id	sid	tid	token	lemma	pos
fcis-14812	1	1	frontiers	frontier	NOUN
fcis-14812	1	2	in	in	ADP
fcis-14812	1	3	computing	computing	NOUN
fcis-14812	1	4	and	and	CCONJ
fcis-14812	1	5	intelligent	intelligent	ADJ
fcis-14812	1	6	systems	system	NOUN
fcis-14812	1	7	issn	issn	VERB
fcis-14812	1	8	:	:	PUNCT
fcis-14812	1	9	2832	2832	NUM
fcis-14812	1	10	-	-	SYM
fcis-14812	1	11	6024	6024	NUM
fcis-14812	1	12	|	|	NOUN
fcis-14812	1	13	vol	vol	NOUN
fcis-14812	1	14	.	.	PROPN
fcis-14812	2	1	6	6	NUM
fcis-14812	2	2	,	,	PUNCT
fcis-14812	2	3	no	no	INTJ
fcis-14812	2	4	.	.	NOUN
fcis-14812	2	5	1	1	NUM
fcis-14812	2	6	,	,	PUNCT
fcis-14812	2	7	2023	2023	NUM
fcis-14812	2	8	70	70	NUM
fcis-14812	2	9	dfenet	dfenet	NOUN
fcis-14812	2	10	:	:	PUNCT
fcis-14812	2	11	double	double	ADJ
fcis-14812	2	12	feature	feature	NOUN
fcis-14812	2	13	enhanced	enhance	VERB
fcis-14812	2	14	class	class	NOUN
fcis-14812	2	15	agnostic	agnostic	ADJ
fcis-14812	2	16	counting	counting	NOUN
fcis-14812	2	17	methods	method	NOUN
fcis-14812	2	18	jiakang	jiakang	PROPN
fcis-14812	2	19	liu	liu	PROPN
fcis-14812	2	20	,	,	PUNCT
fcis-14812	2	21	hua	hua	PROPN
fcis-14812	2	22	huo	huo	PROPN
fcis-14812	2	23	college	college	PROPN
fcis-14812	2	24	of	of	ADP
fcis-14812	2	25	information	information	NOUN
fcis-14812	2	26	engineering	engineering	PROPN
fcis-14812	2	27	,	,	PUNCT
fcis-14812	2	28	henan	henan	PROPN
fcis-14812	2	29	university	university	PROPN
fcis-14812	2	30	of	of	ADP
fcis-14812	2	31	science	science	NOUN
fcis-14812	2	32	and	and	CCONJ
fcis-14812	2	33	technology	technology	NOUN
fcis-14812	2	34	,	,	PUNCT
fcis-14812	2	35	luoyang	luoyang	PROPN
fcis-14812	2	36	471000	471000	NUM
fcis-14812	2	37	,	,	PUNCT
fcis-14812	2	38	china	china	PROPN
fcis-14812	2	39	abstract	abstract	NOUN
fcis-14812	2	40	:	:	PUNCT
fcis-14812	2	41	object	object	NOUN
fcis-14812	2	42	counting	counting	NOUN
fcis-14812	2	43	is	be	AUX
fcis-14812	2	44	a	a	DET
fcis-14812	2	45	basic	basic	ADJ
fcis-14812	2	46	computer	computer	NOUN
fcis-14812	2	47	vision	vision	NOUN
fcis-14812	2	48	task	task	NOUN
fcis-14812	2	49	,	,	PUNCT
fcis-14812	2	50	which	which	PRON
fcis-14812	2	51	can	can	AUX
fcis-14812	2	52	estimate	estimate	VERB
fcis-14812	2	53	the	the	DET
fcis-14812	2	54	number	number	NOUN
fcis-14812	2	55	of	of	ADP
fcis-14812	2	56	each	each	DET
fcis-14812	2	57	object	object	NOUN
fcis-14812	2	58	in	in	ADP
fcis-14812	2	59	an	an	DET
fcis-14812	2	60	image	image	NOUN
fcis-14812	2	61	,	,	PUNCT
fcis-14812	2	62	thus	thus	ADV
fcis-14812	2	63	providing	provide	VERB
fcis-14812	2	64	valuable	valuable	ADJ
fcis-14812	2	65	information	information	NOUN
fcis-14812	2	66	.	.	PUNCT
fcis-14812	3	1	in	in	ADP
fcis-14812	3	2	dense	dense	ADJ
fcis-14812	3	3	scenes	scene	NOUN
fcis-14812	3	4	,	,	PUNCT
fcis-14812	3	5	there	there	PRON
fcis-14812	3	6	are	be	VERB
fcis-14812	3	7	huge	huge	ADJ
fcis-14812	3	8	differences	difference	NOUN
fcis-14812	3	9	in	in	ADP
fcis-14812	3	10	target	target	NOUN
fcis-14812	3	11	individual	individual	ADJ
fcis-14812	3	12	scale	scale	NOUN
fcis-14812	3	13	,	,	PUNCT
fcis-14812	3	14	and	and	CCONJ
fcis-14812	3	15	the	the	DET
fcis-14812	3	16	different	different	ADJ
fcis-14812	3	17	target	target	NOUN
fcis-14812	3	18	individual	individual	ADJ
fcis-14812	3	19	scale	scale	NOUN
fcis-14812	3	20	leads	lead	VERB
fcis-14812	3	21	to	to	ADP
fcis-14812	3	22	low	low	ADJ
fcis-14812	3	23	accuracy	accuracy	NOUN
fcis-14812	3	24	of	of	ADP
fcis-14812	3	25	target	target	NOUN
fcis-14812	3	26	count	count	NOUN
fcis-14812	3	27	.	.	PUNCT
fcis-14812	4	1	in	in	ADP
fcis-14812	4	2	addition	addition	NOUN
fcis-14812	4	3	,	,	PUNCT
fcis-14812	4	4	most	most	ADJ
fcis-14812	4	5	of	of	ADP
fcis-14812	4	6	the	the	DET
fcis-14812	4	7	existing	exist	VERB
fcis-14812	4	8	target	target	NOUN
fcis-14812	4	9	count	count	NOUN
fcis-14812	4	10	datasets	dataset	NOUN
fcis-14812	4	11	in	in	ADP
fcis-14812	4	12	the	the	DET
fcis-14812	4	13	field	field	NOUN
fcis-14812	4	14	require	require	VERB
fcis-14812	4	15	a	a	DET
fcis-14812	4	16	lot	lot	NOUN
fcis-14812	4	17	of	of	ADP
fcis-14812	4	18	manual	manual	ADJ
fcis-14812	4	19	creation	creation	NOUN
fcis-14812	4	20	and	and	CCONJ
fcis-14812	4	21	annotation	annotation	NOUN
fcis-14812	4	22	,	,	PUNCT
fcis-14812	4	23	which	which	PRON
fcis-14812	4	24	increases	increase	VERB
fcis-14812	4	25	the	the	DET
fcis-14812	4	26	cost	cost	NOUN
fcis-14812	4	27	and	and	CCONJ
fcis-14812	4	28	difficulty	difficulty	NOUN
fcis-14812	4	29	of	of	ADP
fcis-14812	4	30	the	the	DET
fcis-14812	4	31	dataset	dataset	NOUN
fcis-14812	4	32	,	,	PUNCT
fcis-14812	4	33	lack	lack	NOUN
fcis-14812	4	34	of	of	ADP
fcis-14812	4	35	ease	ease	NOUN
fcis-14812	4	36	of	of	ADP
fcis-14812	4	37	use	use	NOUN
fcis-14812	4	38	and	and	CCONJ
fcis-14812	4	39	portability	portability	NOUN
fcis-14812	4	40	.	.	PUNCT
fcis-14812	5	1	to	to	PART
fcis-14812	5	2	solve	solve	VERB
fcis-14812	5	3	these	these	DET
fcis-14812	5	4	problems	problem	NOUN
fcis-14812	5	5	,	,	PUNCT
fcis-14812	5	6	this	this	DET
fcis-14812	5	7	paper	paper	NOUN
fcis-14812	5	8	proposes	propose	VERB
fcis-14812	5	9	a	a	DET
fcis-14812	5	10	class	class	NOUN
fcis-14812	5	11	agnostic	agnostic	ADJ
fcis-14812	5	12	counting	counting	NOUN
fcis-14812	5	13	method	method	NOUN
fcis-14812	5	14	double	double	ADJ
fcis-14812	5	15	feature	feature	NOUN
fcis-14812	5	16	enhancement	enhancement	NOUN
fcis-14812	5	17	net	net	NOUN
fcis-14812	5	18	based	base	VERB
fcis-14812	5	19	on	on	ADP
fcis-14812	5	20	improved	improve	VERB
fcis-14812	5	21	bilinear	bilinear	NOUN
fcis-14812	5	22	matching	match	VERB
fcis-14812	5	23	network+	network+	ADP
fcis-14812	5	24	(	(	PUNCT
fcis-14812	5	25	bmnet+	bmnet+	PROPN
fcis-14812	5	26	)	)	PUNCT
fcis-14812	5	27	.	.	PUNCT
fcis-14812	6	1	by	by	ADP
fcis-14812	6	2	introducing	introduce	VERB
fcis-14812	6	3	the	the	DET
fcis-14812	6	4	feature	feature	NOUN
fcis-14812	6	5	enhancement	enhancement	NOUN
fcis-14812	6	6	module	module	NOUN
fcis-14812	6	7	based	base	VERB
fcis-14812	6	8	on	on	ADP
fcis-14812	6	9	the	the	DET
fcis-14812	6	10	principle	principle	NOUN
fcis-14812	6	11	of	of	ADP
fcis-14812	6	12	conditional	conditional	ADJ
fcis-14812	6	13	random	random	ADJ
fcis-14812	6	14	field	field	NOUN
fcis-14812	6	15	and	and	CCONJ
fcis-14812	6	16	the	the	DET
fcis-14812	6	17	adaptively	adaptively	ADV
fcis-14812	6	18	spatial	spatial	ADJ
fcis-14812	6	19	feature	feature	NOUN
fcis-14812	6	20	fusion	fusion	NOUN
fcis-14812	6	21	module	module	NOUN
fcis-14812	6	22	,	,	PUNCT
fcis-14812	6	23	combined	combine	VERB
fcis-14812	6	24	with	with	ADP
fcis-14812	6	25	the	the	DET
fcis-14812	6	26	feature	feature	NOUN
fcis-14812	6	27	similarity	similarity	NOUN
fcis-14812	6	28	measurement	measurement	NOUN
fcis-14812	6	29	strategy	strategy	NOUN
fcis-14812	6	30	of	of	ADP
fcis-14812	6	31	bilinear	bilinear	PROPN
fcis-14812	6	32	matching	matching	NOUN
fcis-14812	6	33	network	network	NOUN
fcis-14812	6	34	,	,	PUNCT
fcis-14812	6	35	the	the	DET
fcis-14812	6	36	method	method	NOUN
fcis-14812	6	37	can	can	AUX
fcis-14812	6	38	effectively	effectively	ADV
fcis-14812	6	39	extract	extract	VERB
fcis-14812	6	40	the	the	DET
fcis-14812	6	41	target	target	NOUN
fcis-14812	6	42	features	feature	NOUN
fcis-14812	6	43	of	of	ADP
fcis-14812	6	44	different	different	ADJ
fcis-14812	6	45	scales	scale	NOUN
fcis-14812	6	46	,	,	PUNCT
fcis-14812	6	47	enhance	enhance	VERB
fcis-14812	6	48	the	the	DET
fcis-14812	6	49	adaptability	adaptability	NOUN
fcis-14812	6	50	to	to	ADP
fcis-14812	6	51	the	the	DET
fcis-14812	6	52	targets	target	NOUN
fcis-14812	6	53	with	with	ADP
fcis-14812	6	54	large	large	ADJ
fcis-14812	6	55	scale	scale	NOUN
fcis-14812	6	56	changes	change	NOUN
fcis-14812	6	57	,	,	PUNCT
fcis-14812	6	58	and	and	CCONJ
fcis-14812	6	59	improve	improve	VERB
fcis-14812	6	60	the	the	DET
fcis-14812	6	61	counting	counting	NOUN
fcis-14812	6	62	performance	performance	NOUN
fcis-14812	6	63	of	of	ADP
fcis-14812	6	64	the	the	DET
fcis-14812	6	65	network	network	NOUN
fcis-14812	6	66	.	.	PUNCT
fcis-14812	7	1	experiments	experiment	NOUN
fcis-14812	7	2	were	be	AUX
fcis-14812	7	3	carried	carry	VERB
fcis-14812	7	4	out	out	ADP
fcis-14812	7	5	on	on	ADP
fcis-14812	7	6	fsc-147	fsc-147	NOUN
fcis-14812	7	7	data	datum	NOUN
fcis-14812	7	8	set	set	VERB
fcis-14812	7	9	,	,	PUNCT
fcis-14812	7	10	and	and	CCONJ
fcis-14812	7	11	the	the	DET
fcis-14812	7	12	experimental	experimental	ADJ
fcis-14812	7	13	results	result	NOUN
fcis-14812	7	14	show	show	VERB
fcis-14812	7	15	that	that	SCONJ
fcis-14812	7	16	the	the	DET
fcis-14812	7	17	proposed	propose	VERB
fcis-14812	7	18	model	model	NOUN
fcis-14812	7	19	has	have	AUX
fcis-14812	7	20	been	be	AUX
fcis-14812	7	21	further	far	ADV
fcis-14812	7	22	improved	improve	VERB
fcis-14812	7	23	in	in	ADP
fcis-14812	7	24	counting	count	VERB
fcis-14812	7	25	accuracy	accuracy	NOUN
fcis-14812	7	26	.	.	PUNCT
fcis-14812	8	1	the	the	DET
fcis-14812	8	2	mae	mae	PROPN
fcis-14812	8	3	and	and	CCONJ
fcis-14812	8	4	mse	mse	PROPN
fcis-14812	8	5	of	of	ADP
fcis-14812	8	6	the	the	DET
fcis-14812	8	7	verification	verification	NOUN
fcis-14812	8	8	set	set	NOUN
fcis-14812	8	9	are	be	AUX
fcis-14812	8	10	15.03	15.03	NUM
fcis-14812	8	11	and	and	CCONJ
fcis-14812	8	12	54.53	54.53	NUM
fcis-14812	8	13	respectively	respectively	ADV
fcis-14812	8	14	.	.	PUNCT
fcis-14812	9	1	in	in	ADP
fcis-14812	9	2	the	the	DET
fcis-14812	9	3	test	test	NOUN
fcis-14812	9	4	set	set	NOUN
fcis-14812	9	5	,	,	PUNCT
fcis-14812	9	6	mae	mae	PROPN
fcis-14812	9	7	reaches	reach	VERB
fcis-14812	9	8	13.65	13.65	NUM
fcis-14812	9	9	,	,	PUNCT
fcis-14812	9	10	mse	mse	PROPN
fcis-14812	9	11	reaches	reach	VERB
fcis-14812	9	12	89.54	89.54	NUM
fcis-14812	9	13	,	,	PUNCT
fcis-14812	9	14	and	and	CCONJ
fcis-14812	9	15	the	the	DET
fcis-14812	9	16	counting	counting	NOUN
fcis-14812	9	17	performance	performance	NOUN
fcis-14812	9	18	is	be	AUX
fcis-14812	9	19	at	at	ADP
fcis-14812	9	20	the	the	DET
fcis-14812	9	21	advanced	advanced	ADJ
fcis-14812	9	22	level	level	NOUN
fcis-14812	9	23	in	in	ADP
fcis-14812	9	24	the	the	DET
fcis-14812	9	25	field	field	NOUN
fcis-14812	9	26	.	.	PUNCT
fcis-14812	10	1	keywords	keyword	NOUN
fcis-14812	10	2	:	:	PUNCT
fcis-14812	10	3	object	object	NOUN
fcis-14812	10	4	count	count	NOUN
fcis-14812	10	5	;	;	PUNCT
fcis-14812	10	6	class	class	NOUN
fcis-14812	10	7	agnostic	agnostic	ADJ
fcis-14812	10	8	counting	counting	NOUN
fcis-14812	10	9	;	;	PUNCT
fcis-14812	10	10	conditional	conditional	ADJ
fcis-14812	10	11	random	random	ADJ
fcis-14812	10	12	field	field	NOUN
fcis-14812	10	13	;	;	PUNCT
fcis-14812	10	14	similarity	similarity	NOUN
fcis-14812	10	15	measure	measure	NOUN
fcis-14812	10	16	.	.	PUNCT
fcis-14812	11	1	1	1	X
fcis-14812	11	2	.	.	X
fcis-14812	11	3	introduction	introduction	NOUN
fcis-14812	11	4	object	object	NOUN
fcis-14812	11	5	counting	counting	NOUN
fcis-14812	11	6	is	be	AUX
fcis-14812	11	7	a	a	DET
fcis-14812	11	8	fundamental	fundamental	ADJ
fcis-14812	11	9	and	and	CCONJ
fcis-14812	11	10	important	important	ADJ
fcis-14812	11	11	computer	computer	NOUN
fcis-14812	11	12	vision	vision	NOUN
fcis-14812	11	13	task	task	NOUN
fcis-14812	11	14	that	that	PRON
fcis-14812	11	15	provides	provide	VERB
fcis-14812	11	16	quantitative	quantitative	ADJ
fcis-14812	11	17	estimation	estimation	NOUN
fcis-14812	11	18	of	of	ADP
fcis-14812	11	19	targets	target	NOUN
fcis-14812	11	20	in	in	ADP
fcis-14812	11	21	images	image	NOUN
fcis-14812	11	22	,	,	PUNCT
fcis-14812	11	23	thus	thus	ADV
fcis-14812	11	24	providing	provide	VERB
fcis-14812	11	25	valuable	valuable	ADJ
fcis-14812	11	26	information	information	NOUN
fcis-14812	11	27	for	for	ADP
fcis-14812	11	28	various	various	ADJ
fcis-14812	11	29	fields	field	NOUN
fcis-14812	11	30	of	of	ADP
fcis-14812	11	31	social	social	ADJ
fcis-14812	11	32	production	production	NOUN
fcis-14812	11	33	.	.	PUNCT
fcis-14812	12	1	this	this	DET
fcis-14812	12	2	task	task	NOUN
fcis-14812	12	3	is	be	AUX
fcis-14812	12	4	important	important	ADJ
fcis-14812	12	5	in	in	ADP
fcis-14812	12	6	many	many	ADJ
fcis-14812	12	7	practical	practical	ADJ
fcis-14812	12	8	applications	application	NOUN
fcis-14812	12	9	,	,	PUNCT
fcis-14812	12	10	including	include	VERB
fcis-14812	12	11	traffic	traffic	NOUN
fcis-14812	12	12	management	management	NOUN
fcis-14812	12	13	,	,	PUNCT
fcis-14812	12	14	public	public	ADJ
fcis-14812	12	15	safety	safety	NOUN
fcis-14812	12	16	,	,	PUNCT
fcis-14812	12	17	medical	medical	ADJ
fcis-14812	12	18	diagnosis	diagnosis	NOUN
fcis-14812	12	19	,	,	PUNCT
fcis-14812	12	20	and	and	CCONJ
fcis-14812	12	21	agricultural	agricultural	ADJ
fcis-14812	12	22	monitoring	monitoring	NOUN
fcis-14812	12	23	.	.	PUNCT
fcis-14812	13	1	for	for	ADP
fcis-14812	13	2	example	example	NOUN
fcis-14812	13	3	,	,	PUNCT
fcis-14812	13	4	in	in	ADP
fcis-14812	13	5	the	the	DET
fcis-14812	13	6	field	field	NOUN
fcis-14812	13	7	of	of	ADP
fcis-14812	13	8	traffic	traffic	NOUN
fcis-14812	13	9	management	management	NOUN
fcis-14812	13	10	,	,	PUNCT
fcis-14812	13	11	object	object	NOUN
fcis-14812	13	12	counting	counting	NOUN
fcis-14812	13	13	for	for	ADP
fcis-14812	13	14	road	road	NOUN
fcis-14812	13	15	traffic	traffic	NOUN
fcis-14812	13	16	can	can	AUX
fcis-14812	13	17	monitor	monitor	VERB
fcis-14812	13	18	the	the	DET
fcis-14812	13	19	congestion	congestion	NOUN
fcis-14812	13	20	of	of	ADP
fcis-14812	13	21	various	various	ADJ
fcis-14812	13	22	roads	road	NOUN
fcis-14812	13	23	in	in	ADP
fcis-14812	13	24	a	a	DET
fcis-14812	13	25	city	city	NOUN
fcis-14812	13	26	,	,	PUNCT
fcis-14812	13	27	thus	thus	ADV
fcis-14812	13	28	assisting	assist	VERB
fcis-14812	13	29	in	in	ADP
fcis-14812	13	30	traffic	traffic	NOUN
fcis-14812	13	31	management	management	NOUN
fcis-14812	13	32	;	;	PUNCT
fcis-14812	13	33	in	in	ADP
fcis-14812	13	34	the	the	DET
fcis-14812	13	35	field	field	NOUN
fcis-14812	13	36	of	of	ADP
fcis-14812	13	37	public	public	ADJ
fcis-14812	13	38	safety	safety	NOUN
fcis-14812	13	39	,	,	PUNCT
fcis-14812	13	40	object	object	NOUN
fcis-14812	13	41	counting	counting	NOUN
fcis-14812	13	42	for	for	ADP
fcis-14812	13	43	crowds	crowd	NOUN
fcis-14812	13	44	can	can	AUX
fcis-14812	13	45	detect	detect	VERB
fcis-14812	13	46	the	the	DET
fcis-14812	13	47	number	number	NOUN
fcis-14812	13	48	of	of	ADP
fcis-14812	13	49	pedestrians	pedestrian	NOUN
fcis-14812	13	50	on	on	ADP
fcis-14812	13	51	the	the	DET
fcis-14812	13	52	streets	street	NOUN
fcis-14812	13	53	in	in	ADP
fcis-14812	13	54	real	real	ADJ
fcis-14812	13	55	time	time	NOUN
fcis-14812	13	56	,	,	PUNCT
fcis-14812	13	57	and	and	CCONJ
fcis-14812	13	58	detect	detect	VERB
fcis-14812	13	59	safety	safety	NOUN
fcis-14812	13	60	hazards	hazard	NOUN
fcis-14812	13	61	in	in	ADP
fcis-14812	13	62	advance	advance	NOUN
fcis-14812	13	63	in	in	ADP
fcis-14812	13	64	order	order	NOUN
fcis-14812	13	65	to	to	PART
fcis-14812	13	66	avoid	avoid	VERB
fcis-14812	13	67	potential	potential	ADJ
fcis-14812	13	68	hazards	hazard	NOUN
fcis-14812	13	69	,	,	PUNCT
fcis-14812	13	70	such	such	ADJ
fcis-14812	13	71	as	as	ADP
fcis-14812	13	72	stampede	stampede	NOUN
fcis-14812	13	73	incidents	incident	NOUN
fcis-14812	13	74	.	.	PUNCT
fcis-14812	14	1	however	however	ADV
fcis-14812	14	2	,	,	PUNCT
fcis-14812	14	3	object	object	VERB
fcis-14812	14	4	counting	counting	NOUN
fcis-14812	14	5	datasets	dataset	NOUN
fcis-14812	14	6	often	often	ADV
fcis-14812	14	7	require	require	VERB
fcis-14812	14	8	a	a	DET
fcis-14812	14	9	lot	lot	NOUN
fcis-14812	14	10	of	of	ADP
fcis-14812	14	11	manual	manual	ADJ
fcis-14812	14	12	creation	creation	NOUN
fcis-14812	14	13	and	and	CCONJ
fcis-14812	14	14	annotation	annotation	NOUN
fcis-14812	14	15	,	,	PUNCT
fcis-14812	14	16	which	which	PRON
fcis-14812	14	17	not	not	PART
fcis-14812	14	18	only	only	ADV
fcis-14812	14	19	increases	increase	VERB
fcis-14812	14	20	the	the	DET
fcis-14812	14	21	cost	cost	NOUN
fcis-14812	14	22	and	and	CCONJ
fcis-14812	14	23	difficulty	difficulty	NOUN
fcis-14812	14	24	of	of	ADP
fcis-14812	14	25	the	the	DET
fcis-14812	14	26	dataset	dataset	NOUN
fcis-14812	14	27	,	,	PUNCT
fcis-14812	14	28	but	but	CCONJ
fcis-14812	14	29	also	also	ADV
fcis-14812	14	30	restricts	restrict	VERB
fcis-14812	14	31	the	the	DET
fcis-14812	14	32	diversity	diversity	NOUN
fcis-14812	14	33	and	and	CCONJ
fcis-14812	14	34	versatility	versatility	NOUN
fcis-14812	14	35	of	of	ADP
fcis-14812	14	36	the	the	DET
fcis-14812	14	37	dataset	dataset	NOUN
fcis-14812	14	38	,	,	PUNCT
fcis-14812	14	39	so	so	SCONJ
fcis-14812	14	40	that	that	SCONJ
fcis-14812	14	41	traditional	traditional	ADJ
fcis-14812	14	42	object	object	NOUN
fcis-14812	14	43	counting	counting	NOUN
fcis-14812	14	44	methods	method	NOUN
fcis-14812	14	45	can	can	AUX
fcis-14812	14	46	often	often	ADV
fcis-14812	14	47	only	only	ADV
fcis-14812	14	48	be	be	AUX
fcis-14812	14	49	used	use	VERB
fcis-14812	14	50	for	for	ADP
fcis-14812	14	51	counting	count	VERB
fcis-14812	14	52	specific	specific	ADJ
fcis-14812	14	53	categories	category	NOUN
fcis-14812	14	54	of	of	ADP
fcis-14812	14	55	targets	target	NOUN
fcis-14812	14	56	,	,	PUNCT
fcis-14812	14	57	and	and	CCONJ
fcis-14812	14	58	lack	lack	NOUN
fcis-14812	14	59	of	of	ADP
fcis-14812	14	60	ease	ease	NOUN
fcis-14812	14	61	of	of	ADP
fcis-14812	14	62	use	use	NOUN
fcis-14812	14	63	and	and	CCONJ
fcis-14812	14	64	transferability	transferability	NOUN
fcis-14812	14	65	.	.	PUNCT
fcis-14812	15	1	to	to	PART
fcis-14812	15	2	solve	solve	VERB
fcis-14812	15	3	this	this	DET
fcis-14812	15	4	problem	problem	NOUN
fcis-14812	15	5	,	,	PUNCT
fcis-14812	15	6	lu	lu	PROPN
fcis-14812	15	7	et	et	NOUN
fcis-14812	15	8	al	al	PROPN
fcis-14812	16	1	[	[	X
fcis-14812	16	2	13	13	NUM
fcis-14812	16	3	]	]	PUNCT
fcis-14812	16	4	first	first	ADV
fcis-14812	16	5	proposed	propose	VERB
fcis-14812	16	6	the	the	DET
fcis-14812	16	7	concept	concept	NOUN
fcis-14812	16	8	of	of	ADP
fcis-14812	16	9	class	class	NOUN
fcis-14812	16	10	-	-	PUNCT
fcis-14812	16	11	agnostic	agnostic	ADJ
fcis-14812	16	12	counting	counting	NOUN
fcis-14812	16	13	(	(	PUNCT
fcis-14812	16	14	cac	cac	PROPN
fcis-14812	16	15	)	)	PUNCT
fcis-14812	16	16	,	,	PUNCT
fcis-14812	16	17	which	which	PRON
fcis-14812	16	18	aims	aim	VERB
fcis-14812	16	19	to	to	PART
fcis-14812	16	20	create	create	VERB
fcis-14812	16	21	a	a	DET
fcis-14812	16	22	generalized	generalized	ADJ
fcis-14812	16	23	counting	counting	NOUN
fcis-14812	16	24	model	model	NOUN
fcis-14812	16	25	that	that	PRON
fcis-14812	16	26	can	can	AUX
fcis-14812	16	27	adapt	adapt	VERB
fcis-14812	16	28	to	to	ADP
fcis-14812	16	29	any	any	DET
fcis-14812	16	30	class	class	NOUN
fcis-14812	16	31	of	of	ADP
fcis-14812	16	32	targets	target	NOUN
fcis-14812	16	33	without	without	ADP
fcis-14812	16	34	the	the	DET
fcis-14812	16	35	need	need	NOUN
fcis-14812	16	36	to	to	PART
fcis-14812	16	37	know	know	VERB
fcis-14812	16	38	the	the	DET
fcis-14812	16	39	target	target	NOUN
fcis-14812	16	40	's	's	PART
fcis-14812	16	41	class	class	NOUN
fcis-14812	16	42	information	information	NOUN
fcis-14812	16	43	in	in	ADP
fcis-14812	16	44	advance	advance	NOUN
fcis-14812	16	45	.	.	PUNCT
fcis-14812	17	1	shi	shi	PROPN
fcis-14812	17	2	et	et	PROPN
fcis-14812	17	3	al	al	PROPN
fcis-14812	18	1	[	[	X
fcis-14812	18	2	15	15	NUM
fcis-14812	18	3	]	]	PUNCT
fcis-14812	18	4	designed	design	VERB
fcis-14812	18	5	a	a	DET
fcis-14812	18	6	network	network	NOUN
fcis-14812	18	7	based	base	VERB
fcis-14812	18	8	on	on	ADP
fcis-14812	18	9	similarity	similarity	NOUN
fcis-14812	18	10	measures	measure	NOUN
fcis-14812	18	11	based	base	VERB
fcis-14812	18	12	on	on	ADP
fcis-14812	18	13	this	this	DET
fcis-14812	18	14	foundation	foundation	NOUN
fcis-14812	18	15	bmnet+	bmnet+	PROPN
fcis-14812	18	16	,	,	PUNCT
fcis-14812	18	17	by	by	ADP
fcis-14812	18	18	comparing	compare	VERB
fcis-14812	18	19	the	the	DET
fcis-14812	18	20	feature	feature	NOUN
fcis-14812	18	21	similarity	similarity	NOUN
fcis-14812	18	22	between	between	ADP
fcis-14812	18	23	the	the	DET
fcis-14812	18	24	sample	sample	NOUN
fcis-14812	18	25	image	image	NOUN
fcis-14812	18	26	and	and	CCONJ
fcis-14812	18	27	the	the	DET
fcis-14812	18	28	query	query	NOUN
fcis-14812	18	29	image	image	NOUN
fcis-14812	18	30	in	in	ADP
fcis-14812	18	31	order	order	NOUN
fcis-14812	18	32	to	to	PART
fcis-14812	18	33	generate	generate	VERB
fcis-14812	18	34	a	a	DET
fcis-14812	18	35	target	target	NOUN
fcis-14812	18	36	count	count	NOUN
fcis-14812	18	37	density	density	NOUN
fcis-14812	18	38	map	map	NOUN
fcis-14812	18	39	.	.	PUNCT
fcis-14812	19	1	however	however	ADV
fcis-14812	19	2	,	,	PUNCT
fcis-14812	19	3	when	when	SCONJ
fcis-14812	19	4	confronted	confront	VERB
fcis-14812	19	5	with	with	ADP
fcis-14812	19	6	multiple	multiple	ADJ
fcis-14812	19	7	targets	target	NOUN
fcis-14812	19	8	with	with	ADP
fcis-14812	19	9	large	large	ADJ
fcis-14812	19	10	scale	scale	NOUN
fcis-14812	19	11	variations	variation	NOUN
fcis-14812	19	12	,	,	PUNCT
fcis-14812	19	13	this	this	DET
fcis-14812	19	14	network	network	NOUN
fcis-14812	19	15	often	often	ADV
fcis-14812	19	16	fails	fail	VERB
fcis-14812	19	17	to	to	PART
fcis-14812	19	18	perform	perform	VERB
fcis-14812	19	19	the	the	DET
fcis-14812	19	20	task	task	NOUN
fcis-14812	19	21	of	of	ADP
fcis-14812	19	22	feature	feature	NOUN
fcis-14812	19	23	comparison	comparison	NOUN
fcis-14812	19	24	of	of	ADP
fcis-14812	19	25	multi	multi	ADJ
fcis-14812	19	26	-	-	ADJ
fcis-14812	19	27	scale	scale	ADJ
fcis-14812	19	28	targets	target	NOUN
fcis-14812	19	29	well	well	ADV
fcis-14812	19	30	,	,	PUNCT
fcis-14812	19	31	resulting	result	VERB
fcis-14812	19	32	in	in	ADP
fcis-14812	19	33	inaccurate	inaccurate	ADJ
fcis-14812	19	34	counting	counting	NOUN
fcis-14812	19	35	results	result	NOUN
fcis-14812	19	36	.	.	PUNCT
fcis-14812	20	1	therefore	therefore	ADV
fcis-14812	20	2	,	,	PUNCT
fcis-14812	20	3	in	in	ADP
fcis-14812	20	4	this	this	DET
fcis-14812	20	5	paper	paper	NOUN
fcis-14812	20	6	,	,	PUNCT
fcis-14812	20	7	we	we	PRON
fcis-14812	20	8	propose	propose	VERB
fcis-14812	20	9	a	a	DET
fcis-14812	20	10	conditional	conditional	ADJ
fcis-14812	20	11	random	random	ADJ
fcis-14812	20	12	field	field	NOUN
fcis-14812	20	13	-	-	PUNCT
fcis-14812	20	14	based	base	VERB
fcis-14812	20	15	feature	feature	NOUN
fcis-14812	20	16	extraction	extraction	NOUN
fcis-14812	20	17	and	and	CCONJ
fcis-14812	20	18	enhancement	enhancement	NOUN
fcis-14812	20	19	method	method	NOUN
fcis-14812	20	20	for	for	ADP
fcis-14812	20	21	image	image	NOUN
fcis-14812	20	22	counting	counting	NOUN
fcis-14812	20	23	task	task	NOUN
fcis-14812	20	24	.	.	PUNCT
fcis-14812	21	1	we	we	PRON
fcis-14812	21	2	extract	extract	VERB
fcis-14812	21	3	two	two	NUM
fcis-14812	21	4	different	different	ADJ
fcis-14812	21	5	levels	level	NOUN
fcis-14812	21	6	of	of	ADP
fcis-14812	21	7	feature	feature	NOUN
fcis-14812	21	8	maps	map	NOUN
fcis-14812	21	9	from	from	ADP
fcis-14812	21	10	the	the	DET
fcis-14812	21	11	backbone	backbone	NOUN
fcis-14812	21	12	network	network	NOUN
fcis-14812	21	13	,	,	PUNCT
fcis-14812	21	14	which	which	PRON
fcis-14812	21	15	represent	represent	VERB
fcis-14812	21	16	the	the	DET
fcis-14812	21	17	low	low	ADJ
fcis-14812	21	18	and	and	CCONJ
fcis-14812	21	19	medium	medium	ADJ
fcis-14812	21	20	level	level	NOUN
