
ajst-23440	58	5	dataset	dataset	NOUN
ajst-23440	58	6	used	use	VERB
ajst-23440	58	7	in	in	ADP
ajst-23440	58	8	this	this	DET
ajst-23440	58	9	experiment	experiment	NOUN
ajst-23440	58	10	is	be	AUX
ajst-23440	58	11	partly	partly	ADV
ajst-23440	58	12	from	from	ADP
ajst-23440	58	13	video	video	NOUN
ajst-23440	58	14	screenshots	screenshot	NOUN
ajst-23440	58	15	of	of	ADP
ajst-23440	58	16	controlled	control	VERB
ajst-23440	58	17	cutting	cutting	NOUN
ajst-23440	58	18	tools	tool	NOUN
ajst-23440	58	19	,	,	PUNCT
ajst-23440	58	20	and	and	CCONJ
ajst-23440	58	21	partly	partly	ADV
ajst-23440	58	22	from	from	ADP
ajst-23440	58	23	controlled	control	VERB
ajst-23440	58	24	cutting	cutting	NOUN
ajst-23440	58	25	tool	tool	NOUN
ajst-23440	58	26	datasets	dataset	NOUN
ajst-23440	58	27	sold	sell	VERB
ajst-23440	58	28	on	on	ADP
ajst-23440	58	29	the	the	DET
ajst-23440	58	30	internet	internet	NOUN
ajst-23440	58	31	.	.	PUNCT
ajst-23440	59	1	clear	clear	ADJ
ajst-23440	59	2	or	or	CCONJ
ajst-23440	59	3	complex	complex	ADJ
ajst-23440	59	4	scene	scene	NOUN
ajst-23440	59	5	controlled	control	VERB
ajst-23440	59	6	cutting	cut	VERB
ajst-23440	59	7	tool	tool	NOUN
ajst-23440	59	8	images	image	NOUN
ajst-23440	59	9	are	be	AUX
ajst-23440	59	10	selected	select	VERB
ajst-23440	59	11	from	from	ADP
ajst-23440	59	12	them	they	PRON
ajst-23440	59	13	and	and	CCONJ
ajst-23440	59	14	expanded	expand	VERB
ajst-23440	59	15	to	to	ADP
ajst-23440	59	16	600	600	NUM
ajst-23440	59	17	datasets	dataset	NOUN
ajst-23440	59	18	through	through	ADP
ajst-23440	59	19	data	datum	NOUN
ajst-23440	59	20	augmentation	augmentation	NOUN
ajst-23440	59	21	methods	method	NOUN
ajst-23440	59	22	(	(	PUNCT
ajst-23440	59	23	such	such	ADJ
ajst-23440	59	24	as	as	ADP
ajst-23440	59	25	increasing	increase	VERB
ajst-23440	59	26	brightness	brightness	NOUN
ajst-23440	59	27	,	,	PUNCT
ajst-23440	59	28	flipping	flipping	NOUN
ajst-23440	59	29	,	,	PUNCT
ajst-23440	59	30	etc	etc	X
ajst-23440	59	31	.	.	X
ajst-23440	59	32	)	)	PUNCT
ajst-23440	59	33	.	.	PUNCT
ajst-23440	60	1	randomly	randomly	ADV
ajst-23440	60	2	divide	divide	VERB
ajst-23440	60	3	the	the	DET
ajst-23440	60	4	images	image	NOUN
ajst-23440	60	5	into	into	ADP
ajst-23440	60	6	480	480	NUM
ajst-23440	60	7	training	training	NOUN
ajst-23440	60	8	set	set	NOUN
ajst-23440	60	9	images	image	NOUN
ajst-23440	60	10	and	and	CCONJ
ajst-23440	60	11	120	120	NUM
ajst-23440	60	12	validation	validation	NOUN
ajst-23440	60	13	set	set	NOUN
ajst-23440	60	14	images	image	NOUN
ajst-23440	60	15	in	in	ADP
ajst-23440	60	16	an	an	DET
ajst-23440	60	17	8:2	8:2	NUM
ajst-23440	60	18	ratio	ratio	NOUN
ajst-23440	60	19	.	.	PUNCT
ajst-23440	61	1	use	use	VERB
ajst-23440	61	2	labelimg	labelimg	NOUN
ajst-23440	61	3	to	to	PART
ajst-23440	61	4	frame	frame	VERB
ajst-23440	61	5	and	and	CCONJ
ajst-23440	61	6	label	label	VERB
ajst-23440	61	7	the	the	DET
ajst-23440	61	8	controlled	control	VERB
ajst-23440	61	9	tools	tool	NOUN
ajst-23440	61	10	in	in	ADP
ajst-23440	61	11	the	the	DET
ajst-23440	61	12	images	image	NOUN
ajst-23440	61	13	with	with	ADP
ajst-23440	61	14	the	the	DET
ajst-23440	61	15	keyword	keyword	NOUN
ajst-23440	61	16	"	"	PUNCT
ajst-23440	61	17	knife	knife	NOUN
ajst-23440	61	18	"	"	PUNCT
ajst-23440	61	19	.	.	PUNCT
ajst-23440	62	1	4.2	4.2	NUM
ajst-23440	62	2	.	.	PUNCT
ajst-23440	63	1	experimental	experimental	ADJ
ajst-23440	63	2	environment	environment	NOUN
ajst-23440	63	3	:	:	PUNCT
ajst-23440	63	4	after	after	ADP
ajst-23440	63	5	preparing	prepare	VERB
ajst-23440	63	6	the	the	DET
ajst-23440	63	7	dataset	dataset	NOUN
ajst-23440	63	8	,	,	PUNCT
ajst-23440	63	9	a	a	DET
ajst-23440	63	10	remote	remote	ADJ
ajst-23440	63	11	server	server	NOUN
ajst-23440	63	12	with	with	ADP
ajst-23440	63	13	rtx	rtx	PROPN
ajst-23440	63	14	4090	4090	NUM
ajst-23440	63	15	,	,	PUNCT
ajst-23440	63	16	pytorch	pytorch	NOUN
ajst-23440	63	17	1.11.0	1.11.0	NUM
ajst-23440	63	18	,	,	PUNCT
ajst-23440	63	19	cuda	cuda	NOUN
ajst-23440	63	20	11.3	11.3	NUM
ajst-23440	63	21	graphics	graphic	NOUN
ajst-23440	63	22	card	card	NOUN
ajst-23440	63	23	,	,	PUNCT
ajst-23440	63	24	programming	programming	NOUN
ajst-23440	63	25	language	language	NOUN
ajst-23440	63	26	in	in	ADP
ajst-23440	63	27	python	python	PROPN
ajst-23440	63	28	,	,	PUNCT
ajst-23440	63	29	and	and	CCONJ
ajst-23440	63	30	rtx	rtx	PROPN
ajst-23440	63	31	4090	4090	NUM
ajst-23440	63	32	(	(	PUNCT
ajst-23440	63	33	24	24	NUM
ajst-23440	63	34	gb	gb	NOUN
ajst-23440	63	35	)	)	PUNCT
ajst-23440	63	36	*	*	PUNCT
ajst-23440	63	37	1	1	NUM
ajst-23440	63	38	gpu	gpu	NOUN
ajst-23440	63	39	was	be	AUX
ajst-23440	63	40	created	create	VERB
ajst-23440	63	41	on	on	ADP
ajst-23440	63	42	autodl	autodl	ADJ
ajst-23440	63	43	,	,	PUNCT
ajst-23440	63	44	and	and	CCONJ
ajst-23440	63	45	a	a	DET
ajst-23440	63	46	yolov8	yolov8	NOUN
ajst-23440	63	47	gpu	gpu	PROPN
ajst-23440	63	48	runtime	runtime	NOUN
ajst-23440	63	49	environment	environment	NOUN
ajst-23440	63	50	was	be	AUX
ajst-23440	63	51	created	create	VERB
ajst-23440	63	52	on	on	ADP
ajst-23440	63	53	the	the	DET
ajst-23440	63	54	server	server	NOUN
ajst-23440	63	55	.	.	PUNCT
ajst-23440	64	1	4.3	4.3	NUM
ajst-23440	64	2	.	.	PUNCT
ajst-23440	65	1	evaluation	evaluation	NOUN
ajst-23440	65	2	criteria	criterion	NOUN
ajst-23440	65	3	:	:	PUNCT
ajst-23440	65	4	the	the	DET
ajst-23440	65	5	indicator	indicator	NOUN
ajst-23440	65	6	used	use	VERB
ajst-23440	65	7	to	to	PART
ajst-23440	65	8	evaluate	evaluate	VERB
ajst-23440	65	9	the	the	DET
ajst-23440	65	10	improvement	improvement	NOUN
ajst-23440	65	11	effect	effect	NOUN
ajst-23440	65	12	of	of	ADP
ajst-23440	65	13	the	the	DET
ajst-23440	65	14	algorithm	algorithm	NOUN
ajst-23440	65	15	in	in	ADP
ajst-23440	65	16	this	this	DET
ajst-23440	65	17	article	article	NOUN
ajst-23440	65	18	is	be	AUX
ajst-23440	65	19	the	the	DET
ajst-23440	65	20	average	average	ADJ
ajst-23440	65	21	accuracy	accuracy	NOUN
ajst-23440	65	22	map	map	NOUN