fcis-14812	21	21	and	and	CCONJ
fcis-14812	21	22	high	high	ADJ
fcis-14812	21	23	level	level	NOUN
fcis-14812	21	24	features	feature	NOUN
fcis-14812	21	25	of	of	ADP
fcis-14812	21	26	the	the	DET
fcis-14812	21	27	image	image	NOUN
fcis-14812	21	28	,	,	PUNCT
fcis-14812	21	29	respectively	respectively	ADV
fcis-14812	21	30	.	.	PUNCT
fcis-14812	22	1	then	then	ADV
fcis-14812	22	2	we	we	PRON
fcis-14812	22	3	design	design	VERB
fcis-14812	22	4	a	a	DET
fcis-14812	22	5	dual	dual	ADJ
fcis-14812	22	6	feature	feature	NOUN
fcis-14812	22	7	enhancement	enhancement	NOUN
fcis-14812	22	8	module	module	NOUN
fcis-14812	22	9	,	,	PUNCT
fcis-14812	22	10	which	which	PRON
fcis-14812	22	11	consists	consist	VERB
fcis-14812	22	12	of	of	ADP
fcis-14812	22	13	a	a	DET
fcis-14812	22	14	selfsimilarity	selfsimilarity	NOUN
fcis-14812	22	15	module	module	NOUN
fcis-14812	22	16	and	and	CCONJ
fcis-14812	22	17	a	a	DET
fcis-14812	22	18	sfem	sfem	ADJ
fcis-14812	22	19	feature	feature	NOUN
fcis-14812	22	20	enhancement	enhancement	NOUN
fcis-14812	22	21	module	module	NOUN
fcis-14812	22	22	,	,	PUNCT
fcis-14812	22	23	for	for	ADP
fcis-14812	22	24	enhancing	enhance	VERB
fcis-14812	22	25	the	the	DET
fcis-14812	22	26	feature	feature	NOUN
fcis-14812	22	27	representation	representation	NOUN
fcis-14812	22	28	of	of	ADP
fcis-14812	22	29	the	the	DET
fcis-14812	22	30	query	query	NOUN
fcis-14812	22	31	image	image	NOUN
fcis-14812	22	32	by	by	ADP
fcis-14812	22	33	utilizing	utilize	VERB
fcis-14812	22	34	the	the	DET
fcis-14812	22	35	information	information	NOUN
fcis-14812	22	36	of	of	ADP
fcis-14812	22	37	the	the	DET
fcis-14812	22	38	query	query	NOUN
fcis-14812	22	39	image	image	NOUN
fcis-14812	22	40	itself	itself	PRON
fcis-14812	22	41	and	and	CCONJ
fcis-14812	22	42	the	the	DET
fcis-14812	22	43	information	information	NOUN
fcis-14812	22	44	between	between	ADP
fcis-14812	22	45	the	the	DET
fcis-14812	22	46	feature	feature	NOUN
fcis-14812	22	47	maps	map	NOUN
fcis-14812	22	48	at	at	ADP
fcis-14812	22	49	different	different	ADJ
fcis-14812	22	50	levels	level	NOUN
fcis-14812	22	51	.	.	PUNCT
fcis-14812	23	1	finally	finally	ADV
fcis-14812	23	2	,	,	PUNCT
fcis-14812	23	3	we	we	PRON
fcis-14812	23	4	use	use	VERB
fcis-14812	23	5	asff	asff	PROPN
fcis-14812	23	6	adaptive	adaptive	ADJ
fcis-14812	23	7	spatial	spatial	ADJ
fcis-14812	23	8	fusion	fusion	NOUN
fcis-14812	23	9	module	module	NOUN
fcis-14812	23	10	and	and	CCONJ
fcis-14812	23	11	dynamic	dynamic	ADJ
fcis-14812	23	12	similarity	similarity	NOUN
fcis-14812	23	13	metric	metric	ADJ
fcis-14812	23	14	module	module	NOUN
fcis-14812	23	15	for	for	ADP
fcis-14812	23	16	dynamically	dynamically	ADV
fcis-14812	23	17	fusing	fuse	VERB
fcis-14812	23	18	and	and	CCONJ
fcis-14812	23	19	comparing	compare	VERB
fcis-14812	23	20	the	the	DET
fcis-14812	23	21	features	feature	NOUN
fcis-14812	23	22	of	of	ADP
fcis-14812	23	23	query	query	NOUN
fcis-14812	23	24	images	image	NOUN
fcis-14812	23	25	and	and	CCONJ
fcis-14812	23	26	sample	sample	NOUN
fcis-14812	23	27	images	image	NOUN
fcis-14812	23	28	at	at	ADP
fcis-14812	23	29	different	different	ADJ
fcis-14812	23	30	levels	level	NOUN
fcis-14812	23	31	according	accord	VERB
fcis-14812	23	32	to	to	ADP
fcis-14812	23	33	the	the	DET
fcis-14812	23	34	correlation	correlation	NOUN
fcis-14812	23	35	between	between	ADP
fcis-14812	23	36	them	they	PRON
fcis-14812	23	37	and	and	CCONJ
fcis-14812	23	38	get	get	VERB
fcis-14812	23	39	the	the	DET
fcis-14812	23	40	final	final	ADJ
fcis-14812	23	41	counting	counting	NOUN
fcis-14812	23	42	results	result	NOUN
fcis-14812	23	43	to	to	PART
fcis-14812	23	44	improve	improve	VERB
fcis-14812	23	45	the	the	DET
fcis-14812	23	46	counting	counting	NOUN
fcis-14812	23	47	performance	performance	NOUN
fcis-14812	23	48	.	.	PUNCT
fcis-14812	24	1	overall	overall	ADV
fcis-14812	24	2	,	,	PUNCT
fcis-14812	24	3	the	the	DET
fcis-14812	24	4	main	main	ADJ
fcis-14812	24	5	contributions	contribution	NOUN
fcis-14812	24	6	of	of	ADP
fcis-14812	24	7	this	this	DET
fcis-14812	24	8	work	work	NOUN
fcis-14812	24	9	are	be	AUX
fcis-14812	24	10	as	as	SCONJ
fcis-14812	24	11	follows	follow	VERB
fcis-14812	24	12	:	:	PUNCT
fcis-14812	24	13	1	1	X
fcis-14812	24	14	.	.	X
fcis-14812	24	15	a	a	DET
fcis-14812	24	16	modified	modify	VERB
fcis-14812	24	17	network	network	NOUN
fcis-14812	24	18	structure	structure	NOUN
fcis-14812	24	19	dfenet	dfenet	NOUN
fcis-14812	24	20	based	base	VERB
fcis-14812	24	21	on	on	ADP
fcis-14812	24	22	a	a	DET
fcis-14812	24	23	modified	modify	VERB
fcis-14812	24	24	bilinear	bilinear	NOUN
fcis-14812	24	25	matching	match	VERB
fcis-14812	24	26	network+	network+	ADP
fcis-14812	24	27	(	(	PUNCT
fcis-14812	24	28	bmnet+	bmnet+	PROPN
fcis-14812	24	29	)	)	PUNCT
fcis-14812	24	30	is	be	AUX
fcis-14812	24	31	proposed	propose	VERB
fcis-14812	24	32	,	,	PUNCT
fcis-14812	24	33	which	which	DET
fcis-14812	24	34	extracts	extract	NOUN
fcis-14812	24	35	,	,	PUNCT
fcis-14812	24	36	enhances	enhance	NOUN
fcis-14812	24	37	&	&	CCONJ
fcis-14812	24	38	fuses	fuse	NOUN
fcis-14812	24	39	features	feature	NOUN
fcis-14812	24	40	of	of	ADP
fcis-14812	24	41	different	different	ADJ
fcis-14812	24	42	scales	scale	NOUN
fcis-14812	24	43	to	to	PART
fcis-14812	24	44	improve	improve	VERB
fcis-14812	24	45	the	the	DET
fcis-14812	24	46	counting	counting	NOUN
fcis-14812	24	47	performance	performance	NOUN
fcis-14812	24	48	of	of	ADP
fcis-14812	24	49	the	the	DET
fcis-14812	24	50	network	network	NOUN
fcis-14812	24	51	.	.	PUNCT
fcis-14812	25	1	2	2	X
fcis-14812	25	2	.	.	X
fcis-14812	25	3	a	a	DET
fcis-14812	25	4	structured	structured	ADJ
fcis-14812	25	5	feature	feature	NOUN
fcis-14812	25	6	enhancement	enhancement	NOUN
fcis-14812	25	7	module	module	NOUN
fcis-14812	25	8	(	(	PUNCT
fcis-14812	25	9	sfem	sfem	ADJ
fcis-14812	25	10	)	)	PUNCT
fcis-14812	25	11	is	be	AUX
fcis-14812	25	12	introduced	introduce	VERB
fcis-14812	25	13	,	,	PUNCT
fcis-14812	25	14	which	which	PRON
fcis-14812	25	15	enhances	enhance	VERB
fcis-14812	25	16	the	the	DET
fcis-14812	25	17	representation	representation	NOUN
fcis-14812	25	18	of	of	ADP
fcis-14812	25	19	features	feature	NOUN
fcis-14812	25	20	by	by	ADP
fcis-14812	25	21	enabling	enable	VERB
fcis-14812	25	22	features	feature	NOUN
fcis-14812	25	23	of	of	ADP
fcis-14812	25	24	different	different	ADJ
fcis-14812	25	25	scales	scale	NOUN
fcis-14812	25	26	to	to	PART
fcis-14812	25	27	transfer	transfer	VERB
fcis-14812	25	28	information	information	NOUN
fcis-14812	25	29	to	to	ADP
fcis-14812	25	30	each	each	DET
fcis-14812	25	31	other	other	ADJ
fcis-14812	25	32	through	through	ADP
fcis-14812	25	33	an	an	DET
fcis-14812	25	34	algorithm	algorithm	NOUN
fcis-14812	25	35	based	base	VERB
fcis-14812	25	36	on	on	ADP
fcis-14812	25	37	the	the	DET
fcis-14812	25	38	principle	principle	NOUN
fcis-14812	25	39	of	of	ADP
fcis-14812	25	40	conditional	conditional	ADJ
fcis-14812	25	41	random	random	ADJ
fcis-14812	25	42	fields	field	NOUN
fcis-14812	25	43	.	.	PUNCT
fcis-14812	26	1	3	3	X
fcis-14812	26	2	.	.	X
fcis-14812	26	3	adaptively	adaptively	ADV
fcis-14812	26	4	spatial	spatial	ADJ
fcis-14812	26	5	feature	feature	NOUN
fcis-14812	26	6	fusion	fusion	NOUN
fcis-14812	26	7	module	module	NOUN
fcis-14812	26	8	(	(	PUNCT
fcis-14812	26	9	asff	asff	NOUN
fcis-14812	26	10	)	)	PUNCT
fcis-14812	26	11	is	be	AUX
fcis-14812	26	12	introduced	introduce	VERB
fcis-14812	26	13	,	,	PUNCT
fcis-14812	26	14	which	which	PRON
fcis-14812	26	15	can	can	AUX
fcis-14812	26	16	adaptively	adaptively	ADV
fcis-14812	26	17	learn	learn	VERB
fcis-14812	26	18	the	the	DET
fcis-14812	26	19	spatial	spatial	ADJ
fcis-14812	26	20	weights	weight	NOUN
fcis-14812	26	21	of	of	ADP
fcis-14812	26	22	feature	feature	NOUN
fcis-14812	26	23	maps	map	NOUN
fcis-14812	26	24	at	at	ADP
fcis-14812	26	25	different	different	ADJ
fcis-14812	26	26	scales	scale	NOUN
fcis-14812	26	27	,	,	PUNCT
fcis-14812	26	28	thus	thus	ADV
fcis-14812	26	29	improving	improve	VERB
fcis-14812	26	30	the	the	DET
fcis-14812	26	31	robustness	robustness	NOUN
fcis-14812	26	32	of	of	ADP
fcis-14812	26	33	the	the	DET
fcis-14812	26	34	feature	feature	NOUN
fcis-14812	26	35	fusion	fusion	NOUN
fcis-14812	26	36	network	network	NOUN
fcis-14812	26	37	.	.	PUNCT
fcis-14812	27	1	2	2	X
fcis-14812	27	2	.	.	X
fcis-14812	27	3	related	relate	VERB
fcis-14812	27	4	work	work	NOUN
fcis-14812	27	5	2.1	2.1	NUM
fcis-14812	27	6	.	.	PUNCT
fcis-14812	27	7	object	object	NOUN
fcis-14812	27	8	counting	count	VERB
fcis-14812	27	9	related	relate	VERB
fcis-14812	27	10	work	work	NOUN
fcis-14812	27	11	object	object	NOUN
fcis-14812	27	12	counting	counting	NOUN
fcis-14812	27	13	is	be	AUX
fcis-14812	27	14	a	a	DET
fcis-14812	27	15	computer	computer	NOUN
fcis-14812	27	16	vision	vision	NOUN
fcis-14812	27	17	task	task	NOUN
fcis-14812	27	18	involving	involve	VERB
fcis-14812	27	19	the	the	DET
fcis-14812	27	20	estimation	estimation	NOUN
fcis-14812	27	21	of	of	ADP
fcis-14812	27	22	the	the	DET
fcis-14812	27	23	number	number	NOUN
fcis-14812	27	24	of	of	ADP
fcis-14812	27	25	targets	target	NOUN
fcis-14812	27	26	in	in	ADP
fcis-14812	27	27	an	an	DET
fcis-14812	27	28	image	image	NOUN
fcis-14812	27	29	,	,	PUNCT
fcis-14812	27	30	which	which	PRON
fcis-14812	27	31	has	have	VERB
fcis-14812	27	32	a	a	DET
fcis-14812	27	33	wide	wide	ADJ
fcis-14812	27	34	range	range	NOUN
fcis-14812	27	35	of	of	ADP
fcis-14812	27	36	applications	application	NOUN
fcis-14812	27	37	in	in	ADP
fcis-14812	27	38	various	various	ADJ
fcis-14812	27	39	fields	field	NOUN
fcis-14812	27	40	of	of	ADP
fcis-14812	27	41	social	social	ADJ
fcis-14812	27	42	production	production	NOUN
fcis-14812	27	43	,	,	PUNCT
fcis-14812	27	44	such	such	ADJ
fcis-14812	27	45	as	as	ADP
fcis-14812	27	46	traffic	traffic	NOUN
fcis-14812	27	47	counting	counting	NOUN
fcis-14812	27	48	,	,	PUNCT
fcis-14812	27	49	crowd	crowd	VERB
fcis-14812	27	50	density	density	NOUN
fcis-14812	27	51	estimation	estimation	NOUN
fcis-14812	27	52	,	,	PUNCT
fcis-14812	27	53	biological	biological	ADJ
fcis-14812	27	54	population	population	NOUN
fcis-14812	27	55	counting	counting	NOUN
fcis-14812	27	56	and	and	CCONJ
fcis-14812	27	57	so	so	ADV
fcis-14812	27	58	on	on	ADV
fcis-14812	27	59	.	.	PUNCT
fcis-14812	28	1	currently	currently	ADV
fcis-14812	28	2	,	,	PUNCT
fcis-14812	28	3	the	the	DET
fcis-14812	28	4	work	work	NOUN
fcis-14812	28	5	related	relate	VERB
fcis-14812	28	6	to	to	PART
fcis-14812	28	7	object	object	VERB
fcis-14812	28	8	counting	counting	NOUN
fcis-14812	28	9	can	can	AUX
fcis-14812	28	10	be	be	AUX
fcis-14812	28	11	categorized	categorize	VERB
fcis-14812	28	12	into	into	ADP
fcis-14812	28	13	two	two	NUM
fcis-14812	28	14	types	type	NOUN
fcis-14812	28	15	according	accord	VERB
fcis-14812	28	16	to	to	ADP
fcis-14812	28	17	the	the	DET
fcis-14812	28	18	different	different	ADJ
fcis-14812	28	19	counting	counting	NOUN
fcis-14812	28	20	methods	method	NOUN
fcis-14812	28	21	:	:	PUNCT
fcis-14812	28	22	detection	detection	NOUN
fcis-14812	28	23	-	-	PUNCT
fcis-14812	28	24	based	base	VERB
fcis-14812	28	25	71	71	NUM
fcis-14812	28	26	counting	counting	NOUN
fcis-14812	28	27	methods	method	NOUN
fcis-14812	28	28	[	[	X
fcis-14812	28	29	1][2][3][4	1][2][3][4	NUM
fcis-14812	28	30	]	]	X
fcis-14812	28	31	,	,	PUNCT
fcis-14812	28	32	and	and	CCONJ
fcis-14812	28	33	regression	regression	NOUN
fcis-14812	28	34	-	-	PUNCT
fcis-14812	28	35	based	base	VERB
fcis-14812	28	36	counting	counting	NOUN
fcis-14812	28	37	methods	method	NOUN
fcis-14812	28	38	[	[	X
fcis-14812	28	39	5][6][8][9	5][6][8][9	NUM
fcis-14812	28	40	]	]	PUNCT
fcis-14812	28	41	.	.	PUNCT
fcis-14812	29	1	detection	detection	NOUN
fcis-14812	29	2	-	-	PUNCT
fcis-14812	29	3	based	base	VERB
fcis-14812	29	4	counting	counting	NOUN
fcis-14812	29	5	methods	method	NOUN
fcis-14812	29	6	are	be	AUX
fcis-14812	29	7	methods	method	NOUN
fcis-14812	29	8	that	that	PRON
fcis-14812	29	9	use	use	VERB
fcis-14812	29	10	techniques	technique	NOUN
fcis-14812	29	11	of	of	ADP
fcis-14812	29	12	target	target	NOUN
fcis-14812	29	13	detection	detection	NOUN
fcis-14812	29	14	for	for	ADP
fcis-14812	29	15	counting	counting	NOUN
fcis-14812	29	16	,	,	PUNCT
fcis-14812	29	17	such	such	ADJ
fcis-14812	29	18	as	as	ADP
fcis-14812	29	19	histograms	histogram	NOUN
fcis-14812	29	20	of	of	ADP
fcis-14812	29	21	oriented	orient	VERB
fcis-14812	29	22	gradients	gradient	NOUN
fcis-14812	29	23	(	(	PUNCT
fcis-14812	29	24	hog	hog	NOUN
fcis-14812	29	25	)	)	PUNCT
fcis-14812	30	1	[	[	X
fcis-14812	30	2	10	10	NUM
fcis-14812	30	3	]	]	PUNCT
fcis-14812	30	4	,	,	PUNCT
fcis-14812	30	5	region	region	NOUN
fcis-14812	30	6	-	-	PUNCT
fcis-14812	30	7	cnn	cnn	PROPN
fcis-14812	30	8	(	(	PUNCT
fcis-14812	30	9	rcnn	rcnn	PROPN
fcis-14812	30	10	)	)	PUNCT
fcis-14812	31	1	[	[	X
fcis-14812	31	2	11	11	NUM
fcis-14812	31	3	]	]	PUNCT
fcis-14812	31	4	,	,	PUNCT
fcis-14812	31	5	you	you	PRON
fcis-14812	31	6	only	only	ADV
fcis-14812	31	7	look	look	VERB
fcis-14812	31	8	once	once	ADV
fcis-14812	31	9	(	(	PUNCT
fcis-14812	31	10	yolo	yolo	ADJ
fcis-14812	31	11	)	)	PUNCT
fcis-14812	31	12	[	[	PUNCT
fcis-14812	31	13	12	12	NUM
fcis-14812	31	14	]	]	PUNCT
fcis-14812	31	15	,	,	PUNCT
fcis-14812	31	16	etc	etc	X
fcis-14812	31	17	.	.	X
fcis-14812	32	1	the	the	DET
fcis-14812	32	2	basic	basic	ADJ
fcis-14812	32	3	idea	idea	NOUN
fcis-14812	32	4	of	of	ADP
fcis-14812	32	5	this	this	DET
fcis-14812	32	6	method	method	NOUN
fcis-14812	32	7	is	be	AUX
fcis-14812	32	8	to	to	PART
fcis-14812	32	9	first	first	ADV
fcis-14812	32	10	detect	detect	VERB
fcis-14812	32	11	each	each	DET
fcis-14812	32	12	target	target	NOUN
fcis-14812	32	13	in	in	ADP
fcis-14812	32	14	the	the	DET
fcis-14812	32	15	image	image	NOUN
fcis-14812	32	16	and	and	CCONJ
fcis-14812	32	17	then	then	ADV
fcis-14812	32	18	count	count	VERB
fcis-14812	32	19	the	the	DET
fcis-14812	32	20	number	number	NOUN
fcis-14812	32	21	of	of	ADP
fcis-14812	32	22	targets	target	NOUN
fcis-14812	32	23	in	in	ADP
fcis-14812	32	24	each	each	DET
fcis-14812	32	25	category	category	NOUN
fcis-14812	32	26	based	base	VERB
fcis-14812	32	27	on	on	ADP
fcis-14812	32	28	the	the	DET
fcis-14812	32	29	detection	detection	NOUN
fcis-14812	32	30	results	result	VERB
fcis-14812	32	31	.	.	PUNCT
fcis-14812	33	1	the	the	DET
fcis-14812	33	2	advantage	advantage	NOUN
fcis-14812	33	3	of	of	ADP
fcis-14812	33	4	this	this	DET
fcis-14812	33	5	method	method	NOUN
fcis-14812	33	6	is	be	AUX
fcis-14812	33	7	that	that	SCONJ
fcis-14812	33	8	the	the	DET
fcis-14812	33	9	position	position	NOUN
fcis-14812	33	10	information	information	NOUN
fcis-14812	33	11	and	and	CCONJ
fcis-14812	33	12	target	target	NOUN
fcis-14812	33	13	scale	scale	NOUN
fcis-14812	33	14	information	information	NOUN
fcis-14812	33	15	of	of	ADP
fcis-14812	33	16	different	different	ADJ
fcis-14812	33	17	targets	target	NOUN
fcis-14812	33	18	can	can	AUX
fcis-14812	33	19	be	be	AUX
fcis-14812	33	20	obtained	obtain	VERB
fcis-14812	33	21	while	while	SCONJ
fcis-14812	33	22	counting	counting	NOUN
fcis-14812	33	23	,	,	PUNCT
fcis-14812	33	24	so	so	SCONJ
fcis-14812	33	25	that	that	SCONJ
fcis-14812	33	26	the	the	DET
fcis-14812	33	27	targets	target	NOUN
fcis-14812	33	28	can	can	AUX
fcis-14812	33	29	be	be	AUX
fcis-14812	33	30	analyzed	analyze	VERB
fcis-14812	33	31	in	in	ADP
fcis-14812	33	32	more	more	ADJ
fcis-14812	33	33	detail	detail	NOUN
fcis-14812	33	34	;	;	PUNCT
fcis-14812	33	35	moreover	moreover	ADV
fcis-14812	33	36	,	,	PUNCT
fcis-14812	33	37	this	this	DET
fcis-14812	33	38	method	method	NOUN
fcis-14812	33	39	is	be	AUX
fcis-14812	33	40	better	well	ADV
fcis-14812	33	41	adapted	adapt	VERB
fcis-14812	33	42	to	to	ADP
fcis-14812	33	43	the	the	DET
fcis-14812	33	44	targets	target	NOUN
fcis-14812	33	45	with	with	ADP
fcis-14812	33	46	multi	multi	ADJ
fcis-14812	33	47	-	-	ADJ
fcis-14812	33	48	scale	scale	ADJ
fcis-14812	33	49	variations	variation	NOUN
fcis-14812	33	50	,	,	PUNCT
fcis-14812	33	51	and	and	CCONJ
fcis-14812	33	52	it	it	PRON
fcis-14812	33	53	can	can	AUX
fcis-14812	33	54	deal	deal	VERB
fcis-14812	33	55	with	with	ADP
fcis-14812	33	56	targets	target	NOUN
fcis-14812	33	57	of	of	ADP
fcis-14812	33	58	different	different	ADJ
fcis-14812	33	59	sizes	size	NOUN
fcis-14812	33	60	.	.	PUNCT
fcis-14812	34	1	however	however	ADV
fcis-14812	34	2	,	,	PUNCT
fcis-14812	34	3	the	the	DET
fcis-14812	34	4	disadvantages	disadvantage	NOUN
fcis-14812	34	5	of	of	ADP
fcis-14812	34	6	this	this	DET
fcis-14812	34	7	type	type	NOUN
fcis-14812	34	8	of	of	ADP
fcis-14812	34	9	method	method	NOUN
fcis-14812	34	10	are	be	AUX
fcis-14812	34	11	that	that	SCONJ
fcis-14812	34	12	it	it	PRON
fcis-14812	34	13	has	have	VERB
fcis-14812	34	14	a	a	DET
fcis-14812	34	15	large	large	ADJ
fcis-14812	34	16	amount	amount	NOUN
fcis-14812	34	17	of	of	ADP
fcis-14812	34	18	computation	computation	NOUN
fcis-14812	34	19	and	and	CCONJ
fcis-14812	34	20	lacks	lack	VERB
fcis-14812	34	21	real	real	ADJ
fcis-14812	34	22	-	-	PUNCT
fcis-14812	34	23	time	time	NOUN
fcis-14812	34	24	and	and	CCONJ
fcis-14812	34	25	convenience	convenience	NOUN
fcis-14812	34	26	for	for	ADP
fcis-14812	34	27	the	the	DET
fcis-14812	34	28	counting	counting	NOUN
fcis-14812	34	29	task	task	NOUN
fcis-14812	34	30	;	;	PUNCT
fcis-14812	34	31	moreover	moreover	ADV
fcis-14812	34	32	,	,	PUNCT
fcis-14812	34	33	this	this	DET
fcis-14812	34	34	method	method	NOUN
fcis-14812	34	35	also	also	ADV
fcis-14812	34	36	tends	tend	VERB
fcis-14812	34	37	to	to	PART
fcis-14812	34	38	show	show	VERB
fcis-14812	34	39	poor	poor	ADJ
fcis-14812	34	40	counting	counting	NOUN
fcis-14812	34	41	performance	performance	NOUN
fcis-14812	34	42	when	when	SCONJ
fcis-14812	34	43	the	the	DET
fcis-14812	34	44	target	target	NOUN
fcis-14812	34	45	receives	receive	VERB
fcis-14812	34	46	an	an	DET
fcis-14812	34	47	occlusion	occlusion	NOUN
fcis-14812	34	48	,	,	PUNCT
fcis-14812	34	49	and	and	CCONJ
fcis-14812	34	50	is	be	AUX
fcis-14812	34	51	unable	unable	ADJ
fcis-14812	34	52	to	to	PART
fcis-14812	34	53	accurately	accurately	ADV
fcis-14812	34	54	recognize	recognize	VERB
fcis-14812	34	55	the	the	DET
fcis-14812	34	56	occluded	occluded	ADJ
fcis-14812	34	57	target	target	NOUN
fcis-14812	34	58	.	.	PUNCT
fcis-14812	35	1	regression	regression	NOUN
fcis-14812	35	2	-	-	PUNCT
fcis-14812	35	3	based	base	VERB
fcis-14812	35	4	counting	counting	NOUN
fcis-14812	35	5	method	method	NOUN
fcis-14812	35	6	refers	refer	VERB
fcis-14812	35	7	to	to	ADP
fcis-14812	35	8	the	the	DET
fcis-14812	35	9	method	method	NOUN
fcis-14812	35	10	that	that	PRON
fcis-14812	35	11	generally	generally	ADV
fcis-14812	35	12	relies	rely	VERB
fcis-14812	35	13	on	on	ADP
fcis-14812	35	14	some	some	DET
fcis-14812	35	15	texture	texture	NOUN
fcis-14812	35	16	features	feature	NOUN
fcis-14812	35	17	in	in	ADP
fcis-14812	35	18	the	the	DET
fcis-14812	35	19	image	image	NOUN
fcis-14812	35	20	,	,	PUNCT
fcis-14812	35	21	and	and	CCONJ
fcis-14812	35	22	then	then	ADV
fcis-14812	35	23	uses	use	VERB
fcis-14812	35	24	linear	linear	PROPN
fcis-14812	35	25	regression	regression	NOUN
fcis-14812	35	26	,	,	PUNCT
fcis-14812	35	27	neural	neural	ADJ
fcis-14812	35	28	networks	network	NOUN
fcis-14812	35	29	,	,	PUNCT
fcis-14812	35	30	and	and	CCONJ
fcis-14812	35	31	other	other	ADJ
fcis-14812	35	32	ways	way	NOUN
fcis-14812	35	33	to	to	PART
fcis-14812	35	34	establish	establish	VERB
fcis-14812	35	35	the	the	DET
fcis-14812	35	36	mapping	mapping	NOUN
fcis-14812	35	37	relationship	relationship	NOUN
fcis-14812	35	38	between	between	ADP
fcis-14812	35	39	the	the	DET
fcis-14812	35	40	features	feature	NOUN
fcis-14812	35	41	and	and	CCONJ
fcis-14812	35	42	the	the	DET
fcis-14812	35	43	target	target	NOUN
fcis-14812	35	44	,	,	PUNCT
fcis-14812	35	45	to	to	PART
fcis-14812	35	46	directly	directly	ADV
fcis-14812	35	47	obtain	obtain	VERB
fcis-14812	35	48	the	the	DET
fcis-14812	35	49	statistical	statistical	ADJ
fcis-14812	35	50	results	result	NOUN
fcis-14812	35	51	of	of	ADP
fcis-14812	35	52	the	the	DET
fcis-14812	35	53	number	number	NOUN
fcis-14812	35	54	of	of	ADP
fcis-14812	35	55	targets	target	NOUN
fcis-14812	35	56	.	.	PUNCT
fcis-14812	36	1	the	the	DET
fcis-14812	36	2	advantage	advantage	NOUN
fcis-14812	36	3	of	of	ADP
fcis-14812	36	4	the	the	DET
fcis-14812	36	5	regressionbased	regressionbase	VERB
fcis-14812	36	6	counting	counting	NOUN
fcis-14812	36	7	method	method	NOUN
fcis-14812	36	8	is	be	AUX
fcis-14812	36	9	that	that	SCONJ
fcis-14812	36	10	it	it	PRON
fcis-14812	36	11	can	can	AUX
fcis-14812	36	12	directly	directly	ADV
fcis-14812	36	13	output	output	VERB
fcis-14812	36	14	the	the	DET
fcis-14812	36	15	number	number	NOUN
fcis-14812	36	16	of	of	ADP
fcis-14812	36	17	targets	target	NOUN
fcis-14812	36	18	without	without	ADP
fcis-14812	36	19	detecting	detect	VERB
fcis-14812	36	20	the	the	DET
fcis-14812	36	21	targets	target	NOUN
fcis-14812	36	22	,	,	PUNCT
fcis-14812	36	23	so	so	SCONJ
fcis-14812	36	24	it	it	PRON
fcis-14812	36	25	tends	tend	VERB
fcis-14812	36	26	to	to	PART
fcis-14812	36	27	have	have	VERB
fcis-14812	36	28	a	a	DET
fcis-14812	36	29	small	small	ADJ
fcis-14812	36	30	amount	amount	NOUN
fcis-14812	36	31	of	of	ADP
fcis-14812	36	32	computation	computation	NOUN
fcis-14812	36	33	and	and	CCONJ
fcis-14812	36	34	high	high	ADJ
fcis-14812	36	35	real	real	ADJ
fcis-14812	36	36	-	-	PUNCT
fcis-14812	36	37	time	time	NOUN
fcis-14812	36	38	,	,	PUNCT
fcis-14812	36	39	which	which	PRON
fcis-14812	36	40	is	be	AUX
fcis-14812	36	41	suitable	suitable	ADJ
fcis-14812	36	42	for	for	ADP
fcis-14812	36	43	fast	fast	ADJ
fcis-14812	36	44	processing	processing	NOUN
fcis-14812	36	45	of	of	ADP
fcis-14812	36	46	counting	counting	NOUN
fcis-14812	36	47	tasks	task	NOUN
fcis-14812	36	48	.	.	PUNCT
fcis-14812	37	1	however	however	ADV
fcis-14812	37	2	,	,	PUNCT
fcis-14812	37	3	the	the	DET
fcis-14812	37	4	disadvantage	disadvantage	NOUN
fcis-14812	37	5	is	be	AUX
fcis-14812	37	6	that	that	SCONJ
fcis-14812	37	7	it	it	PRON
fcis-14812	37	8	is	be	AUX
fcis-14812	37	9	less	less	ADV
fcis-14812	37	10	adaptable	adaptable	ADJ
fcis-14812	37	11	to	to	ADP
fcis-14812	37	12	targets	target	NOUN
fcis-14812	37	13	of	of	ADP
fcis-14812	37	14	different	different	ADJ
fcis-14812	37	15	scales	scale	NOUN
fcis-14812	37	16	,	,	PUNCT
fcis-14812	37	17	and	and	CCONJ
fcis-14812	37	18	can	can	AUX
fcis-14812	37	19	not	not	PART
fcis-14812	37	20	count	count	VERB
fcis-14812	37	21	well	well	ADV
fcis-14812	37	22	in	in	ADP
fcis-14812	37	23	the	the	DET
fcis-14812	37	24	face	face	NOUN
fcis-14812	37	25	of	of	ADP
fcis-14812	37	26	images	image	NOUN
fcis-14812	37	27	with	with	ADP
fcis-14812	37	28	large	large	ADJ
fcis-14812	37	29	target	target	NOUN
fcis-14812	37	30	scales	scale	NOUN
fcis-14812	37	31	(	(	PUNCT
fcis-14812	37	32	e.g.	e.g.	ADV
fcis-14812	37	33	,	,	PUNCT
fcis-14812	37	34	dense	dense	ADJ
fcis-14812	37	35	crowds	crowd	NOUN
fcis-14812	37	36	in	in	ADP
fcis-14812	37	37	perspective	perspective	NOUN
fcis-14812	37	38	)	)	PUNCT
fcis-14812	37	39	,	,	PUNCT
fcis-14812	37	40	because	because	SCONJ
fcis-14812	37	41	the	the	DET
fcis-14812	37	42	target	target	NOUN
fcis-14812	37	43	features	feature	VERB
fcis-14812	37	44	in	in	ADP
fcis-14812	37	45	this	this	DET
fcis-14812	37	46	case	case	NOUN