ajst-23440	65	23	,	,	PUNCT
ajst-23440	65	24	as	as	SCONJ
ajst-23440	65	25	shown	show	VERB
ajst-23440	65	26	in	in	ADP
ajst-23440	65	27	(	(	PUNCT
ajst-23440	65	28	5	5	NUM
ajst-23440	65	29	):	):	PUNCT
ajst-23440	65	30	map=∑ap	map=∑ap	PROPN
ajst-23440	65	31	/	/	SYM
ajst-23440	65	32	n	n	CCONJ
ajst-23440	65	33	(	(	PUNCT
ajst-23440	65	34	5	5	NUM
ajst-23440	65	35	)	)	PUNCT
ajst-23440	65	36	map	map	NOUN
ajst-23440	65	37	stands	stand	VERB
ajst-23440	65	38	for	for	ADP
ajst-23440	65	39	mean	mean	ADJ
ajst-23440	65	40	average	average	ADJ
ajst-23440	65	41	accuracy	accuracy	NOUN
ajst-23440	65	42	,	,	PUNCT
ajst-23440	65	43	which	which	PRON
ajst-23440	65	44	generally	generally	ADV
ajst-23440	65	45	refers	refer	VERB
ajst-23440	65	46	to	to	ADP
ajst-23440	65	47	the	the	DET
ajst-23440	65	48	average	average	ADJ
ajst-23440	65	49	value	value	NOUN
ajst-23440	65	50	of	of	ADP
ajst-23440	65	51	ap	ap	PROPN
ajst-23440	65	52	for	for	ADP
ajst-23440	65	53	all	all	DET
ajst-23440	65	54	categories	category	NOUN
ajst-23440	65	55	within	within	ADP
ajst-23440	65	56	all	all	DET
ajst-23440	65	57	images	image	NOUN
ajst-23440	65	58	.	.	PUNCT
ajst-23440	66	1	the	the	DET
ajst-23440	66	2	p	p	PROPN
ajst-23440	66	3	-	-	PUNCT
ajst-23440	66	4	r	r	NOUN
ajst-23440	66	5	curve	curve	NOUN
ajst-23440	66	6	is	be	AUX
ajst-23440	66	7	a	a	DET
ajst-23440	66	8	curve	curve	NOUN
ajst-23440	66	9	enclosed	enclose	VERB
ajst-23440	66	10	by	by	ADP
ajst-23440	66	11	precision	precision	NOUN
ajst-23440	66	12	and	and	CCONJ
ajst-23440	66	13	recall	recall	PROPN
ajst-23440	66	14	coordinates	coordinate	NOUN
ajst-23440	66	15	,	,	PUNCT
ajst-23440	66	16	respectively	respectively	ADV
ajst-23440	66	17	.	.	PUNCT
ajst-23440	67	1	by	by	ADP
ajst-23440	67	2	comparing	compare	VERB
ajst-23440	67	3	the	the	DET
ajst-23440	67	4	pr	pr	NOUN
ajst-23440	67	5	curve	curve	NOUN
ajst-23440	67	6	and	and	CCONJ
ajst-23440	67	7	map	map	NOUN
ajst-23440	67	8	obtained	obtain	VERB
ajst-23440	67	9	from	from	ADP
ajst-23440	67	10	the	the	DET
ajst-23440	67	11	experiment	experiment	NOUN
ajst-23440	67	12	,	,	PUNCT
ajst-23440	67	13	the	the	DET
ajst-23440	67	14	efficiency	efficiency	NOUN
ajst-23440	67	15	of	of	ADP
ajst-23440	67	16	the	the	DET
ajst-23440	67	17	improved	improve	VERB
ajst-23440	67	18	algorithm	algorithm	NOUN
ajst-23440	67	19	model	model	NOUN
ajst-23440	67	20	can	can	AUX
ajst-23440	67	21	be	be	AUX
ajst-23440	67	22	determined	determine	VERB
ajst-23440	67	23	.	.	PUNCT
ajst-23440	68	1	4.4	4.4	NUM
ajst-23440	68	2	.	.	PUNCT
ajst-23440	69	1	result	result	VERB
ajst-23440	69	2	analysis	analysis	NOUN
ajst-23440	69	3	:	:	PUNCT
ajst-23440	69	4	after	after	ADP
ajst-23440	69	5	200	200	NUM
ajst-23440	69	6	rounds	round	NOUN
ajst-23440	69	7	,	,	PUNCT
ajst-23440	69	8	the	the	DET
ajst-23440	69	9	map	map	NOUN
ajst-23440	69	10	of	of	ADP
ajst-23440	69	11	the	the	DET
ajst-23440	69	12	original	original	ADJ
ajst-23440	69	13	model	model	NOUN
ajst-23440	69	14	is	be	AUX