fcis-14812	37	47	are	be	AUX
fcis-14812	37	48	often	often	ADV
fcis-14812	37	49	not	not	PART
fcis-14812	37	50	obvious	obvious	ADJ
fcis-14812	37	51	,	,	PUNCT
fcis-14812	37	52	and	and	CCONJ
fcis-14812	37	53	it	it	PRON
fcis-14812	37	54	is	be	AUX
fcis-14812	37	55	difficult	difficult	ADJ
fcis-14812	37	56	to	to	PART
fcis-14812	37	57	establish	establish	VERB
fcis-14812	37	58	an	an	DET
fcis-14812	37	59	effective	effective	ADJ
fcis-14812	37	60	mapping	mapping	NOUN
fcis-14812	37	61	relationship	relationship	NOUN
fcis-14812	37	62	.	.	PUNCT
fcis-14812	38	1	2.2	2.2	NUM
fcis-14812	38	2	.	.	PUNCT
fcis-14812	39	1	class	class	NOUN
fcis-14812	39	2	agnostic	agnostic	ADJ
fcis-14812	39	3	counting	counting	NOUN
fcis-14812	39	4	related	relate	VERB
fcis-14812	39	5	work	work	NOUN
fcis-14812	39	6	most	most	ADJ
fcis-14812	39	7	of	of	ADP
fcis-14812	39	8	the	the	DET
fcis-14812	39	9	current	current	ADJ
fcis-14812	39	10	object	object	NOUN
fcis-14812	39	11	counting	counting	NOUN
fcis-14812	39	12	methods	method	NOUN
fcis-14812	39	13	need	need	VERB
fcis-14812	39	14	to	to	PART
fcis-14812	39	15	know	know	VERB
fcis-14812	39	16	the	the	DET
fcis-14812	39	17	class	class	NOUN
fcis-14812	39	18	information	information	NOUN
fcis-14812	39	19	of	of	ADP
fcis-14812	39	20	the	the	DET
fcis-14812	39	21	target	target	NOUN
fcis-14812	39	22	,	,	PUNCT
fcis-14812	39	23	such	such	ADJ
fcis-14812	39	24	as	as	ADP
fcis-14812	39	25	detection	detection	NOUN
fcis-14812	39	26	-	-	PUNCT
fcis-14812	39	27	based	base	VERB
fcis-14812	39	28	counting	counting	NOUN
fcis-14812	39	29	methods	method	NOUN
fcis-14812	39	30	and	and	CCONJ
fcis-14812	39	31	regression	regression	NOUN
fcis-14812	39	32	-	-	PUNCT
fcis-14812	39	33	based	base	VERB
fcis-14812	39	34	counting	counting	NOUN
fcis-14812	39	35	methods	method	NOUN
fcis-14812	39	36	.	.	PUNCT
fcis-14812	40	1	these	these	DET
fcis-14812	40	2	methods	method	NOUN
fcis-14812	40	3	have	have	VERB
fcis-14812	40	4	some	some	DET
fcis-14812	40	5	limitations	limitation	NOUN
fcis-14812	40	6	,	,	PUNCT
fcis-14812	40	7	such	such	ADJ
fcis-14812	40	8	as	as	ADP
fcis-14812	40	9	requiring	require	VERB
fcis-14812	40	10	a	a	DET
fcis-14812	40	11	large	large	ADJ
fcis-14812	40	12	amount	amount	NOUN
fcis-14812	40	13	of	of	ADP
fcis-14812	40	14	labeled	label	VERB
fcis-14812	40	15	data	datum	NOUN
fcis-14812	40	16	,	,	PUNCT
fcis-14812	40	17	and	and	CCONJ
fcis-14812	40	18	difficulty	difficulty	NOUN
fcis-14812	40	19	in	in	ADP
fcis-14812	40	20	dealing	deal	VERB
fcis-14812	40	21	with	with	ADP
fcis-14812	40	22	occlusion	occlusion	NOUN
fcis-14812	40	23	and	and	CCONJ
fcis-14812	40	24	overlapping	overlapping	NOUN
fcis-14812	40	25	.	.	PUNCT
fcis-14812	41	1	therefore	therefore	ADV
fcis-14812	41	2	,	,	PUNCT
fcis-14812	41	3	some	some	DET
fcis-14812	41	4	researchers	researcher	NOUN
fcis-14812	41	5	have	have	AUX
fcis-14812	41	6	proposed	propose	VERB
fcis-14812	41	7	the	the	DET
fcis-14812	41	8	concept	concept	NOUN
fcis-14812	41	9	of	of	ADP
fcis-14812	41	10	class	class	NOUN
fcis-14812	41	11	-	-	PUNCT
fcis-14812	41	12	agnostic	agnostic	ADJ
fcis-14812	41	13	counting	counting	NOUN
fcis-14812	41	14	(	(	PUNCT
fcis-14812	41	15	cac	cac	PROPN
fcis-14812	41	16	)	)	PUNCT
fcis-14812	41	17	,	,	PUNCT
fcis-14812	41	18	which	which	PRON
fcis-14812	41	19	does	do	AUX
fcis-14812	41	20	not	not	PART
fcis-14812	41	21	need	need	VERB
fcis-14812	41	22	to	to	PART
fcis-14812	41	23	know	know	VERB
fcis-14812	41	24	the	the	DET
fcis-14812	41	25	class	class	NOUN
fcis-14812	41	26	of	of	ADP
fcis-14812	41	27	the	the	DET
fcis-14812	41	28	target	target	NOUN
fcis-14812	41	29	,	,	PUNCT
fcis-14812	41	30	but	but	CCONJ
fcis-14812	41	31	only	only	ADV
fcis-14812	41	32	how	how	SCONJ
fcis-14812	41	33	many	many	ADJ
fcis-14812	41	34	targets	target	NOUN
fcis-14812	41	35	are	be	AUX
fcis-14812	41	36	in	in	ADP
fcis-14812	41	37	the	the	DET
fcis-14812	41	38	image	image	NOUN
fcis-14812	41	39	.	.	PUNCT
fcis-14812	42	1	this	this	DET
fcis-14812	42	2	method	method	NOUN
fcis-14812	42	3	can	can	AUX
fcis-14812	42	4	be	be	AUX
fcis-14812	42	5	applied	apply	VERB
fcis-14812	42	6	to	to	ADP
fcis-14812	42	7	a	a	DET
fcis-14812	42	8	variety	variety	NOUN
fcis-14812	42	9	of	of	ADP
fcis-14812	42	10	scenarios	scenario	NOUN
fcis-14812	42	11	,	,	PUNCT
fcis-14812	42	12	such	such	ADJ
fcis-14812	42	13	as	as	ADP
fcis-14812	42	14	medical	medical	ADJ
fcis-14812	42	15	images	image	NOUN
fcis-14812	42	16	,	,	PUNCT
fcis-14812	42	17	satellite	satellite	NOUN
fcis-14812	42	18	images	image	NOUN
fcis-14812	42	19	,	,	PUNCT
fcis-14812	42	20	uav	uav	PROPN
fcis-14812	42	21	images	image	NOUN
fcis-14812	42	22	,	,	PUNCT
fcis-14812	42	23	etc	etc	X
fcis-14812	42	24	.	.	X
fcis-14812	42	25	,	,	PUNCT
fcis-14812	42	26	with	with	ADP
fcis-14812	42	27	better	well	ADJ
fcis-14812	42	28	generalization	generalization	NOUN
fcis-14812	42	29	and	and	CCONJ
fcis-14812	42	30	robustness	robustness	NOUN
fcis-14812	42	31	.	.	PUNCT
fcis-14812	43	1	currently	currently	ADV
fcis-14812	43	2	,	,	PUNCT
fcis-14812	43	3	there	there	PRON
fcis-14812	43	4	are	be	VERB
fcis-14812	43	5	several	several	ADJ
fcis-14812	43	6	related	relate	VERB
fcis-14812	43	7	works	work	NOUN
fcis-14812	43	8	on	on	ADP
fcis-14812	43	9	class	class	NOUN
fcis-14812	43	10	agnostic	agnostic	ADJ
fcis-14812	43	11	counting	counting	NOUN
fcis-14812	43	12	:	:	PUNCT
fcis-14812	43	13	lu	lu	VERB
fcis-14812	43	14	et	et	NOUN
fcis-14812	43	15	al	al	PROPN
fcis-14812	44	1	[	[	X
fcis-14812	44	2	13	13	NUM
fcis-14812	44	3	]	]	PUNCT
fcis-14812	44	4	were	be	AUX
fcis-14812	44	5	the	the	DET
fcis-14812	44	6	first	first	ADJ
fcis-14812	44	7	to	to	PART
fcis-14812	44	8	propose	propose	VERB
fcis-14812	44	9	the	the	DET
fcis-14812	44	10	concept	concept	NOUN
fcis-14812	44	11	of	of	ADP
fcis-14812	44	12	class	class	NOUN
fcis-14812	44	13	agnostic	agnostic	ADJ
fcis-14812	44	14	counting	counting	NOUN
fcis-14812	44	15	,	,	PUNCT
fcis-14812	44	16	and	and	CCONJ
fcis-14812	44	17	they	they	PRON
fcis-14812	44	18	implemented	implement	VERB
fcis-14812	44	19	class	class	NOUN
fcis-14812	44	20	agnostic	agnostic	ADJ
fcis-14812	44	21	counting	counting	NOUN
fcis-14812	44	22	through	through	ADP
fcis-14812	44	23	a	a	DET
fcis-14812	44	24	generic	generic	ADJ
fcis-14812	44	25	matching	matching	NOUN
fcis-14812	44	26	network	network	NOUN
fcis-14812	44	27	(	(	PUNCT
fcis-14812	44	28	gmn	gmn	NOUN
fcis-14812	44	29	)	)	PUNCT
fcis-14812	44	30	model	model	NOUN
fcis-14812	44	31	that	that	PRON
fcis-14812	44	32	was	be	AUX
fcis-14812	44	33	pre	pre	VERB
fcis-14812	44	34	-	-	VERB
fcis-14812	44	35	trained	train	VERB
fcis-14812	44	36	on	on	ADP
fcis-14812	44	37	a	a	DET
fcis-14812	44	38	labeled	label	VERB
fcis-14812	44	39	target	target	NOUN
fcis-14812	44	40	tracking	tracking	NOUN
fcis-14812	44	41	video	video	NOUN
fcis-14812	44	42	dataset	dataset	NOUN
fcis-14812	44	43	.	.	PUNCT
fcis-14812	45	1	the	the	DET
fcis-14812	45	2	advantage	advantage	NOUN
fcis-14812	45	3	of	of	ADP
fcis-14812	45	4	this	this	DET
fcis-14812	45	5	model	model	NOUN
fcis-14812	45	6	is	be	AUX
fcis-14812	45	7	that	that	SCONJ
fcis-14812	45	8	with	with	ADP
fcis-14812	45	9	less	less	ADJ
fcis-14812	45	10	sample	sample	NOUN
fcis-14812	45	11	learning	learning	NOUN
fcis-14812	45	12	,	,	PUNCT
fcis-14812	45	13	it	it	PRON
fcis-14812	45	14	does	do	AUX
fcis-14812	45	15	not	not	PART
fcis-14812	45	16	need	need	VERB
fcis-14812	45	17	to	to	PART
fcis-14812	45	18	be	be	AUX
fcis-14812	45	19	trained	train	VERB
fcis-14812	45	20	with	with	ADP
fcis-14812	45	21	a	a	DET
fcis-14812	45	22	large	large	ADJ
fcis-14812	45	23	training	training	NOUN
fcis-14812	45	24	set	set	VERB
fcis-14812	45	25	like	like	ADP
fcis-14812	45	26	most	most	ADJ
fcis-14812	45	27	other	other	ADJ
fcis-14812	45	28	convolutional	convolutional	ADJ
fcis-14812	45	29	neural	neural	ADJ
fcis-14812	45	30	network	network	NOUN
fcis-14812	45	31	-	-	PUNCT
fcis-14812	45	32	based	base	VERB
fcis-14812	45	33	crowd	crowd	NOUN
fcis-14812	45	34	density	density	NOUN
fcis-14812	45	35	estimation	estimation	NOUN
fcis-14812	45	36	methods	method	NOUN
fcis-14812	45	37	,	,	PUNCT
fcis-14812	45	38	but	but	CCONJ
fcis-14812	45	39	only	only	ADV
fcis-14812	45	40	requires	require	VERB
fcis-14812	45	41	one	one	NUM
fcis-14812	45	42	pre	pre	ADJ
fcis-14812	45	43	-	-	ADJ
fcis-14812	45	44	training	training	ADJ
fcis-14812	45	45	session	session	NOUN
fcis-14812	45	46	to	to	PART
fcis-14812	45	47	count	count	VERB
fcis-14812	45	48	different	different	ADJ
fcis-14812	45	49	objects	object	NOUN
fcis-14812	45	50	.	.	PUNCT
fcis-14812	46	1	at	at	ADP
fcis-14812	46	2	the	the	DET
fcis-14812	46	3	end	end	NOUN
fcis-14812	46	4	of	of	ADP
fcis-14812	46	5	the	the	DET
fcis-14812	46	6	paper	paper	NOUN
fcis-14812	46	7	,	,	PUNCT
fcis-14812	46	8	the	the	DET
fcis-14812	46	9	authors	author	NOUN
fcis-14812	46	10	discuss	discuss	VERB
fcis-14812	46	11	the	the	DET
fcis-14812	46	12	feasibility	feasibility	NOUN
fcis-14812	46	13	of	of	ADP
fcis-14812	46	14	class	class	NOUN
fcis-14812	46	15	agnostic	agnostic	ADJ
fcis-14812	46	16	counting	counting	NOUN
fcis-14812	46	17	in	in	ADP
fcis-14812	46	18	different	different	ADJ
fcis-14812	46	19	domains	domain	NOUN
fcis-14812	46	20	such	such	ADJ
fcis-14812	46	21	as	as	ADP
fcis-14812	46	22	cells	cell	NOUN
fcis-14812	46	23	,	,	PUNCT
fcis-14812	46	24	vehicles	vehicle	NOUN
fcis-14812	46	25	,	,	PUNCT
fcis-14812	46	26	crowds	crowd	NOUN
fcis-14812	46	27	,	,	PUNCT
fcis-14812	46	28	etc	etc	X
fcis-14812	46	29	.	.	X
fcis-14812	46	30	and	and	CCONJ
fcis-14812	46	31	show	show	VERB
fcis-14812	46	32	the	the	DET
fcis-14812	46	33	excellent	excellent	ADJ
fcis-14812	46	34	performance	performance	NOUN
fcis-14812	46	35	of	of	ADP
fcis-14812	46	36	the	the	DET
fcis-14812	46	37	model	model	NOUN
fcis-14812	46	38	on	on	ADP
fcis-14812	46	39	various	various	ADJ
fcis-14812	46	40	datasets	dataset	NOUN
fcis-14812	46	41	.	.	PUNCT
fcis-14812	47	1	ranjan	ranjan	PROPN
fcis-14812	47	2	et	et	PROPN
fcis-14812	47	3	al	al	PROPN
fcis-14812	48	1	[	[	X
fcis-14812	48	2	14	14	NUM
fcis-14812	48	3	]	]	PUNCT
fcis-14812	48	4	proposed	propose	VERB
fcis-14812	48	5	a	a	DET
fcis-14812	48	6	brand	brand	NOUN
fcis-14812	48	7	-	-	PUNCT
fcis-14812	48	8	new	new	ADJ
fcis-14812	48	9	network	network	NOUN
fcis-14812	48	10	architecture	architecture	NOUN
fcis-14812	48	11	called	call	VERB
fcis-14812	48	12	few	few	ADJ
fcis-14812	48	13	-	-	PUNCT
fcis-14812	48	14	shot	shot	NOUN
fcis-14812	48	15	adaptation	adaptation	NOUN
fcis-14812	48	16	&	&	CCONJ
fcis-14812	48	17	matching	matching	NOUN
fcis-14812	48	18	network	network	NOUN
fcis-14812	48	19	(	(	PUNCT
fcis-14812	48	20	famnet	famnet	PROPN
fcis-14812	48	21	)	)	PUNCT
fcis-14812	48	22	to	to	PART
fcis-14812	48	23	solve	solve	VERB
fcis-14812	48	24	the	the	DET
fcis-14812	48	25	cac	cac	PROPN
fcis-14812	48	26	problem	problem	NOUN
fcis-14812	48	27	.	.	PUNCT
fcis-14812	49	1	meanwhile	meanwhile	ADV
fcis-14812	49	2	,	,	PUNCT
fcis-14812	49	3	the	the	DET
fcis-14812	49	4	authors	author	NOUN
fcis-14812	49	5	built	build	VERB
fcis-14812	49	6	an	an	DET
fcis-14812	49	7	fsc-147	fsc-147	NOUN
fcis-14812	49	8	dataset	dataset	NOUN
fcis-14812	49	9	.	.	PUNCT
fcis-14812	50	1	the	the	DET
fcis-14812	50	2	dataset	dataset	NOUN
fcis-14812	50	3	consists	consist	VERB
fcis-14812	50	4	of	of	ADP
fcis-14812	50	5	6135	6135	NUM
fcis-14812	50	6	images	image	NOUN
fcis-14812	50	7	covering	cover	VERB
fcis-14812	50	8	147	147	NUM
fcis-14812	50	9	different	different	ADJ
fcis-14812	50	10	object	object	NOUN
fcis-14812	50	11	categories	category	NOUN
fcis-14812	50	12	ranging	range	VERB
fcis-14812	50	13	from	from	ADP
fcis-14812	50	14	kitchen	kitchen	NOUN
fcis-14812	50	15	utensils	utensil	NOUN
fcis-14812	50	16	and	and	CCONJ
fcis-14812	50	17	office	office	NOUN
fcis-14812	50	18	stationery	stationery	NOUN
fcis-14812	50	19	to	to	ADP
fcis-14812	50	20	vehicles	vehicle	NOUN
fcis-14812	50	21	and	and	CCONJ
fcis-14812	50	22	animals.the	animals.the	DET
fcis-14812	50	23	fsc-147	fsc-147	NOUN
fcis-14812	50	24	dataset	dataset	NOUN
fcis-14812	50	25	has	have	VERB
fcis-14812	50	26	a	a	DET
fcis-14812	50	27	large	large	ADJ
fcis-14812	50	28	variation	variation	NOUN
fcis-14812	50	29	in	in	ADP
fcis-14812	50	30	the	the	DET
fcis-14812	50	31	number	number	NOUN
fcis-14812	50	32	of	of	ADP
fcis-14812	50	33	objects	object	NOUN
fcis-14812	50	34	and	and	CCONJ
fcis-14812	50	35	a	a	DET
fcis-14812	50	36	large	large	ADJ
fcis-14812	50	37	number	number	NOUN
fcis-14812	50	38	of	of	ADP
fcis-14812	50	39	image	image	NOUN
fcis-14812	50	40	categories	category	NOUN
fcis-14812	50	41	,	,	PUNCT
fcis-14812	50	42	which	which	PRON
fcis-14812	50	43	are	be	AUX
fcis-14812	50	44	constructed	construct	VERB
fcis-14812	50	45	specifically	specifically	ADV
fcis-14812	50	46	for	for	ADP
fcis-14812	50	47	the	the	DET
fcis-14812	50	48	cac	cac	PROPN
fcis-14812	50	49	problem	problem	NOUN
fcis-14812	50	50	.	.	PUNCT
fcis-14812	51	1	famnet	famnet	PROPN
fcis-14812	51	2	is	be	AUX
fcis-14812	51	3	an	an	DET
fcis-14812	51	4	end	end	NOUN
fcis-14812	51	5	-	-	PUNCT
fcis-14812	51	6	to	to	ADP
fcis-14812	51	7	-	-	PUNCT
fcis-14812	51	8	end	end	NOUN
fcis-14812	51	9	network	network	NOUN
fcis-14812	51	10	that	that	PRON
fcis-14812	51	11	consists	consist	VERB
fcis-14812	51	12	of	of	ADP
fcis-14812	51	13	a	a	DET
fcis-14812	51	14	feature	feature	NOUN
fcis-14812	51	15	extractor	extractor	NOUN
fcis-14812	51	16	and	and	CCONJ
fcis-14812	51	17	a	a	DET
fcis-14812	51	18	feature	feature	NOUN
fcis-14812	51	19	aggregator	aggregator	NOUN
fcis-14812	51	20	.	.	PUNCT
fcis-14812	52	1	the	the	DET
fcis-14812	52	2	feature	feature	NOUN
fcis-14812	52	3	extractor	extractor	NOUN
fcis-14812	52	4	uses	use	VERB
fcis-14812	52	5	resnet-50	resnet-50	PROPN
fcis-14812	52	6	as	as	ADP
fcis-14812	52	7	the	the	DET
fcis-14812	52	8	backbone	backbone	NOUN
fcis-14812	52	9	network	network	NOUN
fcis-14812	52	10	and	and	CCONJ
fcis-14812	52	11	outputs	output	VERB
fcis-14812	52	12	multi	multi	ADJ
fcis-14812	52	13	-	-	ADJ
fcis-14812	52	14	scale	scale	ADJ
fcis-14812	52	15	feature	feature	NOUN
fcis-14812	52	16	maps	map	NOUN
fcis-14812	52	17	at	at	ADP
fcis-14812	52	18	each	each	DET
fcis-14812	52	19	stage	stage	NOUN
fcis-14812	52	20	.	.	PUNCT
fcis-14812	53	1	the	the	DET
fcis-14812	53	2	feature	feature	NOUN
fcis-14812	53	3	aggregator	aggregator	NOUN
fcis-14812	53	4	uses	use	VERB
fcis-14812	53	5	a	a	DET
fcis-14812	53	6	selfattention	selfattention	NOUN
fcis-14812	53	7	mechanism	mechanism	NOUN
fcis-14812	53	8	to	to	PART
fcis-14812	53	9	fuse	fuse	VERB
fcis-14812	53	10	features	feature	NOUN
fcis-14812	53	11	from	from	ADP
fcis-14812	53	12	different	different	ADJ
fcis-14812	53	13	scales	scale	NOUN
fcis-14812	53	14	and	and	CCONJ
fcis-14812	53	15	channels	channel	NOUN
fcis-14812	53	16	and	and	CCONJ
fcis-14812	53	17	outputs	output	VERB
fcis-14812	53	18	a	a	DET
fcis-14812	53	19	global	global	ADJ
fcis-14812	53	20	feature	feature	NOUN
fcis-14812	53	21	vector	vector	NOUN
fcis-14812	53	22	.	.	PUNCT
fcis-14812	54	1	this	this	DET
fcis-14812	54	2	global	global	ADJ
fcis-14812	54	3	feature	feature	NOUN
fcis-14812	54	4	vector	vector	NOUN
fcis-14812	54	5	is	be	AUX
fcis-14812	54	6	fed	feed	VERB
fcis-14812	54	7	into	into	ADP
fcis-14812	54	8	a	a	DET
fcis-14812	54	9	fully	fully	ADV
fcis-14812	54	10	connected	connect	VERB
fcis-14812	54	11	layer	layer	NOUN
fcis-14812	54	12	to	to	PART
fcis-14812	54	13	predict	predict	VERB
fcis-14812	54	14	the	the	DET
fcis-14812	54	15	number	number	NOUN
fcis-14812	54	16	of	of	ADP
fcis-14812	54	17	targets	target	NOUN
fcis-14812	54	18	.	.	PUNCT
fcis-14812	55	1	the	the	DET
fcis-14812	55	2	authors	author	NOUN
fcis-14812	55	3	conducted	conduct	VERB
fcis-14812	55	4	experiments	experiment	NOUN
fcis-14812	55	5	on	on	ADP
fcis-14812	55	6	the	the	DET
fcis-14812	55	7	fsc-147	fsc-147	NOUN
fcis-14812	55	8	dataset	dataset	NOUN
fcis-14812	55	9	and	and	CCONJ
fcis-14812	55	10	compared	compare	VERB
fcis-14812	55	11	it	it	PRON
fcis-14812	55	12	with	with	ADP
fcis-14812	55	13	other	other	ADJ
fcis-14812	55	14	methods	method	NOUN
fcis-14812	55	15	,	,	PUNCT
fcis-14812	55	16	which	which	PRON
fcis-14812	55	17	showed	show	VERB
fcis-14812	55	18	that	that	SCONJ
fcis-14812	55	19	famnet	famnet	NOUN
fcis-14812	55	20	has	have	VERB
fcis-14812	55	21	better	well	ADJ
fcis-14812	55	22	generalization	generalization	NOUN
fcis-14812	55	23	ability	ability	NOUN
fcis-14812	55	24	and	and	CCONJ
fcis-14812	55	25	robustness	robustness	NOUN
fcis-14812	55	26	.	.	PUNCT
fcis-14812	56	1	shi	shi	PROPN
fcis-14812	56	2	et	et	PROPN
fcis-14812	56	3	al	al	PROPN
fcis-14812	57	1	[	[	X
fcis-14812	57	2	15	15	NUM
fcis-14812	57	3	]	]	X
fcis-14812	57	4	focused	focus	VERB
fcis-14812	57	5	on	on	ADP
fcis-14812	57	6	similarity	similarity	NOUN
fcis-14812	57	7	modeling	modeling	NOUN
fcis-14812	57	8	and	and	CCONJ
fcis-14812	57	9	constructed	construct	VERB
fcis-14812	57	10	a	a	DET
fcis-14812	57	11	generalized	generalized	ADJ
fcis-14812	57	12	cac	cac	PROPN
fcis-14812	57	13	similarity	similarity	NOUN
fcis-14812	57	14	-	-	PUNCT
fcis-14812	57	15	aware	aware	ADJ
fcis-14812	57	16	framework	framework	NOUN
fcis-14812	57	17	,	,	PUNCT
fcis-14812	57	18	bmnet+	bmnet+	PROPN
fcis-14812	57	19	,	,	PUNCT
fcis-14812	57	20	which	which	PRON
fcis-14812	57	21	jointly	jointly	ADV
fcis-14812	57	22	learns	learn	VERB
fcis-14812	57	23	feature	feature	NOUN
fcis-14812	57	24	representation	representation	NOUN
fcis-14812	57	25	and	and	CCONJ
fcis-14812	57	26	similarity	similarity	NOUN
fcis-14812	57	27	metrics	metric	NOUN
fcis-14812	57	28	in	in	ADP
fcis-14812	57	29	an	an	DET
fcis-14812	57	30	end	end	NOUN
fcis-14812	57	31	-	-	PUNCT
fcis-14812	57	32	to	to	ADP
fcis-14812	57	33	-	-	PUNCT
fcis-14812	57	34	end	end	NOUN
fcis-14812	57	35	manner	manner	NOUN
fcis-14812	57	36	,	,	PUNCT
fcis-14812	57	37	and	and	CCONJ
fcis-14812	57	38	uses	use	VERB
fcis-14812	57	39	a	a	DET
fcis-14812	57	40	composite	composite	ADJ
fcis-14812	57	41	loss	loss	NOUN
fcis-14812	57	42	function	function	NOUN
fcis-14812	57	43	to	to	PART
fcis-14812	57	44	monitor	monitor	VERB
fcis-14812	57	45	the	the	DET
fcis-14812	57	46	counts	count	NOUN
fcis-14812	57	47	and	and	CCONJ
fcis-14812	57	48	similarity	similarity	NOUN
fcis-14812	57	49	metrics	metric	NOUN
fcis-14812	57	50	simultaneously	simultaneously	ADV
fcis-14812	57	51	.	.	PUNCT
fcis-14812	58	1	metrics	metric	NOUN
fcis-14812	58	2	are	be	AUX
fcis-14812	58	3	simultaneously	simultaneously	ADV
fcis-14812	58	4	supervised	supervise	VERB
fcis-14812	58	5	with	with	ADP
fcis-14812	58	6	a	a	DET
fcis-14812	58	7	composite	composite	ADJ
fcis-14812	58	8	loss	loss	NOUN
fcis-14812	58	9	function	function	NOUN
fcis-14812	58	10	,	,	PUNCT
fcis-14812	58	11	allowing	allow	VERB
fcis-14812	58	12	it	it	PRON
fcis-14812	58	13	to	to	PART
fcis-14812	58	14	achieve	achieve	VERB
fcis-14812	58	15	the	the	DET
fcis-14812	58	16	best	good	ADJ
fcis-14812	58	17	current	current	ADJ
fcis-14812	58	18	performance	performance	NOUN
fcis-14812	58	19	in	in	ADP
fcis-14812	58	20	the	the	DET
fcis-14812	58	21	cac	cac	PROPN
fcis-14812	58	22	domain	domain	NOUN
fcis-14812	58	23	on	on	ADP
fcis-14812	58	24	the	the	DET
fcis-14812	58	25	fsc147	fsc147	NOUN
fcis-14812	58	26	dataset	dataset	NOUN
fcis-14812	58	27	.	.	PUNCT
fcis-14812	59	1	although	although	SCONJ
fcis-14812	59	2	this	this	DET
fcis-14812	59	3	work	work	NOUN
fcis-14812	59	4	proposes	propose	VERB
fcis-14812	59	5	a	a	DET
fcis-14812	59	6	dynamic	dynamic	ADJ
fcis-14812	59	7	similarity	similarity	NOUN
fcis-14812	59	8	metric	metric	ADJ
fcis-14812	59	9	for	for	ADP
fcis-14812	59	10	multi	multi	ADJ
fcis-14812	59	11	-	-	ADJ
fcis-14812	59	12	scale	scale	ADJ
fcis-14812	59	13	targets	target	NOUN
fcis-14812	59	14	which	which	PRON
fcis-14812	59	15	makes	make	VERB
fcis-14812	59	16	the	the	DET
fcis-14812	59	17	adaptability	adaptability	NOUN
fcis-14812	59	18	to	to	ADP
fcis-14812	59	19	multi	multi	ADJ
fcis-14812	59	20	-	-	ADJ
fcis-14812	59	21	scale	scale	ADJ
fcis-14812	59	22	targets	target	NOUN
fcis-14812	59	23	somewhat	somewhat	ADV
fcis-14812	59	24	improved	improve	VERB
fcis-14812	59	25	,	,	PUNCT
fcis-14812	59	26	there	there	PRON
fcis-14812	59	27	is	be	VERB
fcis-14812	59	28	still	still	ADV
fcis-14812	59	29	room	room	NOUN
fcis-14812	59	30	for	for	ADP
fcis-14812	59	31	improvement	improvement	NOUN
fcis-14812	59	32	.	.	PUNCT
fcis-14812	60	1	therefore	therefore	ADV
fcis-14812	60	2	,	,	PUNCT
fcis-14812	60	3	in	in	ADP
fcis-14812	60	4	this	this	DET
fcis-14812	60	5	paper	paper	NOUN
fcis-14812	60	6	,	,	PUNCT
fcis-14812	60	7	we	we	PRON
fcis-14812	60	8	propose	propose	VERB
fcis-14812	60	9	a	a	DET
fcis-14812	60	10	conditional	conditional	ADJ
fcis-14812	60	11	random	random	ADJ
fcis-14812	60	12	field	field	NOUN
fcis-14812	60	13	-	-	PUNCT
fcis-14812	60	14	based	base	VERB
fcis-14812	60	15	feature	feature	NOUN
fcis-14812	60	16	enhancement	enhancement	NOUN
fcis-14812	60	17	method	method	NOUN
fcis-14812	60	18	for	for	ADP
fcis-14812	60	19	improving	improve	VERB
fcis-14812	60	20	the	the	DET
fcis-14812	60	21	adaptability	adaptability	NOUN
fcis-14812	60	22	of	of	ADP
fcis-14812	60	23	similarity	similarity	NOUN
fcis-14812	60	24	metric	metric	ADJ
fcis-14812	60	25	networks	network	NOUN
fcis-14812	60	26	for	for	ADP
fcis-14812	60	27	multi	multi	ADJ
fcis-14812	60	28	-	-	ADJ
fcis-14812	60	29	scale	scale	ADJ
fcis-14812	60	30	targets	target	NOUN
fcis-14812	60	31	and	and	CCONJ
fcis-14812	60	32	improving	improve	VERB
fcis-14812	60	33	the	the	DET
fcis-14812	60	34	training	training	NOUN
fcis-14812	60	35	performance	performance	NOUN
fcis-14812	60	36	of	of	ADP
fcis-14812	60	37	the	the	DET
fcis-14812	60	38	networks	network	NOUN
fcis-14812	60	39	.	.	PUNCT
fcis-14812	61	1	2.3	2.3	NUM
fcis-14812	61	2	.	.	PUNCT
fcis-14812	62	1	conditional	conditional	ADJ
fcis-14812	62	2	random	random	ADJ
fcis-14812	62	3	field	field	NOUN
fcis-14812	62	4	related	relate	VERB
fcis-14812	62	5	work	work	NOUN
fcis-14812	62	6	in	in	ADP
fcis-14812	62	7	the	the	DET
fcis-14812	62	8	field	field	NOUN
fcis-14812	62	9	of	of	ADP
fcis-14812	62	10	computer	computer	NOUN
fcis-14812	62	11	vision	vision	NOUN
fcis-14812	62	12	,	,	PUNCT
fcis-14812	62	13	conditional	conditional	ADJ
fcis-14812	62	14	random	random	ADJ
fcis-14812	62	15	field	field	NOUN
fcis-14812	62	16	(	(	PUNCT
fcis-14812	62	17	crf	crf	NOUN
fcis-14812	62	18	)	)	PUNCT
fcis-14812	62	19	is	be	AUX
fcis-14812	62	20	a	a	DET
fcis-14812	62	21	commonly	commonly	ADV
fcis-14812	62	22	used	use	VERB
fcis-14812	62	23	probabilistic	probabilistic	ADJ
fcis-14812	62	24	graphical	graphical	ADJ
fcis-14812	62	25	model	model	NOUN
fcis-14812	62	26	that	that	PRON
fcis-14812	62	27	can	can	AUX
fcis-14812	62	28	be	be	AUX
fcis-14812	62	29	used	use	VERB
fcis-14812	62	30	to	to	PART
fcis-14812	62	31	improve	improve	VERB
fcis-14812	62	32	the	the	DET
fcis-14812	62	33	features	feature	NOUN
fcis-14812	62	34	and	and	CCONJ
fcis-14812	62	35	outputs	output	NOUN
fcis-14812	62	36	of	of	ADP
fcis-14812	62	37	convolutional	convolutional	ADJ
fcis-14812	62	38	neural	neural	ADJ
fcis-14812	62	39	networks	network	NOUN
fcis-14812	62	40	(	(	PUNCT
fcis-14812	62	41	cnns	cnns	PROPN