ajst-23440	69	15	shown	show	VERB
ajst-23440	69	16	in	in	ADP
ajst-23440	69	17	figure	figure	NOUN
ajst-23440	69	18	4	4	NUM
ajst-23440	69	19	,	,	PUNCT
ajst-23440	69	20	and	and	CCONJ
ajst-23440	69	21	the	the	DET
ajst-23440	69	22	map	map	NOUN
ajst-23440	69	23	of	of	ADP
ajst-23440	69	24	the	the	DET
ajst-23440	69	25	improved	improved	ADJ
ajst-23440	69	26	model	model	NOUN
ajst-23440	69	27	is	be	AUX
ajst-23440	69	28	shown	show	VERB
ajst-23440	69	29	in	in	ADP
ajst-23440	69	30	figure	figure	NOUN
ajst-23440	69	31	5	5	NUM
ajst-23440	69	32	:	:	PUNCT
ajst-23440	69	33	figure	figure	NOUN
ajst-23440	69	34	4	4	NUM
ajst-23440	69	35	.	.	PUNCT
ajst-23440	70	1	the	the	DET
ajst-23440	70	2	map	map	NOUN
ajst-23440	70	3	of	of	ADP
ajst-23440	70	4	the	the	DET
ajst-23440	70	5	original	original	ADJ
ajst-23440	70	6	model	model	NOUN
ajst-23440	70	7	figure	figure	NOUN
ajst-23440	70	8	5	5	NUM
ajst-23440	70	9	.	.	PUNCT
ajst-23440	71	1	the	the	DET
ajst-23440	71	2	map	map	NOUN
ajst-23440	71	3	of	of	ADP
ajst-23440	71	4	the	the	DET
ajst-23440	71	5	improved	improved	ADJ
ajst-23440	71	6	model	model	NOUN
ajst-23440	71	7	the	the	DET
ajst-23440	71	8	improved	improved	ADJ
ajst-23440	71	9	yolov8	yolov8	NOUN
ajst-23440	71	10	model	model	NOUN
ajst-23440	71	11	has	have	AUX
ajst-23440	71	12	been	be	AUX
ajst-23440	71	13	trained	train	VERB
ajst-23440	71	14	a	a	DET
ajst-23440	71	15	total	total	NOUN
ajst-23440	71	16	of	of	ADP
ajst-23440	71	17	200	200	NUM
ajst-23440	71	18	rounds	round	NOUN
ajst-23440	71	19	,	,	PUNCT
ajst-23440	71	20	and	and	CCONJ
ajst-23440	71	21	the	the	DET
ajst-23440	71	22	map	map	NOUN
ajst-23440	71	23	has	have	AUX
ajst-23440	71	24	increased	increase	VERB
ajst-23440	71	25	from	from	ADP
ajst-23440	71	26	the	the	DET
ajst-23440	71	27	initial	initial	ADJ
ajst-23440	71	28	80.8	80.8	NUM
ajst-23440	71	29	%	%	NOUN
ajst-23440	71	30	to	to	ADP
ajst-23440	71	31	the	the	DET
ajst-23440	71	32	improved	improved	ADJ
ajst-23440	71	33	85	85	NUM
ajst-23440	71	34	%	%	NOUN
ajst-23440	71	35	,	,	PUNCT
ajst-23440	71	36	an	an	DET
ajst-23440	71	37	increase	increase	NOUN
ajst-23440	71	38	of	of	ADP
ajst-23440	71	39	4.2	4.2	NUM
ajst-23440	71	40	%	%	NOUN
ajst-23440	71	41	.	.	PUNCT
ajst-23440	72	1	therefore	therefore	ADV
ajst-23440	72	2	,	,	PUNCT
ajst-23440	72	3	experimental	experimental	ADJ
ajst-23440	72	4	results	result	NOUN
ajst-23440	72	5	have	have	AUX
ajst-23440	72	6	shown	show	VERB
ajst-23440	72	7	that	that	SCONJ
ajst-23440	72	8	the	the	DET
ajst-23440	72	9	improved	improved	ADJ
ajst-23440	72	10	algorithm	algorithm	NOUN
ajst-23440	72	11	performs	perform	VERB
ajst-23440	72	12	well	well	ADV
ajst-23440	72	13	in	in	ADP
ajst-23440	72	14	detecting	detect	VERB
ajst-23440	72	15	controlled	control	VERB
ajst-23440	72	16	cutting	cutting	NOUN
ajst-23440	72	17	tools	tool	NOUN
ajst-23440	72	18	.	.	PUNCT
ajst-23440	73	1	5	5	X
ajst-23440	73	2	.	.	X
ajst-23440	73	3	summary	summary	NOUN
ajst-23440	73	4	this	this	DET
ajst-23440	73	5	article	article	NOUN
ajst-23440	73	6	proposes	propose	VERB
ajst-23440	73	7	a	a	DET
ajst-23440	73	8	controlled	control	VERB
ajst-23440	73	9	tool	tool	NOUN
ajst-23440	73	10	detection	detection	NOUN
ajst-23440	73	11	algorithm	algorithm	NOUN
ajst-23440	73	12	based	base	VERB
ajst-23440	73	13	on	on	ADP
ajst-23440	73	14	the	the	DET
ajst-23440	73	15	improved	improved	ADJ
ajst-23440	73	16	yolov8	yolov8	NOUN
ajst-23440	73	17	model	model	NOUN
ajst-23440	73	18	.	.	PUNCT
ajst-23440	74	1	after	after	ADP
ajst-23440	74	2	multiple	multiple	ADJ
ajst-23440	74	3	experiments	experiment	NOUN
ajst-23440	74	4	,	,	PUNCT
ajst-23440	74	5	se	se	X
ajst-23440	74	6	attention	attention	NOUN
ajst-23440	74	7	mechanism	mechanism	NOUN
ajst-23440	74	8	was	be	AUX
ajst-23440	74	9	added	add	VERB
ajst-23440	74	10	to	to	ADP
ajst-23440	74	11	the	the	DET
ajst-23440	74	12	original	original	ADJ
ajst-23440	74	13	yolov8	yolov8	NOUN
ajst-23440	74	14	algorithm	algorithm	PROPN
ajst-23440	74	15	model	model	NOUN
ajst-23440	74	16	,	,	PUNCT
ajst-23440	74	17	and	and	CCONJ
ajst-23440	74	18	the	the	DET
ajst-23440	74	19	loss	loss	NOUN
ajst-23440	74	20	function	function	NOUN
ajst-23440	74	21	siou	siou	NOUN
ajst-23440	74	22	was	be	AUX
ajst-23440	74	23	introduced	introduce	VERB
ajst-23440	74	24	into	into	ADP
ajst-23440	74	25	the	the	DET
ajst-23440	74	26	yolov8	yolov8	NOUN
ajst-23440	74	27	algorithm	algorithm	PROPN
ajst-23440	74	28	.	.	PUNCT
ajst-23440	75	1	the	the	DET
ajst-23440	75	2	improved	improved	ADJ
ajst-23440	75	3	map	map	NOUN
ajst-23440	75	4	increased	increase	VERB
ajst-23440	75	5	from	from	ADP
ajst-23440	75	6	the	the	DET
ajst-23440	75	7	initial	initial	ADJ
ajst-23440	75	8	80.8	80.8	NUM
ajst-23440	75	9	%	%	NOUN
ajst-23440	75	10	to	to	ADP
ajst-23440	75	11	the	the	DET
ajst-23440	75	12	improved	improved	ADJ
ajst-23440	75	13	85	85	NUM
ajst-23440	75	14	%	%	NOUN
ajst-23440	75	15	,	,	PUNCT
ajst-23440	75	16	an	an	DET
ajst-23440	75	17	increase	increase	NOUN
ajst-23440	75	18	of	of	ADP
ajst-23440	75	19	4.2	4.2	NUM
ajst-23440	75	20	%	%	NOUN
ajst-23440	75	21	.	.	PUNCT
ajst-23440	76	1	this	this	PRON
ajst-23440	76	2	provides	provide	VERB
ajst-23440	76	3	a	a	DET
ajst-23440	76	4	certain	certain	ADJ
ajst-23440	76	5	reference	reference	NOUN
ajst-23440	76	6	value	value	NOUN
ajst-23440	76	7	for	for	ADP
ajst-23440	76	8	future	future	ADJ
ajst-23440	76	9	algorithm	algorithm	NOUN
ajst-23440	76	10	models	model	NOUN
ajst-23440	76	11	for	for	ADP
ajst-23440	76	12	identifying	identify	VERB
ajst-23440	76	13	controlled	control	VERB
ajst-23440	76	14	tools	tool	NOUN
ajst-23440	76	15	.	.	PUNCT
ajst-23440	77	1	references	reference	NOUN
ajst-23440	77	2	[	[	X
ajst-23440	77	3	1	1	NUM
ajst-23440	77	4	]	]	X
ajst-23440	77	5	sun	sun	PROPN
ajst-23440	77	6	y	y	PROPN
ajst-23440	77	7	,	,	PUNCT
ajst-23440	77	8	sun	sun	PROPN
ajst-23440	77	9	z	z	PROPN
ajst-23440	77	10	,	,	PUNCT
ajst-23440	77	11	chen	chen	PROPN