fcis-14812	62	42	)	)	PUNCT
fcis-14812	62	43	by	by	ADP
fcis-14812	62	44	using	use	VERB
fcis-14812	62	45	the	the	DET
fcis-14812	62	46	message	message	NOUN
fcis-14812	62	47	passing	pass	VERB
fcis-14812	62	48	mechanism	mechanism	NOUN
fcis-14812	62	49	.	.	PUNCT
fcis-14812	63	1	crfs	crfs	PROPN
fcis-14812	63	2	can	can	AUX
fcis-14812	63	3	model	model	VERB
fcis-14812	63	4	the	the	DET
fcis-14812	63	5	relationships	relationship	NOUN
fcis-14812	63	6	between	between	ADP
fcis-14812	63	7	pixels	pixel	NOUN
fcis-14812	63	8	or	or	CCONJ
fcis-14812	63	9	regions	region	NOUN
fcis-14812	63	10	to	to	PART
fcis-14812	63	11	improve	improve	VERB
fcis-14812	63	12	the	the	DET
fcis-14812	63	13	expressive	expressive	ADJ
fcis-14812	63	14	power	power	NOUN
fcis-14812	63	15	and	and	CCONJ
fcis-14812	63	16	prediction	prediction	NOUN
fcis-14812	63	17	accuracy	accuracy	NOUN
fcis-14812	63	18	of	of	ADP
fcis-14812	63	19	cnns	cnns	PROPN
fcis-14812	63	20	.	.	PUNCT
fcis-14812	64	1	for	for	ADP
fcis-14812	64	2	example	example	NOUN
fcis-14812	64	3	,	,	PUNCT
fcis-14812	64	4	zheng	zheng	PROPN
fcis-14812	64	5	et	et	PROPN
fcis-14812	64	6	al	al	PROPN
fcis-14812	64	7	[	[	X
fcis-14812	64	8	16	16	NUM
fcis-14812	64	9	]	]	PUNCT
fcis-14812	64	10	proposed	propose	VERB
fcis-14812	64	11	a	a	DET
fcis-14812	64	12	crf	crf	NOUN
fcis-14812	64	13	-	-	PUNCT
fcis-14812	64	14	based	base	VERB
fcis-14812	64	15	approach	approach	NOUN
fcis-14812	64	16	that	that	PRON
fcis-14812	64	17	can	can	AUX
fcis-14812	64	18	refine	refine	VERB
fcis-14812	64	19	the	the	DET
fcis-14812	64	20	semantic	semantic	ADJ
fcis-14812	64	21	segmentation	segmentation	NOUN
fcis-14812	64	22	map	map	NOUN
fcis-14812	64	23	of	of	ADP
fcis-14812	64	24	a	a	DET
fcis-14812	64	25	cnn	cnn	NOUN
fcis-14812	64	26	by	by	ADP
fcis-14812	64	27	modeling	model	VERB
fcis-14812	64	28	the	the	DET
fcis-14812	64	29	relationships	relationship	NOUN
fcis-14812	64	30	between	between	ADP
fcis-14812	64	31	pixels	pixel	NOUN
fcis-14812	64	32	for	for	ADP
fcis-14812	64	33	more	more	ADV
fcis-14812	64	34	accurate	accurate	ADJ
fcis-14812	64	35	scene	scene	NOUN
fcis-14812	64	36	understanding	understanding	NOUN
fcis-14812	64	37	.	.	PUNCT
fcis-14812	65	1	xu	xu	INTJ
fcis-14812	65	2	et	et	PROPN
fcis-14812	66	1	al	al	PROPN
fcis-14812	67	1	[	[	X
fcis-14812	67	2	17	17	NUM
fcis-14812	67	3	]	]	PUNCT
fcis-14812	67	4	improved	improve	VERB
fcis-14812	67	5	the	the	DET
fcis-14812	67	6	performance	performance	NOUN
fcis-14812	67	7	of	of	ADP
fcis-14812	67	8	human	human	ADJ
fcis-14812	67	9	pose	pose	NOUN
fcis-14812	67	10	estimation	estimation	NOUN
fcis-14812	67	11	by	by	ADP
fcis-14812	67	12	fusing	fuse	VERB
fcis-14812	67	13	multiple	multiple	ADJ
fcis-14812	67	14	features	feature	NOUN
fcis-14812	67	15	with	with	ADP
fcis-14812	67	16	attentional	attentional	ADJ
fcis-14812	67	17	gating	gate	VERB
fcis-14812	67	18	crfs	crf	NOUN
fcis-14812	67	19	to	to	PART
fcis-14812	67	20	produce	produce	VERB
fcis-14812	67	21	a	a	DET
fcis-14812	67	22	richer	rich	ADJ
fcis-14812	67	23	representation	representation	NOUN
fcis-14812	67	24	of	of	ADP
fcis-14812	67	25	contour	contour	NOUN
fcis-14812	67	26	prediction	prediction	NOUN
fcis-14812	67	27	.	.	PUNCT
fcis-14812	68	1	wang	wang	PROPN
fcis-14812	68	2	et	et	PROPN
fcis-14812	68	3	al	al	PROPN
fcis-14812	69	1	[	[	X
fcis-14812	69	2	18	18	NUM
fcis-14812	69	3	]	]	PUNCT
fcis-14812	69	4	introduced	introduce	VERB
fcis-14812	69	5	an	an	DET
fcis-14812	69	6	inter	inter	ADJ
fcis-14812	69	7	-	-	ADJ
fcis-14812	69	8	view	view	ADJ
fcis-14812	69	9	message	message	NOUN
fcis-14812	69	10	passing	pass	VERB
fcis-14812	69	11	module	module	NOUN
fcis-14812	69	12	based	base	VERB
fcis-14812	69	13	on	on	ADP
fcis-14812	69	14	crfs	crfs	PROPN
fcis-14812	69	15	to	to	PART
fcis-14812	69	16	enhance	enhance	VERB
fcis-14812	69	17	imagespecific	imagespecific	ADJ
fcis-14812	69	18	features	feature	NOUN
fcis-14812	69	19	for	for	ADP
fcis-14812	69	20	action	action	NOUN
fcis-14812	69	21	recognition	recognition	NOUN
fcis-14812	69	22	.	.	PUNCT
fcis-14812	70	1	it	it	PRON
fcis-14812	70	2	can	can	AUX
fcis-14812	70	3	utilize	utilize	VERB
fcis-14812	70	4	the	the	DET
fcis-14812	70	5	information	information	NOUN
fcis-14812	70	6	from	from	ADP
fcis-14812	70	7	different	different	ADJ
fcis-14812	70	8	viewpoints	viewpoint	NOUN
fcis-14812	70	9	to	to	PART
fcis-14812	70	10	enhance	enhance	VERB
fcis-14812	70	11	the	the	DET
fcis-14812	70	12	features	feature	NOUN
fcis-14812	70	13	of	of	ADP
fcis-14812	70	14	each	each	DET
fcis-14812	70	15	viewpoint	viewpoint	NOUN
fcis-14812	70	16	,	,	PUNCT
fcis-14812	70	17	which	which	PRON
fcis-14812	70	18	improves	improve	VERB
fcis-14812	70	19	the	the	DET
fcis-14812	70	20	accuracy	accuracy	NOUN
fcis-14812	70	21	of	of	ADP
fcis-14812	70	22	action	action	NOUN
fcis-14812	70	23	recognition	recognition	NOUN
fcis-14812	70	24	.	.	PUNCT
fcis-14812	71	1	in	in	ADP
fcis-14812	71	2	crowd	crowd	NOUN
fcis-14812	71	3	counting	counting	NOUN
fcis-14812	71	4	,	,	PUNCT
fcis-14812	71	5	deep	deep	ADJ
fcis-14812	71	6	structured	structured	ADJ
fcis-14812	71	7	scale	scale	NOUN
fcis-14812	71	8	integration	integration	NOUN
fcis-14812	71	9	network	network	NOUN
fcis-14812	71	10	(	(	PUNCT
fcis-14812	71	11	dssinet	dssinet	ADV
fcis-14812	71	12	)	)	PUNCT
fcis-14812	71	13	[	[	X
fcis-14812	71	14	19	19	NUM
fcis-14812	71	15	]	]	PUNCT
fcis-14812	71	16	for	for	ADP
fcis-14812	71	17	the	the	DET
fcis-14812	71	18	first	first	ADJ
fcis-14812	71	19	time	time	NOUN
fcis-14812	71	20	utilizes	utilize	VERB
fcis-14812	71	21	crfs	crf	NOUN
fcis-14812	71	22	to	to	PART
fcis-14812	71	23	mutually	mutually	ADV
fcis-14812	71	24	refine	refine	VERB
fcis-14812	71	25	multiple	multiple	ADJ
fcis-14812	71	26	features	feature	NOUN
fcis-14812	71	27	at	at	ADP
fcis-14812	71	28	different	different	ADJ
fcis-14812	71	29	scales	scale	NOUN
fcis-14812	71	30	and	and	CCONJ
fcis-14812	71	31	proves	prove	VERB
fcis-14812	71	32	its	its	PRON
fcis-14812	71	33	effectiveness	effectiveness	NOUN
fcis-14812	71	34	for	for	ADP
fcis-14812	71	35	this	this	DET
fcis-14812	71	36	task	task	NOUN
fcis-14812	71	37	.	.	PUNCT
fcis-14812	72	1	it	it	PRON
fcis-14812	72	2	improves	improve	VERB
fcis-14812	72	3	the	the	DET
fcis-14812	72	4	performance	performance	NOUN
fcis-14812	72	5	of	of	ADP
fcis-14812	72	6	crowd	crowd	NOUN
fcis-14812	72	7	counting	counting	NOUN
fcis-14812	72	8	by	by	ADP
fcis-14812	72	9	utilizing	utilize	VERB
fcis-14812	72	10	the	the	DET
fcis-14812	72	11	message	message	NOUN
fcis-14812	72	12	passing	pass	VERB
fcis-14812	72	13	mechanism	mechanism	NOUN
fcis-14812	72	14	of	of	ADP
fcis-14812	72	15	crfs	crf	NOUN
fcis-14812	72	16	to	to	PART
fcis-14812	72	17	balance	balance	VERB
fcis-14812	72	18	features	feature	NOUN
fcis-14812	72	19	at	at	ADP
fcis-14812	72	20	different	different	ADJ
fcis-14812	72	21	scales	scale	NOUN
fcis-14812	72	22	to	to	PART
fcis-14812	72	23	generate	generate	VERB
fcis-14812	72	24	finer	fine	ADJ
fcis-14812	72	25	density	density	NOUN
fcis-14812	72	26	maps	map	NOUN
fcis-14812	72	27	.	.	PUNCT
fcis-14812	73	1	72	72	NUM
fcis-14812	73	2	2.4	2.4	NUM
fcis-14812	73	3	.	.	PUNCT
fcis-14812	74	1	metric	metric	ADJ
fcis-14812	74	2	learning	learning	NOUN
fcis-14812	74	3	related	relate	VERB
fcis-14812	74	4	work	work	NOUN
fcis-14812	74	5	the	the	DET
fcis-14812	74	6	aim	aim	NOUN
fcis-14812	74	7	of	of	ADP
fcis-14812	74	8	metric	metric	ADJ
fcis-14812	74	9	learning	learning	NOUN
fcis-14812	74	10	is	be	AUX
fcis-14812	74	11	to	to	PART
fcis-14812	74	12	embed	embed	VERB
fcis-14812	74	13	data	datum	NOUN
fcis-14812	74	14	into	into	ADP
fcis-14812	74	15	a	a	DET
fcis-14812	74	16	space	space	NOUN
fcis-14812	74	17	where	where	SCONJ
fcis-14812	74	18	similar	similar	ADJ
fcis-14812	74	19	samples	sample	NOUN
fcis-14812	74	20	are	be	AUX
fcis-14812	74	21	brought	bring	VERB
fcis-14812	74	22	closer	close	ADV
fcis-14812	74	23	together	together	ADV
fcis-14812	74	24	and	and	CCONJ
fcis-14812	74	25	different	different	ADJ
fcis-14812	74	26	samples	sample	NOUN
fcis-14812	74	27	are	be	AUX
fcis-14812	74	28	pushed	push	VERB
fcis-14812	74	29	apart	apart	ADV
fcis-14812	74	30	[	[	X
fcis-14812	74	31	22	22	NUM
fcis-14812	74	32	]	]	PUNCT
fcis-14812	74	33	.	.	PUNCT
fcis-14812	75	1	similarity	similarity	NOUN
fcis-14812	75	2	is	be	AUX
fcis-14812	75	3	measured	measure	VERB
fcis-14812	75	4	in	in	ADP
fcis-14812	75	5	a	a	DET
fcis-14812	75	6	fixed	fix	VERB
fcis-14812	75	7	[	[	X
fcis-14812	75	8	23]24	23]24	NUM
fcis-14812	75	9	]	]	PUNCT
fcis-14812	75	10	or	or	CCONJ
fcis-14812	75	11	learned	learn	VERB
fcis-14812	75	12	[	[	PUNCT
fcis-14812	75	13	25][26	25][26	ADP
fcis-14812	75	14	]	]	PUNCT
fcis-14812	75	15	manner	manner	NOUN
fcis-14812	75	16	.	.	PUNCT
fcis-14812	76	1	a	a	DET
fcis-14812	76	2	common	common	ADJ
fcis-14812	76	3	approach	approach	NOUN
fcis-14812	76	4	is	be	AUX
fcis-14812	76	5	to	to	PART
fcis-14812	76	6	restrict	restrict	VERB
fcis-14812	76	7	the	the	DET
fcis-14812	76	8	similarity	similarity	NOUN
fcis-14812	76	9	between	between	ADP
fcis-14812	76	10	features	feature	NOUN
fcis-14812	76	11	in	in	ADP
fcis-14812	76	12	a	a	DET
fcis-14812	76	13	pair	pair	NOUN
fcis-14812	76	14	[	[	X
fcis-14812	76	15	27	27	NUM
fcis-14812	76	16	]	]	PUNCT
fcis-14812	76	17	or	or	CCONJ
fcis-14812	76	18	trio	trio	NOUN
fcis-14812	76	19	[	[	X
fcis-14812	76	20	28	28	NUM
fcis-14812	76	21	]	]	PUNCT
fcis-14812	76	22	.	.	PUNCT
fcis-14812	77	1	this	this	DET
fcis-14812	77	2	approach	approach	NOUN
fcis-14812	77	3	allows	allow	VERB
fcis-14812	77	4	similar	similar	ADJ
fcis-14812	77	5	samples	sample	NOUN
fcis-14812	77	6	to	to	PART
fcis-14812	77	7	be	be	AUX
fcis-14812	77	8	closer	close	ADJ
fcis-14812	77	9	together	together	ADV
fcis-14812	77	10	in	in	ADP
fcis-14812	77	11	space	space	NOUN
fcis-14812	77	12	and	and	CCONJ
fcis-14812	77	13	different	different	ADJ
fcis-14812	77	14	samples	sample	NOUN
fcis-14812	77	15	to	to	PART
fcis-14812	77	16	be	be	AUX
fcis-14812	77	17	more	more	ADV
fcis-14812	77	18	spread	spread	VERB
fcis-14812	77	19	out	out	ADP
fcis-14812	77	20	in	in	ADP
fcis-14812	77	21	space	space	NOUN
fcis-14812	77	22	.	.	PUNCT
fcis-14812	78	1	another	another	DET
fcis-14812	78	2	approach	approach	NOUN
fcis-14812	78	3	adds	add	VERB
fcis-14812	78	4	constraints	constraint	NOUN
fcis-14812	78	5	based	base	VERB
fcis-14812	78	6	on	on	ADP
fcis-14812	78	7	the	the	DET
fcis-14812	78	8	signal	signal	NOUN
fcis-14812	78	9	-	-	PUNCT
fcis-14812	78	10	to	to	ADP
fcis-14812	78	11	-	-	PUNCT
fcis-14812	78	12	noise	noise	NOUN
fcis-14812	78	13	ratio	ratio	NOUN
fcis-14812	78	14	[	[	X
fcis-14812	78	15	29][30][31	29][30][31	X
fcis-14812	78	16	]	]	X
fcis-14812	78	17	,	,	PUNCT
fcis-14812	78	18	where	where	SCONJ
fcis-14812	78	19	the	the	DET
fcis-14812	78	20	similarity	similarity	NOUN
fcis-14812	78	21	between	between	ADP
fcis-14812	78	22	pairs	pair	NOUN
fcis-14812	78	23	of	of	ADP
fcis-14812	78	24	positive	positive	ADJ
fcis-14812	78	25	samples	sample	NOUN
fcis-14812	78	26	is	be	AUX
fcis-14812	78	27	considered	consider	VERB
fcis-14812	78	28	signal	signal	NOUN
fcis-14812	78	29	and	and	CCONJ
fcis-14812	78	30	the	the	DET
fcis-14812	78	31	similarity	similarity	NOUN
fcis-14812	78	32	between	between	ADP
fcis-14812	78	33	pairs	pair	NOUN
fcis-14812	78	34	of	of	ADP
fcis-14812	78	35	negative	negative	ADJ
fcis-14812	78	36	samples	sample	NOUN
fcis-14812	78	37	is	be	AUX
fcis-14812	78	38	considered	consider	VERB
fcis-14812	78	39	noise	noise	NOUN
fcis-14812	78	40	.	.	PUNCT
fcis-14812	79	1	this	this	DET
fcis-14812	79	2	approach	approach	NOUN
fcis-14812	79	3	allows	allow	VERB
fcis-14812	79	4	the	the	DET
fcis-14812	79	5	signal	signal	NOUN
fcis-14812	79	6	to	to	PART
fcis-14812	79	7	be	be	AUX
fcis-14812	79	8	stronger	strong	ADJ
fcis-14812	79	9	and	and	CCONJ
fcis-14812	79	10	the	the	DET
fcis-14812	79	11	noise	noise	NOUN
fcis-14812	79	12	to	to	PART
fcis-14812	79	13	be	be	AUX
fcis-14812	79	14	weaker	weak	ADJ
fcis-14812	79	15	.	.	PUNCT
fcis-14812	80	1	bmnet+	bmnet+	PROPN
fcis-14812	80	2	reapplies	reapplie	VERB
fcis-14812	80	3	this	this	DET
fcis-14812	80	4	idea	idea	NOUN
fcis-14812	80	5	to	to	ADP
fcis-14812	80	6	cac	cac	PROPN
fcis-14812	80	7	by	by	ADP
fcis-14812	80	8	pulling	pull	VERB
fcis-14812	80	9	features	feature	NOUN
fcis-14812	80	10	closer	close	ADV
fcis-14812	80	11	between	between	ADP
fcis-14812	80	12	examples	example	NOUN
fcis-14812	80	13	and	and	CCONJ
fcis-14812	80	14	target	target	NOUN
fcis-14812	80	15	instances	instance	NOUN
fcis-14812	80	16	while	while	SCONJ
fcis-14812	80	17	pushing	push	VERB
fcis-14812	80	18	features	feature	NOUN
fcis-14812	80	19	away	away	ADV
fcis-14812	80	20	between	between	ADP
fcis-14812	80	21	examples	example	NOUN
fcis-14812	80	22	and	and	CCONJ
fcis-14812	80	23	background	background	NOUN
fcis-14812	80	24	patches	patch	NOUN
fcis-14812	80	25	.	.	PUNCT
fcis-14812	81	1	this	this	DET
fcis-14812	81	2	approach	approach	NOUN
fcis-14812	81	3	allows	allow	VERB
fcis-14812	81	4	for	for	ADP
fcis-14812	81	5	more	more	ADV
fcis-14812	81	6	similar	similar	ADJ
fcis-14812	81	7	features	feature	NOUN
fcis-14812	81	8	between	between	ADP
fcis-14812	81	9	examples	example	NOUN
fcis-14812	81	10	and	and	CCONJ
fcis-14812	81	11	target	target	NOUN
fcis-14812	81	12	instances	instance	NOUN
fcis-14812	81	13	,	,	PUNCT
fcis-14812	81	14	and	and	CCONJ
fcis-14812	81	15	more	more	ADV
fcis-14812	81	16	different	different	ADJ
fcis-14812	81	17	features	feature	NOUN
fcis-14812	81	18	between	between	ADP
fcis-14812	81	19	examples	example	NOUN
fcis-14812	81	20	and	and	CCONJ
fcis-14812	81	21	background	background	NOUN
fcis-14812	81	22	patches	patch	NOUN
fcis-14812	81	23	.	.	PUNCT
fcis-14812	82	1	3	3	X
fcis-14812	82	2	.	.	X
fcis-14812	82	3	method	method	PROPN
fcis-14812	82	4	3.1	3.1	NUM
fcis-14812	82	5	.	.	PUNCT
fcis-14812	83	1	network	network	NOUN
fcis-14812	83	2	structure	structure	NOUN
fcis-14812	83	3	aiming	aim	VERB
fcis-14812	83	4	at	at	ADP
fcis-14812	83	5	the	the	DET
fcis-14812	83	6	insufficient	insufficient	ADJ
fcis-14812	83	7	adaptability	adaptability	NOUN
fcis-14812	83	8	of	of	ADP
fcis-14812	83	9	bmnet+	bmnet+	PROPN
fcis-14812	83	10	to	to	ADP
fcis-14812	83	11	changes	change	NOUN
fcis-14812	83	12	with	with	ADP
fcis-14812	83	13	multi	multi	ADJ
fcis-14812	83	14	-	-	ADJ
fcis-14812	83	15	scale	scale	ADJ
fcis-14812	83	16	targets	target	NOUN
fcis-14812	83	17	,	,	PUNCT
fcis-14812	83	18	this	this	DET
fcis-14812	83	19	paper	paper	NOUN
fcis-14812	83	20	proposes	propose	VERB
fcis-14812	83	21	an	an	DET
fcis-14812	83	22	improved	improved	ADJ
fcis-14812	83	23	network	network	NOUN
fcis-14812	83	24	model	model	NOUN
fcis-14812	83	25	dfenet	dfenet	NOUN
fcis-14812	83	26	based	base	VERB
fcis-14812	83	27	on	on	ADP
fcis-14812	83	28	bmnet+	bmnet+	PROPN
fcis-14812	83	29	,	,	PUNCT
fcis-14812	83	30	as	as	SCONJ
fcis-14812	83	31	shown	show	VERB
fcis-14812	83	32	in	in	ADP
fcis-14812	83	33	fig	fig	NOUN
fcis-14812	83	34	.	.	PUNCT
fcis-14812	84	1	1	1	NUM
fcis-14812	84	2	,	,	PUNCT
fcis-14812	84	3	which	which	PRON
fcis-14812	84	4	consists	consist	VERB
fcis-14812	84	5	of	of	ADP
fcis-14812	84	6	six	six	NUM
fcis-14812	84	7	parts	part	NOUN
fcis-14812	84	8	,	,	PUNCT
fcis-14812	84	9	namely	namely	ADV
fcis-14812	84	10	,	,	PUNCT
fcis-14812	84	11	the	the	DET
fcis-14812	84	12	backbone	backbone	NOUN
fcis-14812	84	13	network	network	NOUN
fcis-14812	84	14	,	,	PUNCT
fcis-14812	84	15	self	self	NOUN
fcis-14812	84	16	-	-	PUNCT
fcis-14812	84	17	similarity	similarity	NOUN
fcis-14812	84	18	module	module	NOUN
fcis-14812	84	19	,	,	PUNCT
fcis-14812	84	20	sfem	sfem	ADJ
fcis-14812	84	21	,	,	PUNCT
fcis-14812	84	22	asff	asff	NOUN
fcis-14812	84	23	,	,	PUNCT
fcis-14812	84	24	and	and	CCONJ
fcis-14812	84	25	dynamic	dynamic	ADJ
fcis-14812	84	26	similarity	similarity	NOUN
fcis-14812	84	27	metric	metric	ADJ
fcis-14812	84	28	and	and	CCONJ
fcis-14812	84	29	counting	counting	NOUN
fcis-14812	84	30	module	module	NOUN
fcis-14812	84	31	.	.	PUNCT
fcis-14812	85	1	our	our	PRON
fcis-14812	85	2	whole	whole	ADJ
fcis-14812	85	3	network	network	NOUN
fcis-14812	85	4	structure	structure	NOUN
fcis-14812	85	5	is	be	AUX
fcis-14812	85	6	shown	show	VERB
fcis-14812	85	7	in	in	ADP
fcis-14812	85	8	fig	fig	NOUN
fcis-14812	85	9	.	.	PUNCT
fcis-14812	86	1	1	1	X
fcis-14812	86	2	.	.	X
fcis-14812	86	3	figure	figure	NOUN
fcis-14812	86	4	1	1	NUM
fcis-14812	86	5	.	.	PUNCT
fcis-14812	87	1	the	the	DET
fcis-14812	87	2	above	above	ADJ
fcis-14812	87	3	figure	figure	NOUN
fcis-14812	87	4	shows	show	VERB
fcis-14812	87	5	the	the	DET
fcis-14812	87	6	overall	overall	ADJ
fcis-14812	87	7	network	network	NOUN
fcis-14812	87	8	structure	structure	NOUN
fcis-14812	87	9	of	of	ADP
fcis-14812	87	10	this	this	DET
fcis-14812	87	11	paper	paper	NOUN
fcis-14812	87	12	.	.	PUNCT
fcis-14812	88	1	the	the	DET
fcis-14812	88	2	red	red	ADJ
fcis-14812	88	3	line	line	NOUN
fcis-14812	88	4	represents	represent	VERB
fcis-14812	88	5	the	the	DET
fcis-14812	88	6	processing	processing	NOUN
fcis-14812	88	7	route	route	NOUN
fcis-14812	88	8	of	of	ADP
fcis-14812	88	9	the	the	DET
fcis-14812	88	10	sample	sample	NOUN
fcis-14812	88	11	image	image	NOUN
fcis-14812	88	12	,	,	PUNCT
fcis-14812	88	13	and	and	CCONJ
fcis-14812	88	14	the	the	DET
fcis-14812	88	15	green	green	ADJ
fcis-14812	88	16	line	line	NOUN
fcis-14812	88	17	represents	represent	VERB
fcis-14812	88	18	the	the	DET
fcis-14812	88	19	processing	processing	NOUN
fcis-14812	88	20	route	route	NOUN
fcis-14812	88	21	of	of	ADP
fcis-14812	88	22	the	the	DET
fcis-14812	88	23	query	query	NOUN
fcis-14812	88	24	image	image	NOUN
fcis-14812	88	25	.	.	PUNCT
fcis-14812	89	1	backbone	backbone	NOUN
fcis-14812	89	2	.	.	PUNCT
fcis-14812	90	1	the	the	DET
fcis-14812	90	2	backbone	backbone	NOUN
fcis-14812	90	3	network	network	NOUN
fcis-14812	90	4	is	be	AUX
fcis-14812	90	5	the	the	DET
fcis-14812	90	6	net	net	ADJ
fcis-14812	90	7	structure	structure	NOUN
fcis-14812	90	8	that	that	PRON
fcis-14812	90	9	the	the	DET
fcis-14812	90	10	model	model	NOUN
fcis-14812	90	11	uses	use	VERB
fcis-14812	90	12	at	at	ADP
fcis-14812	90	13	the	the	DET
fcis-14812	90	14	very	very	ADJ
fcis-14812	90	15	beginning	beginning	NOUN
fcis-14812	90	16	to	to	PART
fcis-14812	90	17	extract	extract	VERB
fcis-14812	90	18	the	the	DET
fcis-14812	90	19	target	target	NOUN
fcis-14812	90	20	features	feature	NOUN
fcis-14812	90	21	,	,	PUNCT
fcis-14812	90	22	and	and	CCONJ
fcis-14812	90	23	the	the	DET
fcis-14812	90	24	design	design	NOUN
fcis-14812	90	25	of	of	ADP
fcis-14812	90	26	the	the	DET
fcis-14812	90	27	backbone	backbone	NOUN
fcis-14812	90	28	network	network	NOUN
fcis-14812	90	29	affects	affect	VERB
fcis-14812	90	30	the	the	DET
fcis-14812	90	31	results	result	NOUN
fcis-14812	90	32	of	of	ADP
fcis-14812	90	33	target	target	NOUN
fcis-14812	90	34	counting	counting	NOUN
fcis-14812	90	35	to	to	ADP
fcis-14812	90	36	some	some	DET
fcis-14812	90	37	extent	extent	NOUN
fcis-14812	90	38	.	.	PUNCT
fcis-14812	91	1	the	the	DET
fcis-14812	91	2	backbone	backbone	NOUN
fcis-14812	91	3	network	network	NOUN
fcis-14812	91	4	used	use	VERB
fcis-14812	91	5	in	in	ADP
fcis-14812	91	6	this	this	DET
fcis-14812	91	7	work	work	NOUN
fcis-14812	91	8	is	be	AUX
fcis-14812	91	9	the	the	DET
fcis-14812	91	10	first	first	ADJ
fcis-14812	91	11	4	4	NUM
fcis-14812	91	12	blocks	block	NOUN
fcis-14812	91	13	of	of	ADP
fcis-14812	91	14	resnet-50	resnet-50	PROPN
fcis-14812	91	15	.	.	PUNCT
fcis-14812	91	16	self	self	NOUN
fcis-14812	91	17	-	-	PUNCT
fcis-14812	91	18	similarity	similarity	NOUN
fcis-14812	91	19	module	module	NOUN
fcis-14812	91	20	.	.	PUNCT
fcis-14812	92	1	the	the	DET
fcis-14812	92	2	self	self	NOUN
fcis-14812	92	3	-	-	PUNCT
fcis-14812	92	4	similarity	similarity	NOUN
fcis-14812	92	5	module	module	NOUN
fcis-14812	92	6	is	be	AUX
fcis-14812	92	7	one	one	NUM
fcis-14812	92	8	of	of	ADP
fcis-14812	92	9	the	the	DET
fcis-14812	92	10	cores	core	NOUN
fcis-14812	92	11	of	of	ADP
fcis-14812	92	12	the	the	DET
fcis-14812	92	13	bmnet+	bmnet+	PROPN
fcis-14812	92	14	framework	framework	NOUN
fcis-14812	92	15	.	.	PUNCT
fcis-14812	93	1	since	since	SCONJ
fcis-14812	93	2	the	the	DET
fcis-14812	93	3	occurrence	occurrence	NOUN
fcis-14812	93	4	of	of	ADP
fcis-14812	93	5	the	the	DET
fcis-14812	93	6	same	same	ADJ
fcis-14812	93	7	type	type	NOUN
fcis-14812	93	8	of	of	ADP
fcis-14812	93	9	targets	target	NOUN
fcis-14812	93	10	is	be	AUX
fcis-14812	93	11	often	often	ADV
fcis-14812	93	12	accompanied	accompany	VERB
fcis-14812	93	13	by	by	ADP
fcis-14812	93	14	different	different	ADJ
fcis-14812	93	15	scales	scale	NOUN
fcis-14812	93	16	and	and	CCONJ
fcis-14812	93	17	poses	pose	NOUN
fcis-14812	93	18	,	,	PUNCT
fcis-14812	93	19	this	this	DET
fcis-14812	93	20	variation	variation	NOUN
fcis-14812	93	21	within	within	ADP
fcis-14812	93	22	the	the	DET
fcis-14812	93	23	target	target	NOUN
fcis-14812	93	24	group	group	NOUN
fcis-14812	93	25	creates	create	VERB
fcis-14812	93	26	a	a	DET
fcis-14812	93	27	great	great	ADJ
fcis-14812	93	28	challenge	challenge	NOUN
fcis-14812	93	29	for	for	ADP
fcis-14812	93	30	the	the	DET
fcis-14812	93	31	similarity	similarity	NOUN
fcis-14812	93	32	measure	measure	NOUN
fcis-14812	93	33	.	.	PUNCT
fcis-14812	94	1	bmnet+	bmnet+	PROPN
fcis-14812	94	2	proposes	propose	VERB
fcis-14812	94	3	a	a	DET
fcis-14812	94	4	self	self	NOUN
fcis-14812	94	5	-	-	PUNCT
fcis-14812	94	6	similarity	similarity	NOUN
fcis-14812	94	7	module	module	NOUN
fcis-14812	94	8	for	for	ADP
fcis-14812	94	9	enhancing	enhance	VERB
fcis-14812	94	10	the	the	DET
fcis-14812	94	11	common	common	ADJ
fcis-14812	94	12	features	feature	NOUN
fcis-14812	94	13	among	among	ADP
fcis-14812	94	14	each	each	DET
fcis-14812	94	15	target	target	NOUN
fcis-14812	94	16	.	.	PUNCT
fcis-14812	95	1	specifically	specifically	ADV
fcis-14812	95	2	,	,	PUNCT
fcis-14812	95	3	the	the	DET
fcis-14812	95	4	module	module	NOUN
fcis-14812	95	5	first	first	ADV
fcis-14812	95	6	collects	collect	VERB
fcis-14812	95	7	sample	sample	NOUN
fcis-14812	95	8	features	feature	VERB
fcis-14812	95	9			NOUN
fcis-14812	95	10	zf	zf	CCONJ
fcis-14812	95	11	with	with	SCONJ
fcis-14812	95	12	each	each	DET
fcis-14812	95	13	feature	feature	NOUN
fcis-14812	95	14	vector	vector	NOUN
fcis-14812	95	15	of	of	ADP
fcis-14812	95	16	the	the	DET
fcis-14812	95	17	query	query	NOUN
fcis-14812	95	18	feature	feature	NOUN
fcis-14812	95	19	image	image	NOUN
fcis-14812	95	20			NOUN
fcis-14812	95	21	xfij	xfij	NOUN
fcis-14812	95	22	,	,	PUNCT
fcis-14812	95	23	and	and	CCONJ
fcis-14812	95	24	put	put	VERB
fcis-14812	95	25	them	they	PRON
fcis-14812	95	26	into	into	ADP
fcis-14812	95	27	a	a	DET
fcis-14812	95	28	feature	feature	NOUN
fcis-14812	95	29	set	set	NOUN
fcis-14812	95	30	.	.	PUNCT
fcis-14812	96	1	each	each	DET
fcis-14812	96	2	vector	vector	NOUN
fcis-14812	96	3	in	in	ADP
fcis-14812	96	4	the	the	DET
fcis-14812	96	5	feature	feature	NOUN
fcis-14812	96	6	set	set	NOUN
fcis-14812	96	7	is	be	AUX
fcis-14812	96	8	then	then	ADV
fcis-14812	96	9	updated	update	VERB
fcis-14812	96	10	through	through	ADP
fcis-14812	96	11	a	a	DET
fcis-14812	96	12	self	self	NOUN
fcis-14812	96	13	-	-	PUNCT