ajst-23440	77	12	w	w	PROPN
ajst-23440	77	13	.	.	PUNCT
ajst-23440	78	1	the	the	DET
ajst-23440	78	2	evolution	evolution	NOUN
ajst-23440	78	3	of	of	ADP
ajst-23440	78	4	object	object	NOUN
ajst-23440	78	5	detection	detection	NOUN
ajst-23440	78	6	methods	method	NOUN
ajst-23440	78	7	[	[	X
ajst-23440	78	8	j	j	X
ajst-23440	78	9	]	]	X
ajst-23440	78	10	.	.	PUNCT
ajst-23440	79	1	engineering	engineering	NOUN
ajst-23440	79	2	applications	application	NOUN
ajst-23440	79	3	of	of	ADP
ajst-23440	79	4	artificial	artificial	ADJ
ajst-23440	79	5	intelligence	intelligence	NOUN
ajst-23440	79	6	,	,	PUNCT
ajst-23440	79	7	2024	2024	NUM
ajst-23440	79	8	,	,	PUNCT
ajst-23440	79	9	133(pe):108458-	133(pe):108458-	NUM
ajst-23440	79	10	.	.	PUNCT
ajst-23440	80	1	[	[	X
ajst-23440	80	2	2	2	X
ajst-23440	80	3	]	]	X
ajst-23440	80	4	liu	liu	PROPN
ajst-23440	80	5	y	y	PROPN
ajst-23440	80	6	,	,	PUNCT
ajst-23440	80	7	chen	chen	PROPN
ajst-23440	80	8	h	h	PROPN
ajst-23440	80	9	.	.	PUNCT
ajst-23440	81	1	a	a	DET
ajst-23440	81	2	lightweight	lightweight	ADJ
ajst-23440	81	3	container	container	NOUN
ajst-23440	81	4	detection	detection	NOUN
ajst-23440	81	5	network	network	NOUN
ajst-23440	81	6	model	model	NOUN
ajst-23440	81	7	based	base	VERB
ajst-23440	81	8	on	on	ADP
ajst-23440	81	9	yolov5s[c]//	yolov5s[c]//	PROPN
ajst-23440	81	10	(	(	PUNCT
ajst-23440	81	11	technical	technical	ADJ
ajst-23440	81	12	committee	committee	NOUN
ajst-23440	81	13	on	on	ADP
ajst-23440	81	14	control	control	PROPN
ajst-23440	81	15	theory	theory	NOUN
ajst-23440	81	16	,	,	PUNCT
ajst-23440	81	17	chinese	chinese	ADJ
ajst-23440	81	18	association	association	NOUN
ajst-23440	81	19	of	of	ADP
ajst-23440	81	20	automation	automation	NOUN
ajst-23440	81	21	)	)	PUNCT
ajst-23440	81	22	,	,	PUNCT
ajst-23440	81	23	(	(	PUNCT
ajst-23440	81	24	chinese	chinese	PROPN
ajst-23440	81	25	association	association	NOUN
ajst-23440	81	26	of	of	ADP
ajst-23440	81	27	automation	automation	NOUN
ajst-23440	81	28	)	)	PUNCT
ajst-23440	81	29	,	,	PUNCT
ajst-23440	81	30	(	(	PUNCT
ajst-23440	81	31	systems	system	NOUN
ajst-23440	81	32	engineering	engineering	NOUN
ajst-23440	81	33	society	society	NOUN
ajst-23440	81	34	of	of	ADP
ajst-23440	81	35	china	china	PROPN
ajst-23440	81	36	)	)	PUNCT
ajst-23440	81	37	.	.	PUNCT
ajst-23440	82	1	proceedings	proceeding	NOUN
ajst-23440	82	2	of	of	ADP
ajst-23440	82	3	the	the	DET
ajst-23440	82	4	42nd	42nd	PROPN
ajst-23440	82	5	chinese	chinese	PROPN
ajst-23440	82	6	control	control	PROPN
ajst-23440	82	7	conference	conference	PROPN
ajst-23440	82	8	(	(	PUNCT
ajst-23440	82	9	8)	8)	NUM
ajst-23440	82	10	.	.	PUNCT
ajst-23440	83	1	school	school	NOUN
ajst-23440	83	2	of	of	ADP
ajst-23440	83	3	artificial	artificial	ADJ
ajst-23440	83	4	intelligence	intelligence	NOUN
ajst-23440	83	5	,	,	PUNCT
ajst-23440	83	6	hebei	hebei	PROPN
ajst-23440	83	7	university	university	PROPN
ajst-23440	83	8	of	of	ADP
ajst-23440	83	9	technology	technology	NOUN
ajst-23440	83	10	;	;	PUNCT