fcis-14812	96	14	attentive	attentive	ADJ
fcis-14812	96	15	mechanism	mechanism	NOUN
fcis-14812	96	16	,	,	PUNCT
fcis-14812	96	17	and	and	CCONJ
fcis-14812	96	18	the	the	DET
fcis-14812	96	19	updated	update	VERB
fcis-14812	96	20	features	feature	NOUN
fcis-14812	96	21	are	be	AUX
fcis-14812	96	22	accompanied	accompany	VERB
fcis-14812	96	23	by	by	ADP
fcis-14812	96	24	a	a	DET
fcis-14812	96	25	learnable	learnable	ADJ
fcis-14812	96	26	γ	γ	X
fcis-14812	96	27	parameter	parameter	NOUN
fcis-14812	96	28	that	that	PRON
fcis-14812	96	29	is	be	AUX
fcis-14812	96	30	added	add	VERB
fcis-14812	96	31	to	to	ADP
fcis-14812	96	32	the	the	DET
fcis-14812	96	33	original	original	ADJ
fcis-14812	96	34	features	feature	NOUN
fcis-14812	96	35	.	.	PUNCT
fcis-14812	97	1	finally	finally	ADV
fcis-14812	97	2	,	,	PUNCT
fcis-14812	97	3	the	the	DET
fcis-14812	97	4	resulting	result	VERB
fcis-14812	97	5	feature	feature	NOUN
fcis-14812	97	6	set	set	NOUN
fcis-14812	97	7	is	be	AUX
fcis-14812	97	8	split	split	VERB
fcis-14812	97	9	to	to	PART
fcis-14812	97	10	obtain	obtain	VERB
fcis-14812	97	11	the	the	DET
fcis-14812	97	12	final	final	ADJ
fcis-14812	97	13			NOUN
fcis-14812	97	14	zf	zf	PUNCT
fcis-14812	97	15	and	and	CCONJ
fcis-14812	97	16			PROPN
fcis-14812	97	17	xf	xf	PROPN
fcis-14812	97	18	sfem	sfem	PROPN
fcis-14812	97	19	.	.	PUNCT
fcis-14812	98	1	sfem	sfem	PROPN
fcis-14812	98	2	(	(	PUNCT
fcis-14812	98	3	structured	structure	VERB
fcis-14812	98	4	feature	feature	NOUN
fcis-14812	98	5	enhancement	enhancement	NOUN
fcis-14812	98	6	module	module	NOUN
fcis-14812	98	7	)	)	PUNCT
fcis-14812	98	8	was	be	AUX
fcis-14812	98	9	first	first	ADV
fcis-14812	98	10	proposed	propose	VERB
fcis-14812	98	11	by	by	ADP
fcis-14812	98	12	liu	liu	PROPN
fcis-14812	98	13	et	et	PROPN
fcis-14812	98	14	al	al	PROPN
fcis-14812	99	1	[	[	X
fcis-14812	99	2	19	19	NUM
fcis-14812	99	3	]	]	PUNCT
fcis-14812	99	4	in	in	ADP
fcis-14812	99	5	the	the	DET
fcis-14812	99	6	network	network	NOUN
fcis-14812	99	7	model	model	NOUN
fcis-14812	99	8	dssinet	dssinet	PROPN
fcis-14812	99	9	.	.	PUNCT
fcis-14812	100	1	inspired	inspire	VERB
fcis-14812	100	2	by	by	ADP
fcis-14812	100	3	intensive	intensive	ADJ
fcis-14812	100	4	forecasting	forecasting	NOUN
fcis-14812	100	5	work	work	NOUN
fcis-14812	100	6	,	,	PUNCT
fcis-14812	100	7	the	the	DET
fcis-14812	100	8	sfem	sfem	NOUN
fcis-14812	100	9	module	module	NOUN
fcis-14812	100	10	mutually	mutually	ADV
fcis-14812	100	11	refines	refine	VERB
fcis-14812	100	12	features	feature	NOUN
fcis-14812	100	13	at	at	ADP
fcis-14812	100	14	different	different	ADJ
fcis-14812	100	15	scales	scale	NOUN
fcis-14812	100	16	by	by	ADP
fcis-14812	100	17	fully	fully	ADV
fcis-14812	100	18	exploring	explore	VERB
fcis-14812	100	19	their	their	PRON
fcis-14812	100	20	complementarity	complementarity	NOUN
fcis-14812	100	21	with	with	ADP
fcis-14812	100	22	conditional	conditional	ADJ
fcis-14812	100	23	random	random	ADJ
fcis-14812	100	24	fields	field	NOUN
fcis-14812	100	25	(	(	PUNCT
fcis-14812	100	26	crfs	crf	NOUN
fcis-14812	100	27	)	)	PUNCT
fcis-14812	100	28	models	model	NOUN
fcis-14812	100	29	.	.	PUNCT
fcis-14812	101	1	in	in	ADP
fcis-14812	101	2	this	this	DET
fcis-14812	101	3	module	module	NOUN
fcis-14812	101	4	,	,	PUNCT
fcis-14812	101	5	each	each	DET
fcis-14812	101	6	scale	scale	NOUN
fcis-14812	101	7	-	-	PUNCT
fcis-14812	101	8	specific	specific	ADJ
fcis-14812	101	9	feature	feature	NOUN
fcis-14812	101	10	passes	pass	VERB
fcis-14812	101	11	its	its	PRON
fcis-14812	101	12	own	own	ADJ
fcis-14812	101	13	information	information	NOUN
fcis-14812	101	14	to	to	ADP
fcis-14812	101	15	features	feature	NOUN
fcis-14812	101	16	from	from	ADP
fcis-14812	101	17	other	other	ADJ
fcis-14812	101	18	scales	scale	NOUN
fcis-14812	101	19	.	.	PUNCT
fcis-14812	102	1	specifically	specifically	ADV
fcis-14812	102	2	,	,	PUNCT
fcis-14812	102	3	it	it	PRON
fcis-14812	102	4	is	be	AUX
fcis-14812	102	5	assumed	assume	VERB
fcis-14812	102	6	that	that	SCONJ
fcis-14812	102	7	the	the	DET
fcis-14812	102	8	ensemble	ensemble	NOUN
fcis-14812	102	9	of	of	ADP
fcis-14812	102	10	features	feature	NOUN
fcis-14812	102	11	from	from	ADP
fcis-14812	102	12	each	each	DET
fcis-14812	102	13	different	different	ADJ
fcis-14812	102	14	sub	sub	NOUN
fcis-14812	102	15	-	-	NOUN
fcis-14812	102	16	network	network	NOUN
fcis-14812	102	17	is	be	AUX
fcis-14812	102	18	denoted	denote	VERB
fcis-14812	102	19	as	as	ADP
fcis-14812	102	20			PROPN
fcis-14812	102	21			PROPN
fcis-14812	102	22	,	,	PUNCT
fcis-14812	102	23	,	,	PUNCT
fcis-14812	102	24	...	...	PUNCT
fcis-14812	102	25	,	,	PUNCT
fcis-14812	102	26	,	,	PUNCT
fcis-14812	102	27	21	21	NUM
fcis-14812	102	28	nffff	nffff	NOUN
fcis-14812	102	29			PROPN
fcis-14812	102	30	,	,	PUNCT
fcis-14812	102	31	our	our	PRON
fcis-14812	102	32	goal	goal	NOUN
fcis-14812	102	33	is	be	AUX
fcis-14812	102	34	to	to	PART
fcis-14812	102	35	estimate	estimate	VERB
fcis-14812	102	36	a	a	DET
fcis-14812	102	37	refined	refined	ADJ
fcis-14812	102	38	set	set	NOUN
fcis-14812	102	39	of	of	ADP
fcis-14812	102	40	features	feature	NOUN
fcis-14812	102	41			PROPN
fcis-14812	103	1			PROPN
fcis-14812	103	2			PROPN
fcis-14812	103	3			PROPN
fcis-14812	103	4			NUM
fcis-14812	103	5			ADV
fcis-14812	103	6	,	,	PUNCT
fcis-14812	103	7	,	,	PUNCT
fcis-14812	103	8	...	...	PUNCT
fcis-14812	103	9	,	,	PUNCT
fcis-14812	103	10	,	,	PUNCT
fcis-14812	103	11	^	^	SYM
fcis-14812	103	12	2	2	NUM
fcis-14812	103	13	^	^	SYM
fcis-14812	103	14	1	1	NUM
fcis-14812	103	15	^^	^^	NUM
fcis-14812	103	16	nffff	nffff	NOUN
fcis-14812	103	17	,	,	PUNCT
fcis-14812	103	18	where	where	SCONJ
fcis-14812	103	19	^	^	PUNCT
fcis-14812	103	20	if	if	SCONJ
fcis-14812	103	21	is	be	AUX
fcis-14812	103	22	the	the	DET
fcis-14812	103	23	corresponding	correspond	VERB
fcis-14812	103	24	improvement	improvement	NOUN
fcis-14812	103	25	feature	feature	NOUN
fcis-14812	103	26	of	of	ADP
fcis-14812	103	27	if	if	SCONJ
fcis-14812	103	28	.	.	PUNCT
fcis-14812	104	1	we	we	PRON
fcis-14812	104	2	formulate	formulate	VERB
fcis-14812	104	3	this	this	PRON
fcis-14812	104	4	in	in	ADP
fcis-14812	104	5	terms	term	NOUN
fcis-14812	104	6	of	of	ADP
fcis-14812	104	7	the	the	DET
fcis-14812	104	8	crfs	crfs	PROPN
fcis-14812	104	9	model	model	NOUN
fcis-14812	104	10	.	.	PUNCT
fcis-14812	105	1	specifically	specifically	ADV
fcis-14812	105	2	,	,	PUNCT
fcis-14812	105	3	define	define	VERB
fcis-14812	105	4	the	the	DET
fcis-14812	105	5	conditional	conditional	ADJ
fcis-14812	105	6	distributions	distribution	NOUN
fcis-14812	105	7	of	of	ADP
fcis-14812	105	8	the	the	DET
fcis-14812	105	9	original	original	ADJ
fcis-14812	105	10	feature	feature	NOUN
fcis-14812	105	11	f	f	NOUN
fcis-14812	105	12	and	and	CCONJ
fcis-14812	105	13	the	the	DET
fcis-14812	105	14	improved	improved	ADJ
fcis-14812	105	15	feature	feature	NOUN
fcis-14812	105	16	^	^	PUNCT
fcis-14812	105	17	f	f	PROPN
fcis-14812	106	1	as	as	ADP
fcis-14812	106	2	:	:	PUNCT
fcis-14812	106	3			NOUN
fcis-14812	106	4			PUNCT
fcis-14812	107	1			PUNCT
fcis-14812	107	2			INTJ
fcis-14812	108	1			PROPN
fcis-14812	108	2			PROPN
fcis-14812	108	3			NUM
fcis-14812	108	4			ADP
fcis-14812	109	1			PROPN
fcis-14812	109	2			NOUN
fcis-14812	109	3			PRON
fcis-14812	109	4			PROPN
fcis-14812	109	5			PROPN
fcis-14812	109	6			NOUN
fcis-14812	109	7			NOUN
fcis-14812	109	8			NOUN
fcis-14812	109	9			PRON
fcis-14812	109	10			PROPN
fcis-14812	109	11			PROPN
fcis-14812	109	12			NOUN
fcis-14812	109	13			PROPN
fcis-14812	109	14	,	,	PUNCT
fcis-14812	109	15	,	,	PUNCT
fcis-14812	109	16	exp	exp	NOUN
fcis-14812	109	17	1	1	NUM
fcis-14812	109	18	,	,	PUNCT
fcis-14812	109	19	|	|	ADV
fcis-14812	109	20	^^	^^	SYM
fcis-14812	109	21	ffe	ffe	PROPN
fcis-14812	109	22	fz	fz	PROPN
fcis-14812	109	23	ffp	ffp	PROPN
fcis-14812	109	24	(	(	PUNCT
fcis-14812	109	25	1	1	NUM
fcis-14812	109	26	)	)	PUNCT
fcis-14812	109	27	where	where	SCONJ
fcis-14812	109	28			NOUN
fcis-14812	109	29			PROPN
fcis-14812	109	30	^^	^^	SYM
fcis-14812	109	31	f	f	PROPN
fcis-14812	109	32	,	,	PUNCT
fcis-14812	109	33	,	,	PUNCT
fcis-14812	109	34	expf	expf	PROPN
fcis-14812	109	35	^	^	PUNCT
fcis-14812	109	36	fdffez	fdffez	VERB
fcis-14812	110	1			PROPN
fcis-14812	111	1			PROPN
fcis-14812	112	1			PROPN
fcis-14812	113	1			PROPN
fcis-14812	114	1			NUM
fcis-14812	114	2			ADP
fcis-14812	115	1			PROPN
fcis-14812	115	2			NOUN
fcis-14812	115	3			PRON
fcis-14812	115	4			PROPN
fcis-14812	115	5			PROPN
fcis-14812	115	6			NOUN
fcis-14812	115	7			PRON
fcis-14812	115	8	is	be	AUX
fcis-14812	115	9	the	the	DET
fcis-14812	115	10	partition	partition	NOUN
fcis-14812	115	11	function	function	NOUN
fcis-14812	115	12	for	for	ADP
fcis-14812	115	13	normalization	normalization	NOUN
fcis-14812	115	14	,	,	PUNCT
fcis-14812	115	15	and	and	CCONJ
fcis-14812	115	16			NOUN
fcis-14812	115	17	is	be	AUX
fcis-14812	115	18	the	the	DET
fcis-14812	115	19	set	set	NOUN
fcis-14812	115	20	of	of	ADP
fcis-14812	115	21	parameters	parameter	NOUN
fcis-14812	115	22	the	the	DET
fcis-14812	115	23	energy	energy	NOUN
fcis-14812	115	24	functions	function	NOUN
fcis-14812	115	25			PROPN
fcis-14812	115	26			NOUN
fcis-14812	115	27			PRON
fcis-14812	115	28			PROPN
fcis-14812	115	29			PROPN
fcis-14812	115	30			NOUN
fcis-14812	115	31	,,e	,,e	NOUN
fcis-14812	115	32	^	^	PUNCT
fcis-14812	115	33	ff	ff	NOUN
fcis-14812	115	34	in	in	ADP
fcis-14812	115	35	crfs	crfs	PROPN
fcis-14812	115	36	is	be	AUX
fcis-14812	115	37	defined	define	VERB
fcis-14812	115	38	as	as	ADP
fcis-14812	115	39	:	:	PUNCT
fcis-14812	115	40			PROPN
fcis-14812	115	41			NUM
fcis-14812	115	42			PROPN
fcis-14812	115	43			PROPN
fcis-14812	115	44			PROPN
fcis-14812	115	45			VERB
fcis-14812	115	46			NUM
fcis-14812	115	47			PRON
fcis-14812	115	48			PROPN
fcis-14812	115	49			PROPN
fcis-14812	115	50			PROPN
fcis-14812	115	51			NOUN
fcis-14812	116	1			PRON
fcis-14812	116	2			PROPN
fcis-14812	116	3			PROPN
fcis-14812	116	4			NOUN
fcis-14812	116	5			PROPN
fcis-14812	116	6	ji	ji	PROPN
fcis-14812	116	7	ji	ji	PROPN
fcis-14812	116	8	ii	ii	PROPN
fcis-14812	116	9	ffffff	ffffff	PROPN
fcis-14812	116	10	^^	^^	PUNCT
fcis-14812	116	11	,	,	PUNCT
fcis-14812	116	12	^	^	PUNCT
fcis-14812	116	13	i	i	PROPN
fcis-14812	116	14	^	^	PUNCT
fcis-14812	116	15	,	,	PUNCT
fcis-14812	116	16	,	,	PUNCT
fcis-14812	116	17	,	,	PUNCT
fcis-14812	116	18	,	,	PUNCT
fcis-14812	116	19	e	e	X
fcis-14812	116	20			X
fcis-14812	116	21	(	(	PUNCT
fcis-14812	116	22	2	2	NUM
fcis-14812	116	23	)	)	PUNCT
fcis-14812	116	24	where	where	SCONJ
fcis-14812	116	25	the	the	DET
fcis-14812	116	26	unitary	unitary	ADJ
fcis-14812	116	27	potential	potential	ADJ
fcis-14812	116	28			PROPN
fcis-14812	116	29			NOUN
fcis-14812	116	30			PROPN
fcis-14812	116	31			PROPN
fcis-14812	116	32			PROPN
fcis-14812	116	33			NOUN
fcis-14812	116	34	ii	ii	NOUN
fcis-14812	116	35	ff	ff	INTJ
fcis-14812	116	36	,	,	PUNCT
fcis-14812	116	37	^	^	PUNCT
fcis-14812	116	38			NOUN
fcis-14812	116	39	representing	represent	VERB
fcis-14812	116	40	the	the	DET
fcis-14812	116	41	similarity	similarity	NOUN
fcis-14812	116	42	between	between	ADP
fcis-14812	116	43	the	the	DET
fcis-14812	116	44	original	original	ADJ
fcis-14812	116	45	and	and	CCONJ
fcis-14812	116	46	improved	improved	ADJ
fcis-14812	116	47	features	feature	NOUN
fcis-14812	116	48	is	be	AUX
fcis-14812	116	49	defined	define	VERB
fcis-14812	116	50	as	as	ADP
fcis-14812	116	51	:	:	PUNCT
fcis-14812	116	52	2	2	NUM
fcis-14812	116	53	^^	^^	NUM
fcis-14812	116	54	||||	||||	X
fcis-14812	116	55	2	2	NUM
fcis-14812	116	56	1	1	NUM
fcis-14812	116	57	,	,	PUNCT
fcis-14812	116	58	iiii	iiii	PROPN
fcis-14812	116	59	ffff	ffff	PROPN
fcis-14812	116	60			NUM
fcis-14812	116	61			NOUN
fcis-14812	116	62			PROPN
fcis-14812	116	63			PROPN
fcis-14812	116	64			PROPN
fcis-14812	116	65			PROPN
fcis-14812	116	66	(	(	PUNCT
fcis-14812	116	67	3	3	X
fcis-14812	116	68	)	)	PUNCT
fcis-14812	116	69	we	we	PRON
fcis-14812	116	70	model	model	VERB
fcis-14812	116	71	the	the	DET
fcis-14812	116	72	correlation	correlation	NOUN
fcis-14812	116	73	between	between	ADP
fcis-14812	116	74	two	two	NUM
fcis-14812	116	75	improved	improved	ADJ
fcis-14812	116	76	features	feature	NOUN
fcis-14812	116	77	with	with	ADP
fcis-14812	116	78	a	a	DET
fcis-14812	116	79	bilinear	bilinear	ADJ
fcis-14812	116	80	potential	potential	ADJ
fcis-14812	116	81	function	function	NOUN
fcis-14812	116	82	,	,	PUNCT
fcis-14812	116	83	and	and	CCONJ
fcis-14812	116	84	thus	thus	ADV
fcis-14812	116	85	define	define	VERB
fcis-14812	116	86	the	the	DET
fcis-14812	116	87	pairwise	pairwise	NOUN
fcis-14812	116	88	potentials	potential	VERB
fcis-14812	116	89	as	as	ADP
fcis-14812	116	90	:	:	PUNCT
fcis-14812	116	91	j	j	PROPN
fcis-14812	117	1	i	i	PRON
fcis-14812	117	2	j	j	PROPN
fcis-14812	117	3	t	t	PROPN
fcis-14812	117	4	iji	iji	VERB
fcis-14812	117	5	ffff	ffff	PROPN
fcis-14812	117	6	^^^^	^^^^	PROPN
fcis-14812	117	7	,	,	PUNCT
fcis-14812	118	1			PUNCT
fcis-14812	118	2			PROPN
fcis-14812	118	3			NOUN
fcis-14812	118	4			PRON
fcis-14812	118	5			PROPN
fcis-14812	118	6			PROPN
fcis-14812	118	7			NOUN
fcis-14812	118	8			VERB
fcis-14812	118	9			PRON
fcis-14812	118	10			PROPN
fcis-14812	118	11			PROPN
fcis-14812	118	12			NOUN
fcis-14812	118	13	(	(	PUNCT
fcis-14812	118	14	4	4	NUM
fcis-14812	118	15	)	)	PUNCT
fcis-14812	118	16	where	where	SCONJ
fcis-14812	118	17	i	i	PRON
fcis-14812	118	18	j	j	PROPN
fcis-14812	118	19	is	be	AUX
fcis-14812	118	20	the	the	DET
fcis-14812	118	21	learnable	learnable	ADJ
fcis-14812	118	22	parameter	parameter	NOUN
fcis-14812	118	23	used	use	VERB
fcis-14812	118	24	to	to	PART
fcis-14812	118	25	calculate	calculate	VERB
fcis-14812	118	26	the	the	DET
fcis-14812	118	27	correlation	correlation	NOUN
fcis-14812	118	28	between	between	ADP
fcis-14812	118	29	i	i	PRON
fcis-14812	118	30	^	^	PUNCT
fcis-14812	118	31	f	f	PROPN
fcis-14812	118	32	and	and	CCONJ
fcis-14812	118	33	j	j	PROPN
fcis-14812	118	34	^	^	PUNCT
fcis-14812	119	1	f	f	PROPN
fcis-14812	119	2	.	.	PUNCT
fcis-14812	120	1	this	this	PRON
fcis-14812	120	2	is	be	AUX
fcis-14812	120	3	a	a	DET
fcis-14812	120	4	typical	typical	ADJ
fcis-14812	120	5	crf	crf	NOUN
fcis-14812	120	6	formula	formula	NOUN
fcis-14812	120	7	that	that	PRON
fcis-14812	120	8	we	we	PRON
fcis-14812	120	9	solve	solve	VERB
fcis-14812	120	10	using	use	VERB
fcis-14812	120	11	mean	mean	ADJ
fcis-14812	120	12	field	field	NOUN
fcis-14812	120	13	extrapolation	extrapolation	NOUN
fcis-14812	120	14	.	.	PUNCT
fcis-14812	121	1	feature	feature	NOUN
fcis-14812	121	2	^	^	PUNCT
fcis-14812	121	3	if	if	SCONJ
fcis-14812	121	4	is	be	AUX
fcis-14812	121	5	calculated	calculate	VERB
fcis-14812	121	6	by	by	ADP
fcis-14812	121	7	the	the	DET
fcis-14812	121	8	formula	formula	NOUN
fcis-14812	121	9	:	:	PUNCT
fcis-14812	122	1	j	j	PROPN
fcis-14812	123	1	i	i	PRON
fcis-14812	123	2	j	j	PROPN
fcis-14812	124	1	ij	ij	NOUN
fcis-14812	124	2	ii	ii	PROPN
fcis-14812	124	3	fff	fff	PROPN
fcis-14812	124	4	^^	^^	PROPN
fcis-14812	124	5			PROPN
fcis-14812	124	6			NOUN
fcis-14812	124	7			X
fcis-14812	124	8	(	(	PUNCT
fcis-14812	124	9	5	5	NUM
fcis-14812	124	10	)	)	PUNCT
fcis-14812	124	11	where	where	SCONJ
fcis-14812	124	12	the	the	DET
fcis-14812	124	13	unary	unary	ADJ
fcis-14812	124	14	term	term	NOUN
fcis-14812	124	15	is	be	AUX
fcis-14812	124	16	the	the	DET
fcis-14812	124	17	feature	feature	NOUN
fcis-14812	124	18	if	if	SCONJ
fcis-14812	124	19	itself	itself	PRON
fcis-14812	124	20	,	,	PUNCT
fcis-14812	124	21	and	and	CCONJ
fcis-14812	124	22	the	the	DET
fcis-14812	124	23	second	second	ADJ
fcis-14812	124	24	term	term	NOUN
fcis-14812	124	25	denotes	denote	VERB
fcis-14812	124	26	the	the	DET
fcis-14812	124	27	information	information	NOUN
fcis-14812	124	28	received	receive	VERB
fcis-14812	124	29	from	from	ADP
fcis-14812	124	30	other	other	ADJ
fcis-14812	124	31	features	feature	NOUN
fcis-14812	124	32	at	at	ADP
fcis-14812	124	33	different	different	ADJ
fcis-14812	124	34	scales	scale	NOUN
fcis-14812	124	35	.	.	PUNCT
fcis-14812	125	1	the	the	DET
fcis-14812	125	2	parameter	parameter	NOUN
fcis-14812	125	3	i	i	PRON
fcis-14812	125	4	j	j	PROPN
fcis-14812	125	5	determines	determine	VERB
fcis-14812	125	6	the	the	DET
fcis-14812	125	7	information	information	NOUN
fcis-14812	125	8	content	content	NOUN
fcis-14812	125	9	passed	pass	VERB
fcis-14812	125	10	from	from	ADP
fcis-14812	125	11	jf	jf	PROPN
fcis-14812	125	12	to	to	ADP
fcis-14812	125	13	if	if	SCONJ
fcis-14812	125	14	.since	.since	PRON
fcis-14812	125	15	i	i	NOUN
fcis-14812	125	16	^	^	PUNCT
fcis-14812	125	17	f	f	PROPN
fcis-14812	125	18	and	and	CCONJ
fcis-14812	125	19	j	j	PROPN
fcis-14812	125	20	^	^	PUNCT
fcis-14812	125	21	f	f	PROPN
fcis-14812	125	22	are	be	AUX
fcis-14812	125	23	73	73	NUM
fcis-14812	125	24	interdependent	interdependent	NOUN
fcis-14812	125	25	in	in	ADP
fcis-14812	125	26	j	j	PROPN
fcis-14812	126	1	i	i	PRON
fcis-14812	126	2	j	j	PROPN
fcis-14812	127	1	ij	ij	NOUN
fcis-14812	127	2	ii	ii	PROPN
fcis-14812	127	3	fff	fff	PROPN
fcis-14812	127	4	^^	^^	PROPN
fcis-14812	127	5			PROPN
fcis-14812	127	6			NOUN
fcis-14812	127	7			X
fcis-14812	127	8	,	,	PUNCT
fcis-14812	127	9	we	we	PRON
fcis-14812	127	10	iteratively	iteratively	ADV
fcis-14812	127	11	obtain	obtain	VERB
fcis-14812	127	12	each	each	DET
fcis-14812	127	13	improvement	improvement	NOUN
fcis-14812	127	14	feature	feature	NOUN
fcis-14812	127	15	using	use	VERB
fcis-14812	127	16	the	the	DET
fcis-14812	127	17	following	follow	VERB
fcis-14812	127	18	formula	formula	NOUN
fcis-14812	127	19	:	:	PUNCT
fcis-14812	127	20	n	n	NUM
fcis-14812	127	21	ii	ii	NOUN
fcis-14812	127	22	t	t	PROPN
fcis-14812	127	23	j	j	PROPN
fcis-14812	128	1	i	i	PRON
fcis-14812	128	2	j	j	INTJ
fcis-14812	129	1	ij	ij	INTJ
fcis-14812	129	2	i	i	PRON
fcis-14812	129	3	t	t	PROPN
fcis-14812	129	4	iii	iii	NUM
fcis-14812	129	5	hfnthfhfh	hfnthfhfh	PROPN
fcis-14812	129	6			PROPN
fcis-14812	129	7			PROPN
fcis-14812	129	8			NOUN
fcis-14812	129	9	^	^	PUNCT
fcis-14812	129	10	10	10	NUM
fcis-14812	129	11	,	,	PUNCT
fcis-14812	129	12	...	...	PUNCT
fcis-14812	129	13	1	1	NUM
fcis-14812	129	14	,	,	PUNCT
fcis-14812	129	15	,	,	PUNCT
fcis-14812	129	16			X
fcis-14812	129	17	(	(	PUNCT
fcis-14812	129	18	6	6	NUM
fcis-14812	129	19	)	)	PUNCT
fcis-14812	129	20	where	where	SCONJ
fcis-14812	129	21	n	n	PRON
fcis-14812	129	22	is	be	AUX
fcis-14812	129	23	the	the	DET
fcis-14812	129	24	total	total	ADJ
fcis-14812	129	25	number	number	NOUN
fcis-14812	129	26	of	of	ADP
fcis-14812	129	27	iterations	iteration	NOUN
fcis-14812	129	28	and	and	CCONJ
fcis-14812	129	29	t	t	X
fcis-14812	129	30	ih	ih	NOUN
fcis-14812	129	31	is	be	AUX
fcis-14812	129	32	the	the	DET
fcis-14812	129	33	intermediate	intermediate	ADJ
fcis-14812	129	34	feature	feature	NOUN
fcis-14812	129	35	of	of	ADP
fcis-14812	129	36	the	the	DET
fcis-14812	129	37	tht	tht	PROPN
fcis-14812	129	38	iteration	iteration	NOUN
fcis-14812	129	39	.	.	PUNCT
fcis-14812	130	1	the	the	DET
fcis-14812	130	2	formula	formula	NOUN
fcis-14812	130	3	:	:	PUNCT
fcis-14812	130	4	n	n	NUM
fcis-14812	130	5	ii	ii	NOUN
fcis-14812	130	6	t	t	PROPN
fcis-14812	131	1	j	j	PROPN
fcis-14812	132	1	i	i	PRON
fcis-14812	132	2	j	j	INTJ
fcis-14812	133	1	ij	ij	INTJ
fcis-14812	133	2	i	i	PRON
fcis-14812	133	3	t	t	PROPN
fcis-14812	133	4	iii	iii	NUM
fcis-14812	133	5	hfnthfhfh	hfnthfhfh	PROPN
fcis-14812	133	6			PROPN
fcis-14812	133	7			PROPN
fcis-14812	133	8			NOUN
fcis-14812	133	9	^	^	PUNCT
fcis-14812	133	10	10	10	NUM
fcis-14812	133	11	,	,	PUNCT
fcis-14812	133	12	...	...	PUNCT
fcis-14812	133	13	1	1	NUM
fcis-14812	133	14	,	,	PUNCT
fcis-14812	133	15	,	,	PUNCT
fcis-14812	133	16			X
fcis-14812	133	17	(	(	PUNCT
fcis-14812	133	18	7	7	NUM
fcis-14812	133	19	)	)	PUNCT
fcis-14812	133	20	can	can	AUX
fcis-14812	133	21	be	be	AUX
fcis-14812	133	22	easily	easily	ADV
fcis-14812	133	23	implemented	implement	VERB
fcis-14812	133	24	in	in	ADP
fcis-14812	133	25	the	the	DET
fcis-14812	133	26	sfem	sfem	NOUN
fcis-14812	133	27	module	module	NOUN
fcis-14812	133	28	.	.	PUNCT
fcis-14812	134	1	specifically	specifically	ADV
fcis-14812	134	2	,	,	PUNCT
fcis-14812	134	3	the	the	DET
fcis-14812	134	4	sfem	sfem	NOUN
fcis-14812	134	5	module	module	NOUN
fcis-14812	134	6	uses	use	VERB
fcis-14812	134	7	a	a	DET
fcis-14812	134	8	1x1	1x1	NUM
fcis-14812	134	9	convolutional	convolutional	ADJ
fcis-14812	134	10	layer	layer	NOUN
fcis-14812	134	11	to	to	PART
fcis-14812	134	12	pass	pass	VERB
fcis-14812	134	13	complementary	complementary	ADJ
fcis-14812	134	14	information	information	NOUN
fcis-14812	134	15	from	from	ADP
fcis-14812	134	16	jf	jf	PROPN
fcis-14812	134	17	to	to	ADP
fcis-14812	134	18	if	if	SCONJ
fcis-14812	134	19	,	,	PUNCT
fcis-14812	134	20	i	i	PRON
fcis-14812	134	21	j	j	PROPN
fcis-14812	134	22	is	be	AUX
fcis-14812	134	23	the	the	DET
fcis-14812	134	24	learning	learn	VERB
fcis-14812	134	25	parameter	parameter	NOUN
fcis-14812	134	26	of	of	ADP
fcis-14812	134	27	the	the	DET
fcis-14812	134	28	convolutional	convolutional	ADJ
fcis-14812	134	29	layer	layer	NOUN
fcis-14812	134	30	and	and	CCONJ
fcis-14812	134	31	is	be	AUX
fcis-14812	134	32	shared	share	VERB
fcis-14812	134	33	for	for	ADP
fcis-14812	134	34	all	all	DET
fcis-14812	134	35	iterations	iteration	NOUN
fcis-14812	134	36	.	.	PUNCT
fcis-14812	135	1	asff	asff	PROPN
fcis-14812	135	2	.	.	PUNCT
fcis-14812	136	1	asff	asff	PROPN
fcis-14812	137	1	[	[	X
fcis-14812	137	2	20	20	NUM
fcis-14812	137	3	]	]	PUNCT
fcis-14812	137	4	is	be	AUX
fcis-14812	137	5	a	a	DET
fcis-14812	137	6	module	module	NOUN
fcis-14812	137	7	for	for	ADP
fcis-14812	137	8	fusing	fuse	VERB
fcis-14812	137	9	feature	feature	NOUN
fcis-14812	137	10	maps	map	NOUN
fcis-14812	137	11	that	that	PRON
fcis-14812	137	12	have	have	AUX
fcis-14812	137	13	been	be	AUX
fcis-14812	137	14	enhanced	enhance	VERB
fcis-14812	137	15	by	by	ADP
fcis-14812	137	16	the	the	DET
fcis-14812	137	17	sfem	sfem	NOUN
fcis-14812	137	18	module	module	NOUN
fcis-14812	137	19	.	.	PUNCT
fcis-14812	138	1	unlike	unlike	ADP
fcis-14812	138	2	previous	previous	ADJ
fcis-14812	138	3	methods	method	NOUN
fcis-14812	138	4	of	of	ADP
fcis-14812	138	5	integrating	integrate	VERB
fcis-14812	138	6	multilevel	multilevel	NOUN
fcis-14812	138	7	features	feature	NOUN
fcis-14812	138	8	using	use	VERB
fcis-14812	138	9	element	element	ADJ
fcis-14812	138	10	summation	summation	NOUN
fcis-14812	138	11	or	or	CCONJ
fcis-14812	138	12	concatenation	concatenation	NOUN
fcis-14812	138	13	,	,	PUNCT
fcis-14812	138	14	the	the	DET
fcis-14812	138	15	key	key	ADJ
fcis-14812	138	16	idea	idea	NOUN
fcis-14812	138	17	of	of	ADP
fcis-14812	138	18	asff	asff	NOUN
fcis-14812	138	19	(	(	PUNCT
fcis-14812	138	20	adaptively	adaptively	ADV
fcis-14812	138	21	spatial	spatial	ADJ
fcis-14812	138	22	feature	feature	NOUN
fcis-14812	138	23	fusion	fusion	NOUN
fcis-14812	138	24	)	)	PUNCT
fcis-14812	138	25	is	be	AUX
fcis-14812	138	26	to	to	PART
fcis-14812	138	27	adaptively	adaptively	ADV
fcis-14812	138	28	learn	learn	VERB
fcis-14812	138	29	the	the	DET
fcis-14812	138	30	spatial	spatial	ADJ
fcis-14812	138	31	weights	weight	NOUN
fcis-14812	138	32	of	of	ADP
fcis-14812	138	33	the	the	DET
fcis-14812	138	34	fused	fuse	VERB
fcis-14812	138	35	feature	feature	NOUN
fcis-14812	138	36	maps	map	NOUN
fcis-14812	138	37	at	at	ADP
fcis-14812	138	38	each	each	DET
fcis-14812	138	39	scale	scale	NOUN
fcis-14812	138	40	.	.	PUNCT
fcis-14812	139	1	the	the	DET
fcis-14812	139	2	process	process	NOUN
fcis-14812	139	3	consists	consist	VERB
fcis-14812	139	4	of	of	ADP
fcis-14812	139	5	two	two	NUM
fcis-14812	139	6	steps	step	NOUN
fcis-14812	139	7	:	:	PUNCT
fcis-14812	139	8	constant	constant	ADJ
fcis-14812	139	9	scaling	scaling	NOUN
fcis-14812	139	10	and	and	CCONJ
fcis-14812	139	11	adaptive	adaptive	ADJ
fcis-14812	139	12	fusion	fusion	NOUN
fcis-14812	139	13	.	.	PUNCT
fcis-14812	140	1	the	the	DET
fcis-14812	140	2	asff	asff	PROPN
fcis-14812	140	3	structure	structure	NOUN
fcis-14812	140	4	is	be	AUX
fcis-14812	140	5	shown	show	VERB
fcis-14812	140	6	in	in	ADP
fcis-14812	140	7	fig	fig	NOUN
fcis-14812	140	8	.	.	PUNCT
fcis-14812	141	1	2	2	NUM
fcis-14812	141	2	figure	figure	NOUN
fcis-14812	141	3	2	2	NUM
fcis-14812	141	4	.	.	PUNCT
fcis-14812	141	5	illustration	illustration	NOUN
fcis-14812	141	6	of	of	ADP
fcis-14812	141	7	the	the	DET
fcis-14812	141	8	adaptively	adaptively	ADV
fcis-14812	141	9	spatial	spatial	ADJ
fcis-14812	141	10	feature	feature	NOUN
fcis-14812	141	11	fusion	fusion	NOUN
fcis-14812	141	12	mechanism	mechanism	NOUN
fcis-14812	141	13	feature	feature	NOUN
fcis-14812	141	14	resizing	resize	VERB
fcis-14812	141	15	.	.	PUNCT
fcis-14812	142	1	we	we	PRON
fcis-14812	142	2	denote	denote	VERB
fcis-14812	142	3	the	the	DET
fcis-14812	142	4	features	feature	NOUN
fcis-14812	142	5	of	of	ADP
fcis-14812	142	6	the	the	DET
fcis-14812	142	7	resolution	resolution	NOUN
fcis-14812	142	8	at	at	ADP
fcis-14812	142	9	level	level	NOUN
fcis-14812	142	10	l	l	NOUN
fcis-14812	142	11	as	as	ADP
fcis-14812	142	12	lx	lx	NOUN
fcis-14812	142	13	.	.	PUNCT
fcis-14812	143	1	for	for	ADP
fcis-14812	143	2	level	level	NOUN
fcis-14812	143	3	l	l	NOUN
fcis-14812	143	4	,	,	PUNCT
fcis-14812	143	5	we	we	PRON
fcis-14812	143	6	resize	resize	VERB
fcis-14812	143	7	the	the	DET
fcis-14812	143	8	features	feature	NOUN
fcis-14812	143	9	nx	nx	NOUN
fcis-14812	143	10	at	at	ADP
fcis-14812	143	11	the	the	DET
fcis-14812	143	12	other	other	ADJ
fcis-14812	143	13	level	level	NOUN
fcis-14812	143	14			NOUN
fcis-14812	143	15	lnn	lnn	NOUN
fcis-14812	143	16			VERB
fcis-14812	143	17	to	to	ADP
fcis-14812	143	18	the	the	DET
fcis-14812	143	19	same	same	ADJ
fcis-14812	143	20	shape	shape	NOUN
fcis-14812	143	21	as	as	ADP
fcis-14812	143	22	that	that	PRON
fcis-14812	143	23	of	of	ADP
fcis-14812	143	24	lx	lx	NOUN
fcis-14812	143	25	.for	.for	PUNCT
fcis-14812	144	1	upsampling	upsample	VERB
fcis-14812	144	2	,	,	PUNCT
fcis-14812	144	3	we	we	PRON
fcis-14812	144	4	first	first	ADV
fcis-14812	144	5	apply	apply	VERB
fcis-14812	144	6	a	a	DET
fcis-14812	144	7	1×1convolution	1×1convolution	NUM
fcis-14812	144	8	layer	layer	NOUN
fcis-14812	144	9	to	to	PART
fcis-14812	144	10	compress	compress	VERB
fcis-14812	144	11	the	the	DET
fcis-14812	144	12	number	number	NOUN
fcis-14812	144	13	of	of	ADP
fcis-14812	144	14	channels	channel	NOUN
fcis-14812	144	15	of	of	ADP
fcis-14812	144	16	features	feature	NOUN
fcis-14812	144	17	to	to	ADP
fcis-14812	144	18	that	that	PRON
fcis-14812	144	19	in	in	ADP
fcis-14812	144	20	level	level	NOUN
fcis-14812	144	21	l	l	NOUN
fcis-14812	144	22	,	,	PUNCT
fcis-14812	144	23	and	and	CCONJ
fcis-14812	144	24	then	then	ADV
fcis-14812	144	25	interpolated	interpolate	VERB
fcis-14812	144	26	to	to	PART
fcis-14812	144	27	increase	increase	VERB
fcis-14812	144	28	the	the	DET
fcis-14812	144	29	resolution	resolution	NOUN
fcis-14812	144	30	respectively	respectively	ADV
fcis-14812	144	31	.	.	PUNCT
fcis-14812	145	1	for	for	ADP
fcis-14812	145	2	1/2ratio	1/2ratio	NUM
fcis-14812	145	3	down	down	ADV
fcis-14812	145	4	-	-	PUNCT
fcis-14812	145	5	sampling	sampling	NOUN
fcis-14812	145	6	,	,	PUNCT
fcis-14812	145	7	a	a	DET
fcis-14812	145	8	3×3	3×3	NUM
fcis-14812	145	9	convolution	convolution	NOUN
fcis-14812	145	10	layer	layer	NOUN
fcis-14812	145	11	with	with	ADP
fcis-14812	145	12	a	a	DET
fcis-14812	145	13	2	2	NUM
fcis-14812	145	14	-	-	PUNCT
fcis-14812	145	15	step	step	NOUN
fcis-14812	145	16	length	length	NOUN
fcis-14812	145	17	is	be	AUX
fcis-14812	145	18	used	use	VERB
fcis-14812	145	19	,	,	PUNCT
fcis-14812	145	20	while	while	SCONJ
fcis-14812	145	21	modifying	modify	VERB
fcis-14812	145	22	the	the	DET
fcis-14812	145	23	number	number	NOUN
fcis-14812	145	24	of	of	ADP
fcis-14812	145	25	channels	channel	NOUN
fcis-14812	145	26	and	and	CCONJ
fcis-14812	145	27	resolution	resolution	NOUN
fcis-14812	145	28	.	.	PUNCT
fcis-14812	146	1	for	for	ADP
fcis-14812	146	2	1/4	1/4	NUM
fcis-14812	146	3	ratio	ratio	NOUN
fcis-14812	146	4	down	down	ADP
fcis-14812	146	5	-	-	PUNCT
fcis-14812	146	6	sampling	sampling	NOUN
fcis-14812	146	7	,	,	PUNCT
fcis-14812	146	8	a	a	DET
fcis-14812	146	9	2	2	NUM
fcis-14812	146	10	-	-	PUNCT
fcis-14812	146	11	step	step	NOUN
fcis-14812	146	12	maximum	maximum	ADJ
fcis-14812	146	13	pooling	pooling	NOUN
fcis-14812	146	14	layer	layer	NOUN
fcis-14812	146	15	is	be	AUX
fcis-14812	146	16	added	add	VERB
fcis-14812	146	17	before	before	ADP
fcis-14812	146	18	the	the	DET
fcis-14812	146	19	2	2	NUM
fcis-14812	146	20	-	-	PUNCT
fcis-14812	146	21	step	step	NOUN
fcis-14812	146	22	convolution	convolution	NOUN
fcis-14812	146	23	.	.	PUNCT
fcis-14812	147	1	adaptive	adaptive	ADJ
fcis-14812	147	2	fusion	fusion	NOUN
fcis-14812	147	3	.	.	PUNCT
fcis-14812	148	1	let	let	VERB
fcis-14812	148	2	ln	ln	ADV
fcis-14812	148	3	ijx	ijx	NOUN
fcis-14812	148	4			NOUN
fcis-14812	148	5	denote	denote	VERB
fcis-14812	148	6	the	the	DET
fcis-14812	148	7	feature	feature	NOUN
fcis-14812	148	8	vector	vector	NOUN
fcis-14812	148	9	at	at	ADP
fcis-14812	148	10	the	the	DET
fcis-14812	148	11	position	position	NOUN
fcis-14812	148	12			NOUN
fcis-14812	148	13	ji	ji	NOUN
fcis-14812	148	14	,	,	PUNCT
fcis-14812	148	15	on	on	ADP
fcis-14812	148	16	the	the	DET
fcis-14812	148	17	feature	feature	NOUN
fcis-14812	148	18	maps	map	NOUN
fcis-14812	148	19	resized	resize	VERB
fcis-14812	148	20	from	from	ADP
fcis-14812	148	21	level	level	NOUN
fcis-14812	148	22	n	n	ADP
fcis-14812	148	23	to	to	PART
fcis-14812	148	24	level	level	VERB
fcis-14812	148	25	l	l	NOUN
fcis-14812	148	26	,	,	PUNCT
fcis-14812	148	27	then	then	ADV
fcis-14812	148	28	the	the	DET
fcis-14812	148	29	feature	feature	NOUN
fcis-14812	148	30	fusion	fusion	NOUN
fcis-14812	148	31	corresponding	correspond	VERB
fcis-14812	148	32	to	to	ADP
fcis-14812	148	33	level	level	NOUN
fcis-14812	148	34	l	l	NOUN
fcis-14812	148	35	is	be	AUX
fcis-14812	148	36	as	as	SCONJ
fcis-14812	148	37	follows	follow	VERB
fcis-14812	148	38	:	:	PUNCT
fcis-14812	149	1	l	l	NOUN
fcis-14812	149	2	ij	ij	INTJ
fcis-14812	149	3	l	l	NOUN
fcis-14812	149	4	ij	ij	X
fcis-14812	149	5	l	l	NOUN
fcis-14812	149	6	ij	ij	X
fcis-14812	149	7	l	l	NOUN
fcis-14812	149	8	ij	ij	X
fcis-14812	149	9	l	l	NOUN
fcis-14812	149	10	ij	ij	X
fcis-14812	149	11	l	l	NOUN
fcis-14812	149	12	ij	ij	INTJ
fcis-14812	149	13	l	l	NOUN
fcis-14812	149	14	ij	ij	NOUN
fcis-14812	149	15	xxxy	xxxy	PROPN
fcis-14812	149	16			PROPN
fcis-14812	149	17			PROPN
fcis-14812	149	18	321	321	NUM
fcis-14812	149	19			NOUN
fcis-14812	149	20	(	(	PUNCT
fcis-14812	149	21	8)	8)	NUM
fcis-14812	149	22	where	where	SCONJ
fcis-14812	149	23	l	l	NOUN
fcis-14812	149	24	ijy	ijy	NOUN
fcis-14812	149	25	implies	imply	VERB
fcis-14812	149	26	the	the	DET
fcis-14812	149	27			PROPN
fcis-14812	149	28	ji	ji	PROPN
fcis-14812	149	29	,	,	PUNCT
fcis-14812	149	30	-th	-th	ADJ
fcis-14812	149	31	vector	vector	NOUN
fcis-14812	149	32	of	of	ADP
fcis-14812	149	33	the	the	DET
fcis-14812	149	34	output	output	NOUN
fcis-14812	149	35	feature	feature	NOUN
fcis-14812	149	36	maps	map	NOUN
fcis-14812	149	37	ly	ly	VERB
fcis-14812	149	38	among	among	ADP
fcis-14812	149	39	channels	channel	NOUN
fcis-14812	149	40	.	.	PUNCT
fcis-14812	150	1	l	l	NOUN
fcis-14812	150	2	ij	ij	NOUN
fcis-14812	150	3	,	,	PUNCT
fcis-14812	150	4	l	l	PROPN
fcis-14812	150	5	ij	ij	PROPN
fcis-14812	150	6	and	and	CCONJ
fcis-14812	150	7	l	l	NOUN
fcis-14812	150	8	ij	ij	NOUN
fcis-14812	150	9	refer	refer	VERB
fcis-14812	150	10	to	to	ADP
fcis-14812	150	11	the	the	DET
fcis-14812	150	12	spatial	spatial	ADJ
fcis-14812	150	13	importance	importance	NOUN
fcis-14812	150	14	weights	weight	VERB
fcis-14812	150	15	for	for	SCONJ
fcis-14812	150	16	the	the	DET
fcis-14812	150	17	feature	feature	NOUN
fcis-14812	150	18	maps	map	NOUN
fcis-14812	150	19	at	at	ADP
fcis-14812	150	20	three	three	NUM
fcis-14812	150	21	different	different	ADJ
fcis-14812	150	22	levels	level	NOUN
fcis-14812	150	23	to	to	PART
fcis-14812	150	24	level	level	VERB
fcis-14812	150	25	l	l	NOUN
fcis-14812	150	26	,	,	PUNCT
fcis-14812	150	27	which	which	PRON
fcis-14812	150	28	are	be	AUX
fcis-14812	150	29	adaptively	adaptively	ADV
fcis-14812	150	30	learned	learn	VERB
fcis-14812	150	31	by	by	ADP
fcis-14812	150	32	the	the	DET
fcis-14812	150	33	network	network	NOUN
fcis-14812	150	34	.	.	PUNCT
fcis-14812	151	1	note	note	VERB
fcis-14812	151	2	that	that	SCONJ
fcis-14812	151	3	l	l	NOUN
fcis-14812	151	4	ij	ij	NOUN
fcis-14812	151	5	,	,	PUNCT
fcis-14812	151	6	l	l	PROPN
fcis-14812	151	7	ij	ij	PROPN
fcis-14812	151	8	and	and	CCONJ
fcis-14812	151	9	l	l	NOUN
fcis-14812	151	10	ij	ij	NOUN
fcis-14812	151	11	can	can	AUX
fcis-14812	151	12	be	be	AUX
fcis-14812	151	13	simple	simple	ADJ
fcis-14812	151	14	scalar	scalar	ADJ
fcis-14812	151	15	variables	variable	NOUN
fcis-14812	151	16	,	,	PUNCT
fcis-14812	151	17	which	which	PRON
fcis-14812	151	18	are	be	AUX
fcis-14812	151	19	shared	share	VERB
fcis-14812	151	20	across	across	ADP
fcis-14812	151	21	all	all	DET
fcis-14812	151	22	the	the	DET
fcis-14812	151	23	channels	channel	NOUN
fcis-14812	151	24	.	.	PUNCT
fcis-14812	152	1	we	we	PRON
fcis-14812	152	2	forced	force	VERB
fcis-14812	152	3	1	1	NUM
fcis-14812	152	4	l	l	NOUN
fcis-14812	152	5	ij	ij	NOUN
fcis-14812	152	6	l	l	NOUN
fcis-14812	152	7	ij	ij	X
fcis-14812	152	8	l	l	NOUN
fcis-14812	152	9	ij	ij	NOUN
fcis-14812	152	10			NOUN
fcis-14812	152	11	and	and	CCONJ
fcis-14812	152	12	l	l	NOUN
fcis-14812	152	13	ij	ij	NOUN
fcis-14812	152	14	,	,	PUNCT
fcis-14812	152	15	l	l	NOUN
fcis-14812	152	16	ij	ij	PROPN
fcis-14812	152	17	,	,	PUNCT
fcis-14812	152	18	l	l	PROPN
fcis-14812	152	19	ij	ij	NOUN
fcis-14812	152	20			ADJ
fcis-14812	152	21	10，	10，	PROPN
fcis-14812	152	22	,	,	PUNCT
fcis-14812	152	23	and	and	CCONJ
fcis-14812	152	24	define	define	VERB
fcis-14812	152	25	:	:	PUNCT
fcis-14812	153	1	l	l	NOUN
fcis-14812	153	2	ij	ij	INTJ
fcis-14812	153	3	l	l	NOUN
fcis-14812	153	4	ij	ij	X
fcis-14812	153	5	l	l	NOUN
fcis-14812	153	6	ij	ij	INTJ
fcis-14812	153	7	l	l	NOUN
fcis-14812	153	8	ij	ij	NOUN
fcis-14812	153	9	eee	eee	PROPN
fcis-14812	153	10	el	el	PROPN
fcis-14812	153	11	ij	ij	PROPN
fcis-14812	153	12			PRON
fcis-14812	153	13			X
fcis-14812	153	14			ADP
fcis-14812	153	15			ADJ
fcis-14812	153	16			X
fcis-14812	153	17			PROPN
fcis-14812	153	18			NOUN
fcis-14812	153	19	(	(	PUNCT
fcis-14812	153	20	9	9	NUM
fcis-14812	153	21	)	)	PUNCT
fcis-14812	153	22	here	here	ADV
fcis-14812	153	23	the	the	DET
fcis-14812	153	24	softmax	softmax	NOUN
fcis-14812	153	25	function	function	NOUN
fcis-14812	153	26	is	be	AUX
fcis-14812	153	27	used	use	VERB
fcis-14812	153	28	to	to	PART
fcis-14812	153	29	define	define	VERB
fcis-14812	153	30	l	l	NOUN
fcis-14812	153	31	ij	ij	NOUN
fcis-14812	153	32	,	,	PUNCT
fcis-14812	153	33	l	l	NOUN
fcis-14812	153	34	ij	ij	NOUN
fcis-14812	153	35	and	and	CCONJ
fcis-14812	153	36	l	l	PROPN
fcis-14812	153	37	ij	ij	NOUN
fcis-14812	153	38	with	with	ADP
fcis-14812	153	39	l	l	NOUN
fcis-14812	153	40	ij	ij	NOUN
fcis-14812	153	41	,	,	PUNCT
fcis-14812	153	42	l	l	PROPN
fcis-14812	153	43	ij	ij	NOUN
fcis-14812	153	44			ADJ
fcis-14812	153	45	and	and	CCONJ
fcis-14812	153	46	l	l	PROPN
fcis-14812	153	47	ij	ij	X
fcis-14812	153	48			ADJ
fcis-14812	153	49	each	each	PRON
fcis-14812	153	50	as	as	ADP
fcis-14812	153	51	a	a	DET
fcis-14812	153	52	control	control	NOUN
fcis-14812	153	53	parameter	parameter	NOUN
fcis-14812	153	54	respectively	respectively	ADV
fcis-14812	153	55	.	.	PUNCT
fcis-14812	154	1	we	we	PRON
fcis-14812	154	2	use	use	VERB
fcis-14812	154	3	1×1	1×1	ADJ
fcis-14812	154	4	convolutional	convolutional	ADJ
fcis-14812	154	5	layers	layer	NOUN
fcis-14812	154	6	to	to	PART
fcis-14812	154	7	compute	compute	VERB
fcis-14812	154	8	weight	weight	NOUN
fcis-14812	154	9	scalar	scalar	ADJ
fcis-14812	154	10	mappings	mapping	NOUN
fcis-14812	154	11	for	for	ADP
fcis-14812	154	12	l	l	NOUN
fcis-14812	154	13			X
fcis-14812	154	14	,	,	PUNCT
fcis-14812	154	15	l	l	NOUN
fcis-14812	154	16			NOUN
fcis-14812	154	17	,	,	PUNCT
fcis-14812	154	18	and	and	CCONJ
fcis-14812	154	19	l	l	NOUN
fcis-14812	154	20			NOUN
fcis-14812	154	21	from	from	ADP
fcis-14812	154	22	lx	lx	ADP
fcis-14812	154	23	1	1	NOUN
fcis-14812	154	24	,	,	PUNCT
fcis-14812	154	25	lx	lx	NOUN
fcis-14812	154	26	2	2	NOUN
fcis-14812	154	27	,	,	PUNCT
fcis-14812	154	28	and	and	CCONJ
fcis-14812	154	29	lx	lx	X
fcis-14812	154	30	3	3	X
fcis-14812	154	31	respectively	respectively	ADV
fcis-14812	154	32	,	,	PUNCT
fcis-14812	154	33	such	such	ADJ
fcis-14812	154	34	that	that	SCONJ
fcis-14812	154	35	they	they	PRON
fcis-14812	154	36	can	can	AUX
fcis-14812	154	37	be	be	AUX
fcis-14812	154	38	learned	learn	VERB
fcis-14812	154	39	by	by	ADP
fcis-14812	154	40	standard	standard	ADJ
fcis-14812	154	41	backpropagation	backpropagation	NOUN
fcis-14812	154	42	.	.	PUNCT
fcis-14812	155	1	dynamic	dynamic	ADJ
fcis-14812	155	2	similarity	similarity	NOUN
fcis-14812	155	3	metrics	metric	NOUN
fcis-14812	155	4	.	.	PUNCT
fcis-14812	156	1	given	give	VERB
fcis-14812	156	2	that	that	SCONJ
fcis-14812	156	3	humans	human	NOUN
fcis-14812	156	4	may	may	AUX
fcis-14812	156	5	base	base	VERB
fcis-14812	156	6	their	their	PRON
fcis-14812	156	7	categorization	categorization	NOUN
fcis-14812	156	8	judgments	judgment	NOUN
fcis-14812	156	9	on	on	ADP
fcis-14812	156	10	the	the	DET
fcis-14812	156	11	typical	typical	ADJ
fcis-14812	156	12	features	feature	NOUN
fcis-14812	156	13	of	of	ADP
fcis-14812	156	14	a	a	DET
fcis-14812	156	15	category	category	NOUN
fcis-14812	156	16	,	,	PUNCT
fcis-14812	156	17	for	for	ADP
fcis-14812	156	18	example	example	NOUN
fcis-14812	156	19	,	,	PUNCT
fcis-14812	156	20	if	if	SCONJ
fcis-14812	156	21	someone	someone	PRON
fcis-14812	156	22	describes	describe	VERB
fcis-14812	156	23	an	an	DET
fcis-14812	156	24	organism	organism	NOUN
fcis-14812	156	25	as	as	ADP
fcis-14812	156	26	having	have	VERB
fcis-14812	156	27	four	four	NUM
fcis-14812	156	28	limbs	limb	NOUN
fcis-14812	156	29	and	and	CCONJ
fcis-14812	156	30	hair	hair	NOUN
fcis-14812	156	31	-	-	PUNCT
fcis-14812	156	32	covered	cover	VERB
fcis-14812	156	33	auditory	auditory	ADJ
fcis-14812	156	34	organs	organ	NOUN
fcis-14812	156	35	,	,	PUNCT
fcis-14812	156	36	you	you	PRON
fcis-14812	156	37	might	might	AUX
fcis-14812	156	38	infer	infer	VERB
fcis-14812	156	39	that	that	SCONJ
fcis-14812	156	40	it	it	PRON
fcis-14812	156	41	is	be	AUX
fcis-14812	156	42	a	a	DET
fcis-14812	156	43	canine	canine	NOUN
fcis-14812	156	44	.	.	PUNCT
fcis-14812	157	1	therefore	therefore	ADV
fcis-14812	157	2	,	,	PUNCT
fcis-14812	157	3	shi	shi	PROPN
fcis-14812	157	4	et	et	PROPN
fcis-14812	157	5	al	al	PROPN
fcis-14812	158	1	[	[	X
fcis-14812	158	2	15	15	NUM
fcis-14812	158	3	]	]	PUNCT
fcis-14812	158	4	proposed	propose	VERB
fcis-14812	158	5	a	a	DET
fcis-14812	158	6	dynamic	dynamic	ADJ
fcis-14812	158	7	similarity	similarity	NOUN
fcis-14812	158	8	metric	metric	ADJ
fcis-14812	158	9	(	(	PUNCT
fcis-14812	158	10	dsm	dsm	PROPN
fcis-14812	158	11	)	)	PUNCT
fcis-14812	158	12	that	that	PRON
fcis-14812	158	13	adaptively	adaptively	ADV
fcis-14812	158	14	learns	learn	VERB
fcis-14812	158	15	the	the	DET
fcis-14812	158	16	category	category	NOUN
fcis-14812	158	17	patterns	pattern	NOUN
fcis-14812	158	18	of	of	ADP
fcis-14812	158	19	target	target	NOUN
fcis-14812	158	20	samples	sample	NOUN
fcis-14812	158	21	without	without	ADP
fcis-14812	158	22	being	be	AUX
fcis-14812	158	23	restricted	restrict	VERB
fcis-14812	158	24	to	to	ADP
fcis-14812	158	25	specific	specific	ADJ
fcis-14812	158	26	categories	category	NOUN
fcis-14812	158	27	.	.	PUNCT
fcis-14812	159	1	specifically	specifically	ADV
fcis-14812	159	2	,	,	PUNCT
fcis-14812	159	3	this	this	PRON
fcis-14812	159	4	is	be	AUX
fcis-14812	159	5	an	an	DET
fcis-14812	159	6	integrated	integrate	VERB
fcis-14812	159	7	feature	feature	NOUN
fcis-14812	159	8	selection	selection	NOUN
fcis-14812	159	9	module	module	NOUN
fcis-14812	159	10	,	,	PUNCT
fcis-14812	159	11	similar	similar	ADJ
fcis-14812	159	12	to	to	ADP
fcis-14812	159	13	senet	senet	NOUN
fcis-14812	159	14	[	[	X
fcis-14812	159	15	32	32	NUM
fcis-14812	159	16	]	]	PUNCT
fcis-14812	159	17	,	,	PUNCT
fcis-14812	159	18	it	it	PRON
fcis-14812	159	19	learns	learn	VERB
fcis-14812	159	20	the	the	DET
fcis-14812	159	21	dynamic	dynamic	ADJ
fcis-14812	159	22	channel	channel	NOUN
fcis-14812	159	23	attention	attention	NOUN
fcis-14812	159	24	weights	weight	VERB
fcis-14812	159	25	a	a	DET
fcis-14812	159	26	conditioned	condition	VERB
fcis-14812	159	27	on	on	ADP
fcis-14812	159	28	zbqz	zbqz	PROPN
fcis-14812	159	29			VERB
fcis-14812	159	30	such	such	ADJ
fcis-14812	159	31	that	that	SCONJ
fcis-14812	159	32	the	the	DET
fcis-14812	159	33	similarity	similarity	NOUN
fcis-14812	159	34	s	s	VERB
fcis-14812	159	35	can	can	AUX
fcis-14812	159	36	be	be	AUX
fcis-14812	159	37	computed	compute	VERB
fcis-14812	159	38	by	by	ADP
fcis-14812	159	39			NOUN
fcis-14812	159	40			SYM
fcis-14812	159	41			NOUN
fcis-14812	159	42			NOUN
fcis-14812	159	43			NOUN
fcis-14812	159	44			NOUN
fcis-14812	159	45			PRON
fcis-14812	159	46	xt	xt	VERB
fcis-14812	159	47	xijij	xijij	PROPN
fcis-14812	159	48	bqzabpxzxs	bqzabpxzxs	PROPN
fcis-14812	159	49			PUNCT
fcis-14812	159	50			PROPN
fcis-14812	159	51	,	,	PUNCT
fcis-14812	159	52	(	(	PUNCT
fcis-14812	159	53	10	10	NUM
fcis-14812	159	54	)	)	PUNCT
fcis-14812	159	55	where	where	PROPN
fcis-14812	159	56	denotes	denote	VERB
fcis-14812	159	57	the	the	DET
fcis-14812	159	58	hadamard	hadamard	ADJ
fcis-14812	159	59	product	product	NOUN
fcis-14812	159	60	.	.	PUNCT
fcis-14812	160	1	counter	counter	NOUN
fcis-14812	160	2	module	module	NOUN
fcis-14812	160	3	.	.	PUNCT
fcis-14812	161	1	the	the	DET
fcis-14812	161	2	counter	counter	NOUN
fcis-14812	161	3	module	module	NOUN
fcis-14812	161	4	consists	consist	VERB
fcis-14812	161	5	of	of	ADP
fcis-14812	161	6	several	several	ADJ
fcis-14812	161	7	convolutional	convolutional	ADJ
fcis-14812	161	8	and	and	CCONJ
fcis-14812	161	9	bilinear	bilinear	NOUN
fcis-14812	161	10	upsampling	upsample	VERB
fcis-14812	161	11	layers	layer	NOUN
fcis-14812	161	12	used	use	VERB
fcis-14812	161	13	to	to	PART
fcis-14812	161	14	regress	regress	VERB
fcis-14812	161	15	a	a	DET
fcis-14812	161	16	density	density	NOUN
fcis-14812	161	17	map	map	NOUN
fcis-14812	161	18	of	of	ADP
fcis-14812	161	19	the	the	DET
fcis-14812	161	20	same	same	ADJ
fcis-14812	161	21	size	size	NOUN
fcis-14812	161	22	as	as	ADP
fcis-14812	161	23	the	the	DET
fcis-14812	161	24	query	query	NOUN
fcis-14812	161	25	image	image	NOUN
fcis-14812	161	26	.	.	PUNCT
fcis-14812	162	1	specifically	specifically	ADV
fcis-14812	162	2	,	,	PUNCT
fcis-14812	162	3	the	the	DET
fcis-14812	162	4	counting	counting	NOUN
fcis-14812	162	5	module	module	NOUN
fcis-14812	162	6	receives	receive	VERB
fcis-14812	162	7	the	the	DET
fcis-14812	162	8	channel	channel	NOUN
fcis-14812	162	9	-	-	PUNCT
fcis-14812	162	10	level	level	NOUN
fcis-14812	162	11	connections	connection	NOUN
fcis-14812	162	12	between	between	ADP
fcis-14812	162	13	query	query	NOUN
fcis-14812	162	14	feature	feature	NOUN
fcis-14812	162	15	map	map	NOUN
fcis-14812	162	16			PROPN
fcis-14812	162	17	xf	xf	PROPN
fcis-14812	162	18	and	and	CCONJ
fcis-14812	162	19	similarity	similarity	NOUN
fcis-14812	162	20	maps	map	NOUN
fcis-14812	162	21	s	s	PART
fcis-14812	162	22	,	,	PUNCT
fcis-14812	162	23	and	and	CCONJ
fcis-14812	162	24	then	then	ADV
fcis-14812	162	25	predicts	predict	VERB
fcis-14812	162	26	a	a	DET
fcis-14812	162	27	density	density	NOUN
fcis-14812	162	28	map	map	NOUN
fcis-14812	162	29	prd	prd	NOUN
fcis-14812	162	30	.	.	PUNCT
fcis-14812	163	1	the	the	DET
fcis-14812	163	2	final	final	ADJ
fcis-14812	163	3	count	count	NOUN
fcis-14812	163	4	results	result	NOUN
fcis-14812	163	5	in	in	ADP
fcis-14812	163	6	an	an	DET
fcis-14812	163	7	integral	integral	NOUN
fcis-14812	163	8	of	of	ADP
fcis-14812	163	9	prd	prd	NOUN
fcis-14812	163	10	.	.	PUNCT
fcis-14812	164	1	3.2	3.2	NUM
fcis-14812	164	2	.	.	PUNCT
fcis-14812	165	1	experiment	experiment	NOUN
fcis-14812	165	2	details	detail	NOUN
fcis-14812	165	3	data	datum	NOUN
fcis-14812	165	4	pre	pre	ADJ
fcis-14812	165	5	-	-	NOUN
fcis-14812	165	6	processing	processing	ADJ
fcis-14812	165	7	.	.	PUNCT
fcis-14812	166	1	in	in	ADP
fcis-14812	166	2	this	this	DET
fcis-14812	166	3	work	work	NOUN
fcis-14812	166	4	,	,	PUNCT
fcis-14812	166	5	we	we	PRON
fcis-14812	166	6	adopt	adopt	VERB
fcis-14812	166	7	the	the	DET
fcis-14812	166	8	same	same	ADJ
fcis-14812	166	9	data	datum	NOUN
fcis-14812	166	10	preprocessing	preprocessing	NOUN
fcis-14812	166	11	method	method	NOUN
fcis-14812	166	12	as	as	ADP
fcis-14812	166	13	bmnet+	bmnet+	NOUN
fcis-14812	166	14	to	to	PART
fcis-14812	166	15	ensure	ensure	VERB
fcis-14812	166	16	the	the	DET
fcis-14812	166	17	consistency	consistency	NOUN
fcis-14812	166	18	and	and	CCONJ
fcis-14812	166	19	comparability	comparability	NOUN
fcis-14812	166	20	of	of	ADP
fcis-14812	166	21	the	the	DET
fcis-14812	166	22	data	datum	NOUN
fcis-14812	166	23	.	.	PUNCT
fcis-14812	167	1	the	the	DET
fcis-14812	167	2	data	datum	NOUN
fcis-14812	167	3	preprocessing	preprocessing	NOUN
fcis-14812	167	4	method	method	NOUN
fcis-14812	167	5	consists	consist	VERB
fcis-14812	167	6	of	of	ADP
fcis-14812	167	7	two	two	NUM
fcis-14812	167	8	main	main	ADJ
fcis-14812	167	9	steps	step	NOUN
fcis-14812	167	10	:	:	PUNCT
fcis-14812	167	11	image	image	NOUN
fcis-14812	167	12	scaling	scaling	NOUN
fcis-14812	167	13	and	and	CCONJ
fcis-14812	167	14	image	image	NOUN
fcis-14812	167	15	cropping	cropping	NOUN
fcis-14812	167	16	.	.	PUNCT
fcis-14812	168	1	the	the	DET
fcis-14812	168	2	purpose	purpose	NOUN
fcis-14812	168	3	of	of	ADP
fcis-14812	168	4	image	image	NOUN
fcis-14812	168	5	scaling	scale	VERB
fcis-14812	168	6	is	be	AUX
fcis-14812	168	7	to	to	PART
fcis-14812	168	8	resize	resize	VERB
fcis-14812	168	9	the	the	DET
fcis-14812	168	10	query	query	NOUN
fcis-14812	168	11	image	image	NOUN
fcis-14812	168	12	to	to	ADP
fcis-14812	168	13	a	a	DET
fcis-14812	168	14	suitable	suitable	ADJ
fcis-14812	168	15	range	range	NOUN
fcis-14812	168	16	to	to	PART
fcis-14812	168	17	accommodate	accommodate	VERB
fcis-14812	168	18	image	image	NOUN
fcis-14812	168	19	inputs	input	NOUN
fcis-14812	168	20	with	with	ADP
fcis-14812	168	21	different	different	ADJ
fcis-14812	168	22	resolutions	resolution	NOUN
fcis-14812	168	23	.	.	PUNCT
fcis-14812	169	1	we	we	PRON
fcis-14812	169	2	use	use	VERB