ajst-23440	83	11	control	control	NOUN
ajst-23440	83	12	engineering	engineering	NOUN
ajst-23440	83	13	technology	technology	NOUN
ajst-23440	83	14	innovation	innovation	NOUN
ajst-23440	83	15	center	center	NOUN
ajst-23440	83	16	of	of	ADP
ajst-23440	83	17	hebei	hebei	PROPN
ajst-23440	83	18	province	province	PROPN
ajst-23440	83	19	,	,	PUNCT
ajst-23440	83	20	hebei	hebei	PROPN
ajst-23440	83	21	university	university	PROPN
ajst-23440	83	22	of	of	ADP
ajst-23440	83	23	technology	technology	PROPN
ajst-23440	83	24	,	,	PUNCT
ajst-23440	83	25	2023:6	2023:6	NUM
ajst-23440	83	26	.	.	PUNCT
ajst-23440	84	1	doi	doi	NOUN
ajst-23440	84	2	:	:	PUNCT
ajst-23440	84	3	10	10	NUM
ajst-23440	84	4	.	.	X
ajst-23440	85	1	26914	26914	NUM
ajst-23440	85	2	/	/	SYM
ajst-23440	85	3	c.	c.	PROPN
ajst-23440	85	4	cnkihy	cnkihy	PROPN
ajst-23440	85	5	.	.	PUNCT
ajst-23440	86	1	2023.036061	2023.036061	X
ajst-23440	86	2	.	.	PUNCT
ajst-23440	87	1	[	[	X
ajst-23440	87	2	3	3	X
ajst-23440	87	3	]	]	X
ajst-23440	87	4	liang	liang	PROPN
ajst-23440	87	5	f	f	PROPN
ajst-23440	87	6	,	,	PUNCT
ajst-23440	87	7	zhou	zhou	PROPN
ajst-23440	87	8	y	y	PROPN
ajst-23440	87	9	,	,	PUNCT
ajst-23440	87	10	chen	chen	PROPN
ajst-23440	87	11	x	x	PROPN
ajst-23440	87	12	,	,	PUNCT
ajst-23440	87	13	et	et	PROPN
ajst-23440	87	14	al	al	PROPN
ajst-23440	87	15	.	.	PROPN
ajst-23440	88	1	review	review	PROPN
ajst-23440	88	2	of	of	ADP
ajst-23440	88	3	target	target	NOUN
ajst-23440	88	4	detection	detection	NOUN
ajst-23440	88	5	technology	technology	NOUN
ajst-23440	88	6	based	base	VERB
ajst-23440	88	7	on	on	ADP
ajst-23440	88	8	deep	deep	ADJ
ajst-23440	88	9	learning[c]//	learning[c]//	PROPN
ajst-23440	88	10	(	(	PUNCT
ajst-23440	88	11	apise	apise	PROPN
ajst-23440	88	12	)	)	PUNCT
ajst-23440	88	13	.	.	PUNCT
ajst-23440	89	1	proceedings	proceeding	NOUN
ajst-23440	89	2	of	of	ADP
ajst-23440	89	3	5th	5th	ADJ
ajst-23440	89	4	international	international	ADJ
ajst-23440	89	5	conference	conference	NOUN
ajst-23440	89	6	on	on	ADP
ajst-23440	89	7	control	control	NOUN
ajst-23440	89	8	engineering	engineering	NOUN
ajst-23440	89	9	and	and	CCONJ
ajst-23440	89	10	artificial	artificial	ADJ
ajst-23440	89	11	intelligence	intelligence	NOUN
ajst-23440	89	12	(	(	PUNCT
ajst-23440	89	13	cceai	cceai	PROPN
ajst-23440	89	14	2021).acm	2021).acm	NUM
ajst-23440	89	15	,	,	PUNCT
ajst-23440	89	16	2021:4.doi	2021:4.doi	PRON
ajst-23440	89	17	:	:	PUNCT
ajst-23440	89	18	10.26914/	10.26914/	NUM
ajst-23440	89	19	c.cnkihy	c.cnkihy	NOUN
ajst-23440	89	20	.	.	PUNCT
ajst-23440	89	21	2021	2021	NUM
ajst-23440	89	22	.	.	PUNCT
ajst-23440	89	23	001576	001576	NUM
ajst-23440	89	24	.	.	PUNCT
ajst-23440	90	1	[	[	X
ajst-23440	90	2	4	4	NUM
ajst-23440	90	3	]	]	X
ajst-23440	90	4	liu	liu	PROPN
ajst-23440	90	5	,	,	PUNCT
ajst-23440	90	6	qingtang	qingtang	PROPN
ajst-23440	90	7	,	,	PUNCT
ajst-23440	90	8	et	et	PROPN
ajst-23440	90	9	al	al	PROPN
ajst-23440	90	10	.	.	PUNCT
ajst-23440	91	1	"	"	PUNCT