fcis-14812	169	3	an	an	DET
fcis-14812	169	4	aspect	aspect	NOUN
fcis-14812	169	5	ratio	ratio	NOUN
fcis-14812	169	6	-	-	PUNCT
fcis-14812	169	7	based	base	VERB
fcis-14812	169	8	scaling	scaling	NOUN
fcis-14812	169	9	method	method	NOUN
fcis-14812	169	10	,	,	PUNCT
fcis-14812	169	11	i.e.	i.e.	X
fcis-14812	169	12	,	,	PUNCT
fcis-14812	169	13	scaling	scale	VERB
fcis-14812	169	14	the	the	DET
fcis-14812	169	15	longest	long	ADJ
fcis-14812	169	16	side	side	NOUN
fcis-14812	169	17	of	of	ADP
fcis-14812	169	18	the	the	DET
fcis-14812	169	19	query	query	NOUN
fcis-14812	169	20	image	image	NOUN
fcis-14812	169	21	to	to	ADP
fcis-14812	169	22	a	a	DET
fcis-14812	169	23	preset	preset	ADJ
fcis-14812	169	24	interval	interval	NOUN
fcis-14812	169	25	,	,	PUNCT
fcis-14812	169	26	i.e.	i.e.	X
fcis-14812	169	27	,	,	PUNCT
fcis-14812	169	28	[	[	X
fcis-14812	169	29	384,1584	384,1584	NUM
fcis-14812	169	30	]	]	PUNCT
fcis-14812	169	31	,	,	PUNCT
fcis-14812	169	32	while	while	SCONJ
fcis-14812	169	33	keeping	keep	VERB
fcis-14812	169	34	the	the	DET
fcis-14812	169	35	original	original	ADJ
fcis-14812	169	36	aspect	aspect	NOUN
fcis-14812	169	37	ratio	ratio	NOUN
fcis-14812	169	38	of	of	ADP
fcis-14812	169	39	the	the	DET
fcis-14812	169	40	query	query	NOUN
fcis-14812	169	41	image	image	NOUN
fcis-14812	169	42	unchanged	unchanged	ADJ
fcis-14812	169	43	.	.	PUNCT
fcis-14812	170	1	this	this	PRON
fcis-14812	170	2	avoids	avoid	VERB
fcis-14812	170	3	the	the	DET
fcis-14812	170	4	problem	problem	NOUN
fcis-14812	170	5	of	of	ADP
fcis-14812	170	6	distortion	distortion	NOUN
fcis-14812	170	7	or	or	CCONJ
fcis-14812	170	8	deformation	deformation	NOUN
fcis-14812	170	9	of	of	ADP
fcis-14812	170	10	the	the	DET
fcis-14812	170	11	image	image	NOUN
fcis-14812	170	12	during	during	ADP
fcis-14812	170	13	the	the	DET
fcis-14812	170	14	scaling	scaling	ADJ
fcis-14812	170	15	process	process	NOUN
fcis-14812	170	16	.	.	PUNCT
fcis-14812	171	1	the	the	DET
fcis-14812	171	2	purpose	purpose	NOUN
fcis-14812	171	3	of	of	ADP
fcis-14812	171	4	image	image	NOUN
fcis-14812	171	5	cropping	cropping	NOUN
fcis-14812	171	6	is	be	AUX
fcis-14812	171	7	to	to	PART
fcis-14812	171	8	unify	unify	VERB
fcis-14812	171	9	the	the	DET
fcis-14812	171	10	size	size	NOUN
fcis-14812	171	11	of	of	ADP
fcis-14812	171	12	the	the	DET
fcis-14812	171	13	sample	sample	NOUN
fcis-14812	171	14	images	image	NOUN
fcis-14812	171	15	to	to	ADP
fcis-14812	171	16	128	128	NUM
fcis-14812	171	17	×	×	NOUN
fcis-14812	171	18	128	128	NUM
fcis-14812	171	19	for	for	ADP
fcis-14812	171	20	subsequent	subsequent	ADJ
fcis-14812	171	21	feature	feature	NOUN
fcis-14812	171	22	extraction	extraction	NOUN
fcis-14812	171	23	and	and	CCONJ
fcis-14812	171	24	similarity	similarity	NOUN
fcis-14812	171	25	metrics	metric	NOUN
fcis-14812	171	26	.	.	PUNCT
fcis-14812	172	1	we	we	PRON
fcis-14812	172	2	use	use	VERB
fcis-14812	172	3	a	a	DET
fcis-14812	172	4	center	center	NOUN
fcis-14812	172	5	-	-	PUNCT
fcis-14812	172	6	based	base	VERB
fcis-14812	172	7	cropping	cropping	NOUN
fcis-14812	172	8	method	method	NOUN
fcis-14812	172	9	,	,	PUNCT
fcis-14812	172	10	i.e.	i.e.	X
fcis-14812	172	11	,	,	PUNCT
fcis-14812	172	12	starting	start	VERB
fcis-14812	172	13	from	from	ADP
fcis-14812	172	14	the	the	DET
fcis-14812	172	15	center	center	NOUN
fcis-14812	172	16	of	of	ADP
fcis-14812	172	17	the	the	DET
fcis-14812	172	18	sample	sample	NOUN
fcis-14812	172	19	image	image	NOUN
fcis-14812	172	20	,	,	PUNCT
fcis-14812	172	21	128	128	NUM
fcis-14812	172	22	pixels	pixel	NOUN
fcis-14812	172	23	are	be	AUX
fcis-14812	172	24	cropped	crop	VERB
fcis-14812	172	25	out	out	ADP
fcis-14812	172	26	horizontally	horizontally	ADV
fcis-14812	172	27	and	and	CCONJ
fcis-14812	172	28	vertically	vertically	ADV
fcis-14812	172	29	along	along	ADP
fcis-14812	172	30	the	the	DET
fcis-14812	172	31	horizontal	horizontal	ADJ
fcis-14812	172	32	and	and	CCONJ
fcis-14812	172	33	vertical	vertical	ADJ
fcis-14812	172	34	directions	direction	NOUN
fcis-14812	172	35	,	,	PUNCT
fcis-14812	172	36	respectively	respectively	ADV
fcis-14812	172	37	,	,	PUNCT
fcis-14812	172	38	to	to	PART
fcis-14812	172	39	obtain	obtain	VERB
fcis-14812	172	40	a	a	DET
fcis-14812	172	41	square	square	ADJ
fcis-14812	172	42	region	region	NOUN
fcis-14812	172	43	as	as	ADP
fcis-14812	172	44	the	the	DET
fcis-14812	172	45	final	final	ADJ
fcis-14812	172	46	sample	sample	NOUN
fcis-14812	172	47	image	image	NOUN
fcis-14812	172	48	.	.	PUNCT
fcis-14812	173	1	this	this	PRON
fcis-14812	173	2	preserves	preserve	VERB
fcis-14812	173	3	the	the	DET
fcis-14812	173	4	most	most	ADV
fcis-14812	173	5	important	important	ADJ
fcis-14812	173	6	parts	part	NOUN
fcis-14812	173	7	of	of	ADP
fcis-14812	173	8	the	the	DET
fcis-14812	173	9	sample	sample	NOUN
fcis-14812	173	10	image	image	NOUN
fcis-14812	173	11	while	while	SCONJ
fcis-14812	173	12	reducing	reduce	VERB
fcis-14812	173	13	the	the	DET
fcis-14812	173	14	interference	interference	NOUN
fcis-14812	173	15	of	of	ADP
fcis-14812	173	16	irrelevant	irrelevant	ADJ
fcis-14812	173	17	information	information	NOUN
fcis-14812	173	18	.	.	PUNCT
fcis-14812	174	1	figure	figure	VERB
fcis-14812	174	2	3	3	NUM
fcis-14812	174	3	.	.	PROPN
fcis-14812	174	4	resnet50	resnet50	PROPN
fcis-14812	174	5	structure	structure	PROPN
fcis-14812	174	6	diagram	diagram	PROPN
fcis-14812	174	7	network	network	NOUN
fcis-14812	174	8	design	design	NOUN
fcis-14812	174	9	.	.	PUNCT
fcis-14812	175	1	to	to	PART
fcis-14812	175	2	extract	extract	VERB
fcis-14812	175	3	effective	effective	ADJ
fcis-14812	175	4	features	feature	NOUN
fcis-14812	175	5	from	from	ADP
fcis-14812	175	6	images	image	NOUN
fcis-14812	175	7	,	,	PUNCT
fcis-14812	175	8	74	74	NUM
fcis-14812	175	9	we	we	PRON
fcis-14812	175	10	employ	employ	VERB
fcis-14812	175	11	a	a	DET
fcis-14812	175	12	deep	deep	ADJ
fcis-14812	175	13	residual	residual	ADJ
fcis-14812	175	14	network	network	NOUN
fcis-14812	175	15	(	(	PUNCT
fcis-14812	175	16	resnet-50	resnet-50	NOUN
fcis-14812	175	17	)	)	PUNCT
fcis-14812	175	18	based	base	VERB
fcis-14812	175	19	backbone	backbone	NOUN
fcis-14812	175	20	network	network	NOUN
fcis-14812	175	21	as	as	ADP
fcis-14812	175	22	a	a	DET
fcis-14812	175	23	feature	feature	NOUN
fcis-14812	175	24	extractor	extractor	NOUN
fcis-14812	175	25	.	.	PUNCT
fcis-14812	176	1	the	the	DET
fcis-14812	176	2	schematic	schematic	ADJ
fcis-14812	176	3	diagram	diagram	NOUN
fcis-14812	176	4	of	of	ADP
fcis-14812	176	5	resnet50	resnet50	NOUN
fcis-14812	176	6	structure	structure	NOUN
fcis-14812	176	7	is	be	AUX
fcis-14812	176	8	shown	show	VERB
fcis-14812	176	9	in	in	ADP
fcis-14812	176	10	fig	fig	NOUN
fcis-14812	176	11	.	.	PUNCT
fcis-14812	177	1	3	3	X
fcis-14812	177	2	.	.	X
fcis-14812	177	3	the	the	DET
fcis-14812	177	4	backbone	backbone	NOUN
fcis-14812	177	5	network	network	NOUN
fcis-14812	177	6	consists	consist	VERB
fcis-14812	177	7	of	of	ADP
fcis-14812	177	8	the	the	DET
fcis-14812	177	9	first	first	ADJ
fcis-14812	177	10	four	four	NUM
fcis-14812	177	11	residual	residual	ADJ
fcis-14812	177	12	blocks	block	NOUN
fcis-14812	177	13	of	of	ADP
fcis-14812	177	14	resnet-50	resnet-50	NUM
fcis-14812	177	15	,	,	PUNCT
fcis-14812	177	16	and	and	CCONJ
fcis-14812	177	17	each	each	DET
fcis-14812	177	18	residual	residual	ADJ
fcis-14812	177	19	block	block	NOUN
fcis-14812	177	20	contains	contain	VERB
fcis-14812	177	21	multiple	multiple	ADJ
fcis-14812	177	22	convolutional	convolutional	ADJ
fcis-14812	177	23	layers	layer	NOUN
fcis-14812	177	24	and	and	CCONJ
fcis-14812	177	25	jump	jump	NOUN
fcis-14812	177	26	connections	connection	NOUN
fcis-14812	177	27	.	.	PUNCT
fcis-14812	178	1	we	we	PRON
fcis-14812	178	2	extract	extract	VERB
fcis-14812	178	3	two	two	NUM
fcis-14812	178	4	different	different	ADJ
fcis-14812	178	5	levels	level	NOUN
fcis-14812	178	6	of	of	ADP
fcis-14812	178	7	feature	feature	NOUN
fcis-14812	178	8	maps	map	NOUN
fcis-14812	178	9	from	from	ADP
fcis-14812	178	10	the	the	DET
fcis-14812	178	11	backbone	backbone	NOUN
fcis-14812	178	12	network	network	NOUN
fcis-14812	178	13	,	,	PUNCT
fcis-14812	178	14	which	which	PRON
fcis-14812	178	15	are	be	AUX
fcis-14812	178	16	the	the	DET
fcis-14812	178	17	outputs	output	NOUN
fcis-14812	178	18	of	of	ADP
fcis-14812	178	19	the	the	DET
fcis-14812	178	20	third	third	ADJ
fcis-14812	178	21	residual	residual	ADJ
fcis-14812	178	22	block	block	NOUN
fcis-14812	178	23	(	(	PUNCT
fcis-14812	178	24	layer3	layer3	PROPN
fcis-14812	178	25	)	)	PUNCT
fcis-14812	178	26	and	and	CCONJ
fcis-14812	178	27	the	the	DET
fcis-14812	178	28	fourth	fourth	ADJ
fcis-14812	178	29	residual	residual	ADJ
fcis-14812	178	30	block	block	NOUN
fcis-14812	178	31	(	(	PUNCT
fcis-14812	178	32	layer4	layer4	NOUN
fcis-14812	178	33	)	)	PUNCT
fcis-14812	178	34	,	,	PUNCT
fcis-14812	178	35	and	and	CCONJ
fcis-14812	178	36	their	their	PRON
fcis-14812	178	37	channel	channel	NOUN
fcis-14812	178	38	counts	count	NOUN
fcis-14812	178	39	are	be	AUX
fcis-14812	178	40	1024	1024	NUM
fcis-14812	178	41	and	and	CCONJ
fcis-14812	178	42	2048	2048	NUM
fcis-14812	178	43	,	,	PUNCT
fcis-14812	178	44	respectively	respectively	ADV
fcis-14812	178	45	.	.	PUNCT
fcis-14812	179	1	these	these	DET
fcis-14812	179	2	two	two	NUM
fcis-14812	179	3	feature	feature	NOUN
fcis-14812	179	4	maps	map	NOUN
fcis-14812	179	5	represent	represent	VERB
fcis-14812	179	6	the	the	DET
fcis-14812	179	7	low	low	ADJ
fcis-14812	179	8	and	and	CCONJ
fcis-14812	179	9	medium	medium	ADJ
fcis-14812	179	10	level	level	NOUN
fcis-14812	179	11	and	and	CCONJ
fcis-14812	179	12	high	high	ADJ
fcis-14812	179	13	level	level	NOUN
fcis-14812	179	14	features	feature	NOUN
fcis-14812	179	15	of	of	ADP
fcis-14812	179	16	the	the	DET
fcis-14812	179	17	image	image	NOUN
fcis-14812	179	18	,	,	PUNCT
fcis-14812	179	19	respectively	respectively	ADV
fcis-14812	179	20	,	,	PUNCT
fcis-14812	179	21	which	which	PRON
fcis-14812	179	22	can	can	AUX
fcis-14812	179	23	capture	capture	VERB
fcis-14812	179	24	different	different	ADJ
fcis-14812	179	25	details	detail	NOUN
fcis-14812	179	26	and	and	CCONJ
fcis-14812	179	27	semantic	semantic	ADJ
fcis-14812	179	28	information	information	NOUN
fcis-14812	179	29	of	of	ADP
fcis-14812	179	30	the	the	DET
fcis-14812	179	31	image	image	NOUN
fcis-14812	179	32	.	.	PUNCT
fcis-14812	180	1	for	for	ADP
fcis-14812	180	2	each	each	DET
fcis-14812	180	3	query	query	NOUN
fcis-14812	180	4	image	image	NOUN
fcis-14812	180	5	,	,	PUNCT
fcis-14812	180	6	we	we	PRON
fcis-14812	180	7	use	use	VERB
fcis-14812	180	8	a	a	DET
fcis-14812	180	9	1×1	1×1	ADJ
fcis-14812	180	10	convolutional	convolutional	ADJ
fcis-14812	180	11	layer	layer	NOUN
fcis-14812	180	12	to	to	PART
fcis-14812	180	13	reduce	reduce	VERB
fcis-14812	180	14	the	the	DET
fcis-14812	180	15	number	number	NOUN
fcis-14812	180	16	of	of	ADP
fcis-14812	180	17	channels	channel	NOUN
fcis-14812	180	18	of	of	ADP
fcis-14812	180	19	the	the	DET
fcis-14812	180	20	two	two	NUM
fcis-14812	180	21	feature	feature	NOUN
fcis-14812	180	22	maps	map	NOUN
fcis-14812	180	23	to	to	ADP
fcis-14812	180	24	256	256	NUM
fcis-14812	180	25	to	to	PART
fcis-14812	180	26	reduce	reduce	VERB
fcis-14812	180	27	the	the	DET
fcis-14812	180	28	computation	computation	NOUN
fcis-14812	180	29	and	and	CCONJ
fcis-14812	180	30	memory	memory	NOUN
fcis-14812	180	31	consumption	consumption	NOUN
fcis-14812	180	32	.	.	PUNCT
fcis-14812	181	1	for	for	ADP
fcis-14812	181	2	each	each	DET
fcis-14812	181	3	sample	sample	NOUN
fcis-14812	181	4	image	image	NOUN
fcis-14812	181	5	,	,	PUNCT
fcis-14812	181	6	we	we	PRON
fcis-14812	181	7	first	first	ADV
fcis-14812	181	8	perform	perform	VERB
fcis-14812	181	9	a	a	DET
fcis-14812	181	10	global	global	ADJ
fcis-14812	181	11	average	average	ADJ
fcis-14812	181	12	pooling	pooling	NOUN
fcis-14812	181	13	(	(	PUNCT
fcis-14812	181	14	gap	gap	NOUN
fcis-14812	181	15	)	)	PUNCT
fcis-14812	181	16	operation	operation	NOUN
fcis-14812	181	17	on	on	ADP
fcis-14812	181	18	the	the	DET
fcis-14812	181	19	two	two	NUM
fcis-14812	181	20	feature	feature	NOUN
fcis-14812	181	21	maps	map	NOUN
fcis-14812	181	22	to	to	PART
fcis-14812	181	23	compress	compress	VERB
fcis-14812	181	24	each	each	DET
fcis-14812	181	25	feature	feature	NOUN
fcis-14812	181	26	map	map	NOUN
fcis-14812	181	27	into	into	ADP
fcis-14812	181	28	a	a	DET
fcis-14812	181	29	single	single	ADJ
fcis-14812	181	30	vector	vector	NOUN
fcis-14812	181	31	,	,	PUNCT
fcis-14812	181	32	and	and	CCONJ
fcis-14812	181	33	then	then	ADV
fcis-14812	181	34	use	use	VERB
fcis-14812	181	35	a	a	DET
fcis-14812	181	36	fully	fully	ADV
fcis-14812	181	37	-	-	PUNCT
fcis-14812	181	38	connected	connect	VERB
fcis-14812	181	39	layer	layer	NOUN
fcis-14812	181	40	to	to	PART
fcis-14812	181	41	stitch	stitch	VERB
fcis-14812	181	42	the	the	DET
fcis-14812	181	43	two	two	NUM
fcis-14812	181	44	vectors	vector	NOUN
fcis-14812	181	45	together	together	ADV
fcis-14812	181	46	and	and	CCONJ
fcis-14812	181	47	map	map	VERB
fcis-14812	181	48	them	they	PRON
fcis-14812	181	49	into	into	ADP
fcis-14812	181	50	a	a	DET
fcis-14812	181	51	256dimensional	256dimensional	PROPN
fcis-14812	181	52	feature	feature	NOUN
fcis-14812	181	53	vector	vector	NOUN
fcis-14812	181	54	.	.	PUNCT
fcis-14812	182	1	in	in	ADP
fcis-14812	182	2	this	this	DET
fcis-14812	182	3	way	way	NOUN
fcis-14812	182	4	,	,	PUNCT
fcis-14812	182	5	we	we	PRON
fcis-14812	182	6	obtain	obtain	VERB
fcis-14812	182	7	two	two	NUM
fcis-14812	182	8	different	different	ADJ
fcis-14812	182	9	forms	form	NOUN
fcis-14812	182	10	of	of	ADP
fcis-14812	182	11	feature	feature	NOUN
fcis-14812	182	12	representations	representation	NOUN
fcis-14812	182	13	,	,	PUNCT
fcis-14812	182	14	i.e.	i.e.	X
fcis-14812	182	15	,	,	PUNCT
fcis-14812	182	16	feature	feature	NOUN
fcis-14812	182	17	maps	map	NOUN
fcis-14812	182	18	and	and	CCONJ
fcis-14812	182	19	feature	feature	NOUN
fcis-14812	182	20	vectors	vector	NOUN
fcis-14812	182	21	,	,	PUNCT
fcis-14812	182	22	for	for	ADP
fcis-14812	182	23	each	each	DET
fcis-14812	182	24	query	query	NOUN
fcis-14812	182	25	image	image	NOUN
fcis-14812	182	26	and	and	CCONJ
fcis-14812	182	27	sample	sample	NOUN
fcis-14812	182	28	image	image	NOUN
fcis-14812	182	29	.	.	PUNCT
fcis-14812	183	1	in	in	ADP
fcis-14812	183	2	order	order	NOUN
fcis-14812	183	3	to	to	PART
fcis-14812	183	4	further	far	ADV
fcis-14812	183	5	enhance	enhance	VERB
fcis-14812	183	6	the	the	DET
fcis-14812	183	7	feature	feature	NOUN
fcis-14812	183	8	representation	representation	NOUN
fcis-14812	183	9	of	of	ADP
fcis-14812	183	10	the	the	DET
fcis-14812	183	11	query	query	NOUN
fcis-14812	183	12	image	image	NOUN
fcis-14812	183	13	,	,	PUNCT
fcis-14812	183	14	we	we	PRON
fcis-14812	183	15	design	design	VERB
fcis-14812	183	16	a	a	DET
fcis-14812	183	17	dual	dual	ADJ
fcis-14812	183	18	feature	feature	NOUN
fcis-14812	183	19	enhancement	enhancement	NOUN
fcis-14812	183	20	module	module	NOUN
fcis-14812	183	21	,	,	PUNCT
fcis-14812	183	22	which	which	PRON
fcis-14812	183	23	consists	consist	VERB
fcis-14812	183	24	of	of	ADP
fcis-14812	183	25	a	a	DET
fcis-14812	183	26	self	self	NOUN
fcis-14812	183	27	-	-	PUNCT
fcis-14812	183	28	similarity	similarity	NOUN
fcis-14812	183	29	module	module	NOUN
fcis-14812	183	30	and	and	CCONJ
fcis-14812	183	31	a	a	DET
fcis-14812	183	32	structured	structured	ADJ
fcis-14812	183	33	feature	feature	NOUN
fcis-14812	183	34	enhancement	enhancement	NOUN
fcis-14812	183	35	module	module	NOUN
fcis-14812	183	36	for	for	ADP
fcis-14812	183	37	sfem	sfem	NOUN
fcis-14812	183	38	.	.	PUNCT
fcis-14812	184	1	the	the	DET
fcis-14812	184	2	role	role	NOUN
fcis-14812	184	3	of	of	ADP
fcis-14812	184	4	the	the	DET
fcis-14812	184	5	selfsimilarity	selfsimilarity	NOUN
fcis-14812	184	6	module	module	NOUN
fcis-14812	184	7	is	be	AUX
fcis-14812	184	8	to	to	PART
fcis-14812	184	9	enhance	enhance	VERB
fcis-14812	184	10	the	the	DET
fcis-14812	184	11	feature	feature	NOUN
fcis-14812	184	12	map	map	NOUN
fcis-14812	184	13	of	of	ADP
fcis-14812	184	14	the	the	DET
fcis-14812	184	15	query	query	NOUN
fcis-14812	184	16	image	image	NOUN
fcis-14812	184	17	by	by	ADP
fcis-14812	184	18	utilizing	utilize	VERB
fcis-14812	184	19	the	the	DET
fcis-14812	184	20	information	information	NOUN
fcis-14812	184	21	of	of	ADP
fcis-14812	184	22	the	the	DET
fcis-14812	184	23	query	query	NOUN
fcis-14812	184	24	image	image	NOUN
fcis-14812	184	25	itself	itself	PRON
fcis-14812	184	26	,	,	PUNCT
fcis-14812	184	27	while	while	SCONJ
fcis-14812	184	28	the	the	DET
fcis-14812	184	29	role	role	NOUN
fcis-14812	184	30	of	of	ADP
fcis-14812	184	31	the	the	DET
fcis-14812	184	32	sfem	sfem	ADJ
fcis-14812	184	33	feature	feature	NOUN
fcis-14812	184	34	enhancement	enhancement	NOUN
fcis-14812	184	35	module	module	NOUN
fcis-14812	184	36	is	be	AUX
fcis-14812	184	37	to	to	PART
fcis-14812	184	38	further	far	ADV
fcis-14812	184	39	enhance	enhance	VERB
fcis-14812	184	40	the	the	DET
fcis-14812	184	41	feature	feature	NOUN
fcis-14812	184	42	map	map	NOUN
fcis-14812	184	43	of	of	ADP
fcis-14812	184	44	the	the	DET
fcis-14812	184	45	query	query	NOUN
fcis-14812	184	46	image	image	NOUN
fcis-14812	184	47	by	by	ADP
fcis-14812	184	48	utilizing	utilize	VERB
fcis-14812	184	49	the	the	DET
fcis-14812	184	50	information	information	NOUN
fcis-14812	184	51	between	between	ADP
fcis-14812	184	52	the	the	DET
fcis-14812	184	53	feature	feature	NOUN
fcis-14812	184	54	maps	map	NOUN
fcis-14812	184	55	of	of	ADP
fcis-14812	184	56	the	the	DET
fcis-14812	184	57	query	query	NOUN
fcis-14812	184	58	images	image	NOUN
fcis-14812	184	59	at	at	ADP
fcis-14812	184	60	different	different	ADJ
fcis-14812	184	61	levels	level	NOUN
fcis-14812	184	62	.	.	PUNCT
fcis-14812	185	1	specifically	specifically	ADV
fcis-14812	185	2	,	,	PUNCT
fcis-14812	185	3	we	we	PRON
fcis-14812	185	4	send	send	VERB
fcis-14812	185	5	the	the	DET
fcis-14812	185	6	query	query	NOUN
fcis-14812	185	7	image	image	NOUN
fcis-14812	185	8	feature	feature	NOUN
fcis-14812	185	9	maps	map	NOUN
fcis-14812	185	10	and	and	CCONJ
fcis-14812	185	11	sample	sample	NOUN
fcis-14812	185	12	image	image	NOUN
fcis-14812	185	13	feature	feature	NOUN
fcis-14812	185	14	vectors	vector	NOUN
fcis-14812	185	15	output	output	VERB
fcis-14812	185	16	from	from	ADP
fcis-14812	185	17	the	the	DET
fcis-14812	185	18	same	same	ADJ
fcis-14812	185	19	layer	layer	NOUN
fcis-14812	185	20	of	of	ADP
fcis-14812	185	21	the	the	DET
fcis-14812	185	22	backbone	backbone	NOUN
fcis-14812	185	23	network	network	NOUN
fcis-14812	185	24	to	to	ADP
fcis-14812	185	25	the	the	DET
fcis-14812	185	26	self	self	NOUN
fcis-14812	185	27	-	-	PUNCT
fcis-14812	185	28	similarity	similarity	NOUN
fcis-14812	185	29	module	module	NOUN
fcis-14812	185	30	for	for	ADP
fcis-14812	185	31	the	the	DET
fcis-14812	185	32	first	first	ADJ
fcis-14812	185	33	feature	feature	NOUN
fcis-14812	185	34	enhancement	enhancement	NOUN
fcis-14812	185	35	to	to	PART
fcis-14812	185	36	get	get	VERB
fcis-14812	185	37	the	the	DET
fcis-14812	185	38	enhanced	enhanced	ADJ
fcis-14812	185	39	query	query	NOUN
fcis-14812	185	40	image	image	NOUN
fcis-14812	185	41	feature	feature	NOUN
fcis-14812	185	42	maps	map	NOUN
fcis-14812	185	43	;	;	PUNCT
fcis-14812	185	44	then	then	ADV
fcis-14812	185	45	we	we	PRON
fcis-14812	185	46	send	send	VERB
fcis-14812	185	47	the	the	DET
fcis-14812	185	48	query	query	NOUN
fcis-14812	185	49	image	image	NOUN
fcis-14812	185	50	feature	feature	NOUN
fcis-14812	185	51	maps	map	NOUN
fcis-14812	185	52	from	from	ADP
fcis-14812	185	53	different	different	ADJ
fcis-14812	185	54	layers	layer	NOUN
fcis-14812	185	55	processed	process	VERB
fcis-14812	185	56	by	by	ADP
fcis-14812	185	57	the	the	DET
fcis-14812	185	58	self	self	NOUN
fcis-14812	185	59	-	-	PUNCT
fcis-14812	185	60	similarity	similarity	NOUN
fcis-14812	185	61	module	module	NOUN
fcis-14812	185	62	to	to	ADP
fcis-14812	185	63	the	the	DET
fcis-14812	185	64	sfem	sfem	ADJ
fcis-14812	185	65	feature	feature	NOUN
fcis-14812	185	66	enhancement	enhancement	NOUN
fcis-14812	185	67	module	module	NOUN
fcis-14812	185	68	for	for	ADP
fcis-14812	185	69	the	the	DET
fcis-14812	185	70	second	second	ADJ
fcis-14812	185	71	feature	feature	NOUN
fcis-14812	185	72	enhancement	enhancement	NOUN
fcis-14812	185	73	to	to	PART
fcis-14812	185	74	get	get	VERB
fcis-14812	185	75	the	the	DET
fcis-14812	185	76	final	final	ADJ
fcis-14812	185	77	query	query	NOUN
fcis-14812	185	78	image	image	NOUN
fcis-14812	185	79	feature	feature	NOUN
fcis-14812	185	80	maps	map	NOUN
fcis-14812	185	81	.	.	PUNCT
fcis-14812	186	1	the	the	DET
fcis-14812	186	2	sfem	sfem	ADJ
fcis-14812	186	3	feature	feature	NOUN
fcis-14812	186	4	enhancement	enhancement	NOUN
fcis-14812	186	5	module	module	NOUN
fcis-14812	186	6	is	be	AUX
fcis-14812	186	7	based	base	VERB
fcis-14812	186	8	on	on	ADP
fcis-14812	186	9	the	the	DET
fcis-14812	186	10	conditional	conditional	ADJ
fcis-14812	186	11	random	random	ADJ
fcis-14812	186	12	field	field	NOUN
fcis-14812	186	13	(	(	PUNCT
fcis-14812	186	14	crf	crf	NOUN
fcis-14812	186	15	)	)	PUNCT
fcis-14812	186	16	method	method	NOUN
fcis-14812	186	17	,	,	PUNCT
fcis-14812	186	18	which	which	PRON
fcis-14812	186	19	can	can	AUX
fcis-14812	186	20	effectively	effectively	ADV
fcis-14812	186	21	utilize	utilize	VERB
fcis-14812	186	22	the	the	DET
fcis-14812	186	23	correlation	correlation	NOUN
fcis-14812	186	24	between	between	ADP
fcis-14812	186	25	spatially	spatially	ADV
fcis-14812	186	26	adjacent	adjacent	ADJ
fcis-14812	186	27	pixels	pixel	NOUN
fcis-14812	186	28	to	to	PART
fcis-14812	186	29	optimize	optimize	VERB
fcis-14812	186	30	the	the	DET
fcis-14812	186	31	feature	feature	NOUN
fcis-14812	186	32	value	value	NOUN
fcis-14812	186	33	of	of	ADP
fcis-14812	186	34	each	each	DET
fcis-14812	186	35	pixel	pixel	PROPN
fcis-14812	186	36	point	point	NOUN
fcis-14812	186	37	.	.	PUNCT
fcis-14812	187	1	after	after	ADP
fcis-14812	187	2	obtaining	obtain	VERB
fcis-14812	187	3	the	the	DET
fcis-14812	187	4	feature	feature	NOUN
fcis-14812	187	5	representations	representation	NOUN
fcis-14812	187	6	of	of	ADP
fcis-14812	187	7	the	the	DET
fcis-14812	187	8	final	final	ADJ
fcis-14812	187	9	query	query	NOUN
fcis-14812	187	10	image	image	NOUN
fcis-14812	187	11	and	and	CCONJ
fcis-14812	187	12	the	the	DET
fcis-14812	187	13	sample	sample	NOUN
fcis-14812	187	14	image	image	NOUN
fcis-14812	187	15	,	,	PUNCT
fcis-14812	187	16	we	we	PRON
fcis-14812	187	17	need	need	VERB
fcis-14812	187	18	to	to	PART
fcis-14812	187	19	fuse	fuse	VERB
fcis-14812	187	20	and	and	CCONJ
fcis-14812	187	21	compare	compare	VERB
fcis-14812	187	22	them	they	PRON
fcis-14812	187	23	to	to	PART
fcis-14812	187	24	obtain	obtain	VERB
fcis-14812	187	25	a	a	DET
fcis-14812	187	26	similarity	similarity	NOUN
fcis-14812	187	27	measure	measure	NOUN
fcis-14812	187	28	between	between	ADP
fcis-14812	187	29	them	they	PRON
fcis-14812	187	30	.	.	PUNCT
fcis-14812	188	1	for	for	ADP
fcis-14812	188	2	this	this	DET
fcis-14812	188	3	purpose	purpose	NOUN
fcis-14812	188	4	,	,	PUNCT
fcis-14812	188	5	we	we	PRON
fcis-14812	188	6	use	use	VERB
fcis-14812	188	7	the	the	DET
fcis-14812	188	8	asff	asff	NOUN
fcis-14812	188	9	adaptively	adaptively	ADV
fcis-14812	188	10	spatial	spatial	ADJ