ajst-23440	91	2	yolov8n_bt	yolov8n_bt	ADJ
ajst-23440	91	3	:	:	PUNCT
ajst-23440	91	4	research	research	NOUN
ajst-23440	91	5	on	on	ADP
ajst-23440	91	6	classroom	classroom	NOUN
ajst-23440	91	7	learning	learn	VERB
ajst-23440	91	8	behavior	behavior	NOUN
ajst-23440	91	9	recognition	recognition	NOUN
ajst-23440	91	10	algorithm	algorithm	NOUN
ajst-23440	91	11	based	base	VERB
ajst-23440	91	12	on	on	ADP
ajst-23440	91	13	improved	improved	ADJ
ajst-23440	91	14	yolov8n	yolov8n	NOUN
ajst-23440	91	15	.	.	PUNCT
ajst-23440	91	16	"	"	PUNCT
ajst-23440	92	1	ieee	ieee	NOUN
ajst-23440	92	2	access	access	NOUN
ajst-23440	92	3	(	(	PUNCT
ajst-23440	92	4	2024	2024	NUM
ajst-23440	92	5	)	)	PUNCT
ajst-23440	92	6	.	.	PUNCT
ajst-23440	93	1	[	[	X
ajst-23440	93	2	5	5	NUM
ajst-23440	93	3	]	]	PUNCT
ajst-23440	93	4	zhigang	zhigang	PROPN
ajst-23440	93	5	l	l	PROPN
ajst-23440	93	6	,	,	PUNCT
ajst-23440	93	7	baoshan	baoshan	PROPN
ajst-23440	93	8	s	s	PROPN
ajst-23440	93	9	,	,	PUNCT
ajst-23440	93	10	kaiyu	kaiyu	NOUN
ajst-23440	93	11	b	b	PROPN
ajst-23440	93	12	.optimization	.optimization	NOUN
ajst-23440	93	13	of	of	ADP
ajst-23440	93	14	yolov7	yolov7	NOUN
ajst-23440	93	15	based	base	VERB
ajst-23440	93	16	on	on	ADP
ajst-23440	93	17	pconv	pconv	NOUN
ajst-23440	93	18	,	,	PUNCT
ajst-23440	93	19	se	se	ADV
ajst-23440	93	20	attention	attention	NOUN
ajst-23440	93	21	and	and	CCONJ
ajst-23440	93	22	wise	wise	ADJ
ajst-23440	93	23	-	-	PUNCT
ajst-23440	93	24	iou[j].international	iou[j].international	ADJ
ajst-23440	93	25	journal	journal	NOUN
ajst-23440	93	26	of	of	ADP
ajst-23440	93	27	computational	computational	ADJ
ajst-23440	93	28	intelligence	intelligence	NOUN
ajst-23440	93	29	and	and	CCONJ
ajst-23440	93	30	applications	application	NOUN
ajst-23440	93	31	,	,	PUNCT
ajst-23440	93	32	2024	2024	NUM
ajst-23440	93	33	,	,	PUNCT
ajst-23440	93	34	23(01	23(01	NUM
ajst-23440	93	35	)	)	PUNCT
ajst-23440	93	36	.	.	PUNCT
ajst-23440	94	1	[	[	X
ajst-23440	94	2	6	6	NUM
ajst-23440	94	3	]	]	X
ajst-23440	94	4	feng	feng	PROPN
ajst-23440	94	5	l	l	PROPN
ajst-23440	94	6	,	,	PUNCT
ajst-23440	94	7	kang	kang	PROPN
ajst-23440	94	8	x	x	PROPN
ajst-23440	94	9	,	,	PUNCT
ajst-23440	94	10	zhengpeng	zhengpeng	PROPN
ajst-23440	94	11	h	h	NOUN
ajst-23440	94	12	,	,	PUNCT
ajst-23440	94	13	et	et	PROPN
ajst-23440	94	14	al	al	PROPN
ajst-23440	94	15	.	.	PROPN
ajst-23440	94	16	fabric	fabric	NOUN
ajst-23440	94	17	defect	defect	NOUN
ajst-23440	94	18	detection	detection	NOUN
ajst-23440	94	19	algorithm	algorithm	NOUN
ajst-23440	94	20	based	base	VERB
ajst-23440	94	21	on	on	ADP
ajst-23440	94	22	improved	improved	ADJ
ajst-23440	94	23	yolov5[j	yolov5[j	PROPN
ajst-23440	94	24	]	]	PUNCT
ajst-23440	94	25	.	.	PUNCT
ajst-23440	95	1	the	the	DET
ajst-23440	95	2	visual	visual	ADJ
ajst-23440	95	3	computer	computer	NOUN
ajst-23440	95	4	,	,	PUNCT
ajst-23440	95	5	2023,40(4):2309	2023,40(4):2309	NUM
ajst-23440	95	6	-	-	SYM
ajst-23440	95	7	2324	2324	NUM
ajst-23440	95	8	.	.	PUNCT