fcis-14812	188	11	fusion	fusion	NOUN
fcis-14812	188	12	module	module	NOUN
fcis-14812	188	13	and	and	CCONJ
fcis-14812	188	14	dynamic	dynamic	ADJ
fcis-14812	188	15	similarity	similarity	NOUN
fcis-14812	188	16	metric	metric	ADJ
fcis-14812	188	17	module	module	NOUN
fcis-14812	188	18	,	,	PUNCT
fcis-14812	188	19	the	the	DET
fcis-14812	188	20	role	role	NOUN
fcis-14812	188	21	of	of	ADP
fcis-14812	188	22	the	the	DET
fcis-14812	188	23	asff	asff	NOUN
fcis-14812	188	24	adaptive	adaptive	PROPN
fcis-14812	188	25	spatial	spatial	ADJ
fcis-14812	188	26	fusion	fusion	NOUN
fcis-14812	188	27	module	module	NOUN
fcis-14812	188	28	is	be	AUX
fcis-14812	188	29	to	to	PART
fcis-14812	188	30	dynamically	dynamically	ADV
fcis-14812	188	31	fuse	fuse	VERB
fcis-14812	188	32	the	the	DET
fcis-14812	188	33	query	query	NOUN
fcis-14812	188	34	and	and	CCONJ
fcis-14812	188	35	sample	sample	NOUN
fcis-14812	188	36	images	image	NOUN
fcis-14812	188	37	at	at	ADP
fcis-14812	188	38	different	different	ADJ
fcis-14812	188	39	levels	level	NOUN
fcis-14812	188	40	based	base	VERB
fcis-14812	188	41	on	on	ADP
fcis-14812	188	42	the	the	DET
fcis-14812	188	43	correlation	correlation	NOUN
fcis-14812	188	44	between	between	ADP
fcis-14812	188	45	them	they	PRON
fcis-14812	188	46	.	.	PUNCT
fcis-14812	189	1	the	the	DET
fcis-14812	189	2	function	function	NOUN
fcis-14812	189	3	of	of	ADP
fcis-14812	189	4	asff	asff	PROPN
fcis-14812	189	5	adaptive	adaptive	PROPN
fcis-14812	189	6	spatial	spatial	ADJ
fcis-14812	189	7	fusion	fusion	NOUN
fcis-14812	189	8	module	module	NOUN
fcis-14812	189	9	is	be	AUX
fcis-14812	189	10	to	to	PART
fcis-14812	189	11	dynamically	dynamically	ADV
fcis-14812	189	12	fuse	fuse	VERB
fcis-14812	189	13	the	the	DET
fcis-14812	189	14	features	feature	NOUN
fcis-14812	189	15	of	of	ADP
fcis-14812	189	16	query	query	NOUN
fcis-14812	189	17	images	image	NOUN
fcis-14812	189	18	and	and	CCONJ
fcis-14812	189	19	sample	sample	NOUN
fcis-14812	189	20	images	image	NOUN
fcis-14812	189	21	at	at	ADP
fcis-14812	189	22	different	different	ADJ
fcis-14812	189	23	levels	level	NOUN
fcis-14812	189	24	according	accord	VERB
fcis-14812	189	25	to	to	ADP
fcis-14812	189	26	their	their	PRON
fcis-14812	189	27	correlations	correlation	NOUN
fcis-14812	189	28	,	,	PUNCT
fcis-14812	189	29	so	so	SCONJ
fcis-14812	189	30	as	as	SCONJ
fcis-14812	189	31	to	to	PART
fcis-14812	189	32	obtain	obtain	VERB
fcis-14812	189	33	a	a	DET
fcis-14812	189	34	more	more	ADV
fcis-14812	189	35	comprehensive	comprehensive	ADJ
fcis-14812	189	36	and	and	CCONJ
fcis-14812	189	37	accurate	accurate	ADJ
fcis-14812	189	38	fusion	fusion	NOUN
fcis-14812	189	39	features	feature	NOUN
fcis-14812	189	40	.	.	PUNCT
fcis-14812	190	1	the	the	DET
fcis-14812	190	2	role	role	NOUN
fcis-14812	190	3	of	of	ADP
fcis-14812	190	4	the	the	DET
fcis-14812	190	5	dynamic	dynamic	ADJ
fcis-14812	190	6	similarity	similarity	NOUN
fcis-14812	190	7	metric	metric	ADJ
fcis-14812	190	8	module	module	NOUN
fcis-14812	190	9	is	be	AUX
fcis-14812	190	10	to	to	PART
fcis-14812	190	11	calculate	calculate	VERB
fcis-14812	190	12	the	the	DET
fcis-14812	190	13	similarity	similarity	NOUN
fcis-14812	190	14	between	between	ADP
fcis-14812	190	15	each	each	DET
fcis-14812	190	16	pixel	pixel	NOUN
fcis-14812	190	17	and	and	CCONJ
fcis-14812	190	18	the	the	DET
fcis-14812	190	19	sample	sample	NOUN
fcis-14812	190	20	image	image	NOUN
fcis-14812	190	21	according	accord	VERB
fcis-14812	190	22	to	to	ADP
fcis-14812	190	23	the	the	DET
fcis-14812	190	24	fused	fuse	VERB
fcis-14812	190	25	feature	feature	NOUN
fcis-14812	190	26	map	map	NOUN
fcis-14812	190	27	,	,	PUNCT
fcis-14812	190	28	and	and	CCONJ
fcis-14812	190	29	dynamically	dynamically	ADV
fcis-14812	190	30	adjust	adjust	VERB
fcis-14812	190	31	the	the	DET
fcis-14812	190	32	similarity	similarity	NOUN
fcis-14812	190	33	threshold	threshold	NOUN
fcis-14812	190	34	according	accord	VERB
fcis-14812	190	35	to	to	ADP
fcis-14812	190	36	the	the	DET
fcis-14812	190	37	distribution	distribution	NOUN
fcis-14812	190	38	of	of	ADP
fcis-14812	190	39	similarity	similarity	NOUN
fcis-14812	190	40	,	,	PUNCT
fcis-14812	190	41	so	so	SCONJ
fcis-14812	190	42	as	as	SCONJ
fcis-14812	190	43	to	to	PART
fcis-14812	190	44	obtain	obtain	VERB
fcis-14812	190	45	a	a	DET
fcis-14812	190	46	more	more	ADV
fcis-14812	190	47	stable	stable	ADJ
fcis-14812	190	48	and	and	CCONJ
fcis-14812	190	49	robust	robust	ADJ
fcis-14812	190	50	similarity	similarity	NOUN
fcis-14812	190	51	metric	metric	NOUN
fcis-14812	190	52	.	.	PUNCT
fcis-14812	191	1	finally	finally	ADV
fcis-14812	191	2	,	,	PUNCT
fcis-14812	191	3	we	we	PRON
fcis-14812	191	4	feed	feed	VERB
fcis-14812	191	5	the	the	DET
fcis-14812	191	6	output	output	NOUN
fcis-14812	191	7	of	of	ADP
fcis-14812	191	8	the	the	DET
fcis-14812	191	9	dynamic	dynamic	ADJ
fcis-14812	191	10	similarity	similarity	NOUN
fcis-14812	191	11	metric	metric	ADJ
fcis-14812	191	12	module	module	NOUN
fcis-14812	191	13	into	into	ADP
fcis-14812	191	14	a	a	DET
fcis-14812	191	15	regression	regression	NOUN
fcis-14812	191	16	layer	layer	NOUN
fcis-14812	191	17	to	to	PART
fcis-14812	191	18	get	get	VERB
fcis-14812	191	19	the	the	DET
fcis-14812	191	20	final	final	ADJ
fcis-14812	191	21	counting	counting	NOUN
fcis-14812	191	22	result	result	NOUN
fcis-14812	191	23	.	.	PUNCT
fcis-14812	192	1	loss	loss	NOUN
fcis-14812	192	2	function	function	NOUN
fcis-14812	192	3	.	.	PUNCT
fcis-14812	193	1	the	the	DET
fcis-14812	193	2	loss	loss	NOUN
fcis-14812	193	3	function	function	NOUN
fcis-14812	193	4	in	in	ADP
fcis-14812	193	5	this	this	DET
fcis-14812	193	6	work	work	NOUN
fcis-14812	193	7	consists	consist	VERB
fcis-14812	193	8	of	of	ADP
fcis-14812	193	9	two	two	NUM
fcis-14812	193	10	parts	part	NOUN
fcis-14812	193	11	,	,	PUNCT
fcis-14812	193	12	namely	namely	ADV
fcis-14812	193	13	similarity	similarity	NOUN
fcis-14812	193	14	loss	loss	NOUN
fcis-14812	193	15	and	and	CCONJ
fcis-14812	193	16	count	count	NOUN
fcis-14812	193	17	loss	loss	NOUN
fcis-14812	193	18	.	.	PUNCT
fcis-14812	194	1	we	we	PRON
fcis-14812	194	2	will	will	AUX
fcis-14812	194	3	assume	assume	VERB
fcis-14812	194	4	that	that	SCONJ
fcis-14812	194	5	the	the	DET
fcis-14812	194	6	size	size	NOUN
fcis-14812	194	7	of	of	ADP
fcis-14812	194	8	s	s	PROPN
fcis-14812	194	9	is	be	AUX
fcis-14812	194	10	1	1	NUM
fcis-14812	194	11	/	/	SYM
fcis-14812	194	12	r	r	NOUN
fcis-14812	194	13	of	of	ADP
fcis-14812	194	14	x	x	PRON
fcis-14812	194	15	,	,	PUNCT
fcis-14812	194	16	i.e.	i.e.	X
fcis-14812	194	17	,	,	PUNCT
fcis-14812	194	18	each	each	DET
fcis-14812	194	19	position	position	NOUN
fcis-14812	194	20	in	in	ADP
fcis-14812	194	21	the	the	DET
fcis-14812	194	22	similarity	similarity	NOUN
fcis-14812	194	23	map	map	NOUN
fcis-14812	194	24	corresponds	correspond	VERB
fcis-14812	194	25	to	to	ADP
fcis-14812	194	26	an	an	DET
fcis-14812	194	27	r×r	r×r	ADJ
fcis-14812	194	28	block	block	NOUN
fcis-14812	194	29	in	in	ADP
fcis-14812	194	30	the	the	DET
fcis-14812	194	31	query	query	NOUN
fcis-14812	194	32	image	image	NOUN
fcis-14812	194	33	.	.	PUNCT
fcis-14812	195	1	for	for	ADP
fcis-14812	195	2	each	each	DET
fcis-14812	195	3	location	location	NOUN
fcis-14812	195	4	in	in	ADP
fcis-14812	195	5	s	s	PROPN
fcis-14812	195	6	,	,	PUNCT
fcis-14812	195	7	a	a	DET
fcis-14812	195	8	positive	positive	ADJ
fcis-14812	195	9	label	label	NOUN
fcis-14812	195	10	is	be	AUX
fcis-14812	195	11	assigned	assign	VERB
fcis-14812	195	12	if	if	SCONJ
fcis-14812	195	13	its	its	PRON
fcis-14812	195	14	corresponding	corresponding	ADJ
fcis-14812	195	15	r×r	r×r	NOUN
fcis-14812	195	16	block	block	NOUN
fcis-14812	195	17	contains	contain	VERB
fcis-14812	195	18	multiple	multiple	ADJ
fcis-14812	195	19	targets	target	NOUN
fcis-14812	195	20	,	,	PUNCT
fcis-14812	195	21	and	and	CCONJ
fcis-14812	195	22	a	a	DET
fcis-14812	195	23	negative	negative	ADJ
fcis-14812	195	24	label	label	NOUN
fcis-14812	195	25	is	be	AUX
fcis-14812	195	26	assigned	assign	VERB
fcis-14812	195	27	if	if	SCONJ
fcis-14812	195	28	it	it	PRON
fcis-14812	195	29	does	do	AUX
fcis-14812	195	30	not	not	PART
fcis-14812	195	31	contain	contain	VERB
fcis-14812	195	32	targets	target	NOUN
fcis-14812	195	33	.	.	PUNCT
fcis-14812	196	1	then	then	ADV
fcis-14812	196	2	,	,	PUNCT
fcis-14812	196	3	we	we	PRON
fcis-14812	196	4	can	can	AUX
fcis-14812	196	5	derive	derive	VERB
fcis-14812	196	6	the	the	DET
fcis-14812	196	7	similarity	similarity	NOUN
fcis-14812	196	8	loss	loss	NOUN
fcis-14812	196	9	based	base	VERB
fcis-14812	196	10	on	on	ADP
fcis-14812	196	11	the	the	DET
fcis-14812	196	12	signal	signal	NOUN
fcis-14812	196	13	-	-	PUNCT
fcis-14812	196	14	to	to	ADP
fcis-14812	196	15	-	-	PUNCT
fcis-14812	196	16	noise	noise	NOUN
fcis-14812	196	17	ratio	ratio	NOUN
fcis-14812	196	18	:	:	PUNCT
fcis-14812	196	19			NOUN
fcis-14812	196	20			SYM
fcis-14812	196	21			NOUN
fcis-14812	196	22			SYM
fcis-14812	196	23			NOUN
fcis-14812	196	24	jnegiiposi	jnegiiposi	DET
fcis-14812	196	25	iposi	iposi	NOUN
fcis-14812	196	26	sim	sim	VERB
fcis-14812	196	27	ss	ss	NOUN
fcis-14812	196	28	s	s	PROPN
fcis-14812	196	29	l	l	NOUN
fcis-14812	196	30	expexp	expexp	NOUN
fcis-14812	196	31	exp	exp	NOUN
fcis-14812	196	32	log	log	NOUN
fcis-14812	196	33			X
fcis-14812	196	34			PROPN
fcis-14812	196	35			ADV
fcis-14812	196	36			X
fcis-14812	196	37			PROPN
fcis-14812	196	38	(	(	PUNCT
fcis-14812	196	39	11	11	NUM
fcis-14812	196	40	)	)	PUNCT
fcis-14812	196	41	where	where	SCONJ
fcis-14812	196	42	pos	pos	NOUN
fcis-14812	196	43	and	and	CCONJ
fcis-14812	196	44	neg	neg	NOUN
fcis-14812	196	45	denote	denote	VERB
fcis-14812	196	46	the	the	DET
fcis-14812	196	47	positive	positive	ADJ
fcis-14812	196	48	labeling	labeling	NOUN
fcis-14812	196	49	position	position	NOUN
fcis-14812	196	50	and	and	CCONJ
fcis-14812	196	51	negative	negative	ADJ
fcis-14812	196	52	labeling	labeling	NOUN
fcis-14812	196	53	position	position	NOUN
fcis-14812	196	54	in	in	ADP
fcis-14812	196	55	s.	s.	PROPN
fcis-14812	196	56	regarding	regard	VERB
fcis-14812	196	57	the	the	DET
fcis-14812	196	58	count	count	NOUN
fcis-14812	196	59	loss	loss	NOUN
fcis-14812	196	60	,	,	PUNCT
fcis-14812	196	61	we	we	PRON
fcis-14812	196	62	choose	choose	VERB
fcis-14812	196	63	to	to	PART
fcis-14812	196	64	use	use	VERB
fcis-14812	196	65	the	the	DET
fcis-14812	196	66	traditional	traditional	ADJ
fcis-14812	196	67	l2	l2	NOUN
fcis-14812	196	68	loss	loss	NOUN
fcis-14812	196	69	as	as	ADP
fcis-14812	196	70	our	our	PRON
fcis-14812	196	71	count	count	NOUN
fcis-14812	196	72	loss	loss	NOUN
fcis-14812	196	73	function	function	NOUN
fcis-14812	196	74	:	:	PUNCT
fcis-14812	196	75			NOUN
fcis-14812	197	1			SYM
fcis-14812	197	2			NOUN
fcis-14812	197	3			PROPN
fcis-14812	197	4	2	2	NUM
fcis-14812	197	5	2	2	NUM
fcis-14812	197	6	,	,	PUNCT
fcis-14812	197	7	,	,	PUNCT
fcis-14812	197	8	zxdzxdl	zxdzxdl	PROPN
fcis-14812	197	9	gtprcount	gtprcount	NOUN
fcis-14812	197	10			PROPN
fcis-14812	197	11	(	(	PUNCT
fcis-14812	197	12	12	12	NUM
fcis-14812	197	13	)	)	PUNCT
fcis-14812	197	14	where	where	SCONJ
fcis-14812	197	15	gtd	gtd	PROPN
fcis-14812	197	16	denotes	denote	VERB
fcis-14812	197	17	the	the	DET
fcis-14812	197	18	ground	ground	NOUN
fcis-14812	197	19	truth	truth	NOUN
fcis-14812	197	20	density	density	NOUN
fcis-14812	197	21	map	map	NOUN
fcis-14812	197	22	.	.	PUNCT
fcis-14812	198	1	thus	thus	ADV
fcis-14812	198	2	,	,	PUNCT
fcis-14812	198	3	our	our	PRON
fcis-14812	198	4	overall	overall	ADJ
fcis-14812	198	5	loss	loss	NOUN
fcis-14812	198	6	function	function	NOUN
fcis-14812	198	7	is	be	AUX
fcis-14812	198	8	:	:	PUNCT
fcis-14812	198	9			NOUN
fcis-14812	198	10			SYM
fcis-14812	198	11			PROPN
fcis-14812	198	12	slddll	slddll	PROPN
fcis-14812	198	13	simgtprcount	simgtprcount	NOUN
fcis-14812	198	14			PROPN
fcis-14812	198	15			NOUN
fcis-14812	198	16	,	,	PUNCT
fcis-14812	198	17	(	(	PUNCT
fcis-14812	198	18	13	13	NUM
fcis-14812	198	19	)	)	PUNCT
fcis-14812	198	20	where	where	SCONJ
fcis-14812	198	21	is	be	AUX
fcis-14812	198	22	used	use	VERB
fcis-14812	198	23	to	to	PART
fcis-14812	198	24	balance	balance	VERB
fcis-14812	198	25	the	the	DET
fcis-14812	198	26	values	value	NOUN
fcis-14812	198	27	of	of	ADP
fcis-14812	198	28	the	the	DET
fcis-14812	198	29	two	two	NUM
fcis-14812	198	30	loss	loss	NOUN
fcis-14812	198	31	functions	function	NOUN
fcis-14812	198	32	.	.	PUNCT
fcis-14812	199	1	4	4	X
fcis-14812	199	2	.	.	X
fcis-14812	199	3	experimental	experimental	ADJ
fcis-14812	199	4	results	result	NOUN
fcis-14812	199	5	&	&	CCONJ
fcis-14812	199	6	analysis	analysis	NOUN
fcis-14812	199	7	4.1	4.1	NUM
fcis-14812	199	8	.	.	PUNCT
fcis-14812	199	9	evaluation	evaluation	NOUN
fcis-14812	199	10	metrics	metric	NOUN
fcis-14812	199	11	this	this	DET
fcis-14812	199	12	study	study	NOUN
fcis-14812	199	13	provides	provide	VERB
fcis-14812	199	14	an	an	DET
fcis-14812	199	15	in	in	ADP
fcis-14812	199	16	-	-	PUNCT
fcis-14812	199	17	depth	depth	NOUN
fcis-14812	199	18	exploration	exploration	NOUN
fcis-14812	199	19	and	and	CCONJ
fcis-14812	199	20	analysis	analysis	NOUN
fcis-14812	199	21	of	of	ADP
fcis-14812	199	22	the	the	DET
fcis-14812	199	23	target	target	NOUN
fcis-14812	199	24	counting	counting	NOUN
fcis-14812	199	25	problem	problem	NOUN
fcis-14812	199	26	based	base	VERB
fcis-14812	199	27	on	on	ADP
fcis-14812	199	28	the	the	DET
fcis-14812	199	29	fsc-147	fsc-147	NOUN
fcis-14812	199	30	dataset	dataset	NOUN
fcis-14812	199	31	.	.	PUNCT
fcis-14812	200	1	this	this	DET
fcis-14812	200	2	dataset	dataset	NOUN
fcis-14812	200	3	was	be	AUX
fcis-14812	200	4	constructed	construct	VERB
fcis-14812	200	5	by	by	ADP
fcis-14812	200	6	ranjan	ranjan	PROPN
fcis-14812	200	7	et	et	PROPN
fcis-14812	200	8	al	al	PROPN
fcis-14812	201	1	[	[	X
fcis-14812	201	2	14	14	NUM
fcis-14812	201	3	]	]	PUNCT
fcis-14812	202	1	and	and	CCONJ
fcis-14812	202	2	contains	contain	VERB
fcis-14812	202	3	high	high	ADJ
fcis-14812	202	4	-	-	PUNCT
fcis-14812	202	5	resolution	resolution	NOUN
fcis-14812	202	6	images	image	NOUN
fcis-14812	202	7	of	of	ADP
fcis-14812	202	8	147	147	NUM
fcis-14812	202	9	different	different	ADJ
fcis-14812	202	10	scenes	scene	NOUN
fcis-14812	202	11	covering	cover	VERB
fcis-14812	202	12	a	a	DET
fcis-14812	202	13	wide	wide	ADJ
fcis-14812	202	14	range	range	NOUN
fcis-14812	202	15	of	of	ADP
fcis-14812	202	16	densities	density	NOUN
fcis-14812	202	17	,	,	PUNCT
fcis-14812	202	18	illumination	illumination	NOUN
fcis-14812	202	19	,	,	PUNCT
fcis-14812	202	20	occlusion	occlusion	NOUN
fcis-14812	202	21	and	and	CCONJ
fcis-14812	202	22	background	background	NOUN
fcis-14812	202	23	complexity	complexity	NOUN
fcis-14812	202	24	.	.	PUNCT
fcis-14812	203	1	in	in	ADP
fcis-14812	203	2	order	order	NOUN
fcis-14812	203	3	to	to	PART
fcis-14812	203	4	objectively	objectively	ADV
fcis-14812	203	5	assess	assess	VERB
fcis-14812	203	6	the	the	DET
fcis-14812	203	7	performance	performance	NOUN
fcis-14812	203	8	of	of	ADP
fcis-14812	203	9	the	the	DET
fcis-14812	203	10	method	method	NOUN
fcis-14812	203	11	proposed	propose	VERB
fcis-14812	203	12	in	in	ADP
fcis-14812	203	13	this	this	DET
fcis-14812	203	14	study	study	NOUN
fcis-14812	203	15	,	,	PUNCT
fcis-14812	203	16	two	two	NUM
fcis-14812	203	17	widely	widely	ADV
fcis-14812	203	18	used	use	VERB
fcis-14812	203	19	evaluation	evaluation	NOUN
fcis-14812	203	20	metrics	metric	NOUN
fcis-14812	203	21	are	be	AUX
fcis-14812	203	22	used	use	VERB
fcis-14812	203	23	,	,	PUNCT
fcis-14812	203	24	namely	namely	ADV
fcis-14812	203	25	mean	mean	VERB
fcis-14812	203	26	absolute	absolute	ADJ
fcis-14812	203	27	error	error	NOUN
fcis-14812	203	28	(	(	PUNCT
fcis-14812	203	29	mae	mae	PROPN
fcis-14812	203	30	)	)	PUNCT
fcis-14812	203	31	and	and	CCONJ
fcis-14812	203	32	mean	mean	VERB
fcis-14812	203	33	square	square	ADJ
fcis-14812	203	34	error	error	NOUN
fcis-14812	203	35	(	(	PUNCT
fcis-14812	203	36	mse	mse	NOUN
fcis-14812	203	37	)	)	PUNCT
fcis-14812	203	38	.	.	PUNCT
fcis-14812	204	1	mae	mae	PROPN
fcis-14812	204	2	measures	measure	VERB
fcis-14812	204	3	the	the	DET
fcis-14812	204	4	average	average	ADJ
fcis-14812	204	5	deviation	deviation	NOUN
fcis-14812	204	6	between	between	ADP
fcis-14812	204	7	the	the	DET
fcis-14812	204	8	predicted	predict	VERB
fcis-14812	204	9	value	value	NOUN
fcis-14812	204	10	and	and	CCONJ
fcis-14812	204	11	the	the	DET
fcis-14812	204	12	true	true	ADJ
fcis-14812	204	13	value	value	NOUN
fcis-14812	204	14	,	,	PUNCT
fcis-14812	204	15	and	and	CCONJ
fcis-14812	204	16	mse	mse	NOUN
fcis-14812	204	17	measures	measure	NOUN
fcis-14812	204	18	the	the	DET
fcis-14812	204	19	average	average	ADJ
fcis-14812	204	20	squared	squared	ADJ
fcis-14812	204	21	difference	difference	NOUN
fcis-14812	204	22	between	between	ADP
fcis-14812	204	23	the	the	DET
fcis-14812	204	24	predicted	predict	VERB
fcis-14812	204	25	value	value	NOUN
fcis-14812	204	26	and	and	CCONJ
fcis-14812	204	27	the	the	DET
fcis-14812	204	28	true	true	ADJ
fcis-14812	204	29	value	value	NOUN
fcis-14812	204	30	.	.	PUNCT
fcis-14812	205	1	these	these	DET
fcis-14812	205	2	two	two	NUM
fcis-14812	205	3	metrics	metric	NOUN
fcis-14812	205	4	reflect	reflect	VERB
fcis-14812	205	5	the	the	DET
fcis-14812	205	6	accuracy	accuracy	NOUN
fcis-14812	205	7	and	and	CCONJ
fcis-14812	205	8	stability	stability	NOUN
fcis-14812	205	9	of	of	ADP
fcis-14812	205	10	the	the	DET
fcis-14812	205	11	target	target	NOUN
fcis-14812	205	12	counting	counting	NOUN
fcis-14812	205	13	method	method	NOUN
fcis-14812	205	14	.	.	PUNCT
fcis-14812	206	1	||	||	NOUN
fcis-14812	207	1	1	1	NUM
fcis-14812	207	2	mae	mae	PROPN
fcis-14812	207	3	^	^	SYM
fcis-14812	207	4	1	1	NUM
fcis-14812	207	5			NUM
fcis-14812	207	6			NOUN
fcis-14812	207	7			PRON
fcis-14812	207	8			PROPN
fcis-14812	207	9			PROPN
fcis-14812	207	10			NOUN
fcis-14812	207	11			NOUN
fcis-14812	207	12			PROPN
fcis-14812	208	1	ii	ii	VERB
fcis-14812	208	2	m	m	VERB
fcis-14812	209	1	i	i	INTJ
fcis-14812	209	2	yy	yy	INTJ
fcis-14812	209	3	m	m	VERB
fcis-14812	209	4	(	(	PUNCT
fcis-14812	209	5	14	14	NUM
fcis-14812	209	6	)	)	PUNCT
fcis-14812	209	7	2^	2^	NUM
fcis-14812	209	8	1	1	NUM
fcis-14812	209	9	1	1	NUM
fcis-14812	209	10	mse	mse	NOUN
fcis-14812	209	11			PROPN
fcis-14812	209	12			NOUN
fcis-14812	209	13			PRON
fcis-14812	209	14			PROPN
fcis-14812	209	15			PROPN
fcis-14812	209	16			NOUN
fcis-14812	209	17			NOUN
fcis-14812	210	1			PROPN
fcis-14812	210	2	ii	ii	VERB
fcis-14812	210	3	m	m	VERB
fcis-14812	211	1	i	i	INTJ
fcis-14812	211	2	yy	yy	INTJ
fcis-14812	211	3	m	m	VERB
fcis-14812	211	4	(	(	PUNCT
fcis-14812	211	5	15	15	NUM
fcis-14812	211	6	)	)	PUNCT
fcis-14812	211	7	4.2	4.2	NUM
fcis-14812	211	8	.	.	PUNCT
fcis-14812	212	1	experiment	experiment	NOUN
fcis-14812	212	2	results	result	NOUN
fcis-14812	212	3	table	table	NOUN
fcis-14812	212	4	1	1	NUM
fcis-14812	212	5	.	.	PUNCT
fcis-14812	212	6	experimental	experimental	ADJ
fcis-14812	212	7	results	result	NOUN
fcis-14812	212	8	and	and	CCONJ
fcis-14812	212	9	comparison	comparison	NOUN
fcis-14812	212	10	of	of	ADP
fcis-14812	212	11	fsc-147	fsc-147	NOUN
fcis-14812	212	12	dataset	dataset	NOUN
fcis-14812	212	13	methods	method	NOUN
fcis-14812	212	14	val	val	NOUN
fcis-14812	212	15	mae	mae	PROPN
fcis-14812	212	16	val	val	PROPN
fcis-14812	212	17	mse	mse	PROPN
fcis-14812	212	18	test	test	PROPN
fcis-14812	212	19	mae	mae	PROPN
fcis-14812	212	20	test	test	NOUN
fcis-14812	212	21	mse	mse	NOUN
fcis-14812	212	22	gmn[4	gmn[4	PROPN
fcis-14812	212	23	]	]	PUNCT
fcis-14812	212	24	29.66	29.66	NUM
fcis-14812	212	25	89.81	89.81	NUM
fcis-14812	212	26	26.52	26.52	NUM
fcis-14812	212	27	124.57	124.57	NUM
fcis-14812	212	28	famnet[4	famnet[4	NOUN
fcis-14812	212	29	]	]	X
fcis-14812	212	30	24.32	24.32	NUM
fcis-14812	212	31	70.94	70.94	NUM
fcis-14812	212	32	22.56	22.56	NUM
fcis-14812	212	33	101.54	101.54	NUM
fcis-14812	212	34	famnet+[4	famnet+[4	PROPN
fcis-14812	212	35	]	]	X
fcis-14812	212	36	23.75	23.75	NUM
fcis-14812	212	37	69.07	69.07	NUM
fcis-14812	212	38	22.08	22.08	NUM
fcis-14812	212	39	99.54	99.54	NUM
fcis-14812	212	40	cfocnet[4	cfocnet[4	X
fcis-14812	212	41	]	]	X
fcis-14812	212	42	21.19	21.19	NUM
fcis-14812	212	43	61.41	61.41	NUM
fcis-14812	212	44	22.10	22.10	NUM
fcis-14812	212	45	112.71	112.71	NUM
fcis-14812	212	46	bmnet[4	bmnet[4	NOUN
fcis-14812	212	47	]	]	X
fcis-14812	212	48	19.06	19.06	NUM
fcis-14812	212	49	67.95	67.95	NUM
fcis-14812	212	50	16.71	16.71	NUM
fcis-14812	212	51	103.31	103.31	NUM
fcis-14812	212	52	bmnet+[4	bmnet+[4	SYM
fcis-14812	212	53	]	]	PUNCT
fcis-14812	212	54	15.74	15.74	NUM
fcis-14812	212	55	58.53	58.53	NUM
fcis-14812	212	56	14.62	14.62	NUM
fcis-14812	212	57	91.83	91.83	NUM
fcis-14812	212	58	defnet	defnet	NOUN
fcis-14812	212	59	(	(	PUNCT
fcis-14812	212	60	ours	ours	ADJ
fcis-14812	212	61	)	)	PUNCT
fcis-14812	212	62	15.03	15.03	NUM
fcis-14812	212	63	54.53	54.53	NUM
fcis-14812	212	64	13.65	13.65	NUM
fcis-14812	212	65	89.54	89.54	NUM
fcis-14812	212	66	table	table	NOUN
fcis-14812	212	67	1	1	NUM
fcis-14812	212	68	demonstrates	demonstrate	VERB
fcis-14812	212	69	the	the	DET
fcis-14812	212	70	experimental	experimental	ADJ
fcis-14812	212	71	results	result	NOUN
fcis-14812	212	72	and	and	CCONJ
fcis-14812	212	73	comparisons	comparison	NOUN
fcis-14812	212	74	of	of	ADP
fcis-14812	212	75	the	the	DET
fcis-14812	212	76	counting	counting	NOUN
fcis-14812	212	77	method	method	NOUN
fcis-14812	212	78	of	of	ADP
fcis-14812	212	79	this	this	DET
fcis-14812	212	80	work	work	NOUN
fcis-14812	212	81	on	on	ADP
fcis-14812	212	82	the	the	DET
fcis-14812	212	83	fsc	fsc	PROPN
fcis-14812	212	84	75	75	NUM
fcis-14812	212	85	147	147	NUM
fcis-14812	212	86	dataset	dataset	NOUN
fcis-14812	212	87	.	.	PUNCT
fcis-14812	213	1	according	accord	VERB
fcis-14812	213	2	to	to	ADP
fcis-14812	213	3	the	the	DET
fcis-14812	213	4	indexes	index	NOUN
fcis-14812	213	5	shown	show	VERB
fcis-14812	213	6	in	in	ADP
fcis-14812	213	7	the	the	DET
fcis-14812	213	8	following	follow	VERB
fcis-14812	213	9	table	table	NOUN
fcis-14812	213	10	,	,	PUNCT
fcis-14812	213	11	the	the	DET
fcis-14812	213	12	proposed	propose	VERB
fcis-14812	213	13	method	method	NOUN
fcis-14812	213	14	of	of	ADP
fcis-14812	213	15	this	this	DET
fcis-14812	213	16	paper	paper	NOUN
fcis-14812	213	17	is	be	AUX
fcis-14812	213	18	compared	compare	VERB
fcis-14812	213	19	with	with	ADP
fcis-14812	213	20	other	other	ADJ
fcis-14812	213	21	related	related	ADJ
fcis-14812	213	22	methods	method	NOUN
fcis-14812	213	23	in	in	ADP
fcis-14812	213	24	the	the	DET
fcis-14812	213	25	experiments	experiment	NOUN
fcis-14812	213	26	on	on	ADP
fcis-14812	213	27	the	the	DET
fcis-14812	213	28	fsc-147	fsc-147	NOUN
fcis-14812	213	29	dataset	dataset	NOUN
fcis-14812	213	30	,	,	PUNCT
fcis-14812	213	31	and	and	CCONJ
fcis-14812	213	32	the	the	DET
fcis-14812	213	33	results	result	NOUN
fcis-14812	213	34	show	show	VERB
fcis-14812	213	35	that	that	SCONJ
fcis-14812	213	36	the	the	DET
fcis-14812	213	37	method	method	NOUN
fcis-14812	213	38	of	of	ADP
fcis-14812	213	39	this	this	DET
fcis-14812	213	40	paper	paper	NOUN
fcis-14812	213	41	achieves	achieve	VERB
fcis-14812	213	42	the	the	DET
fcis-14812	213	43	highest	high	ADJ
fcis-14812	213	44	accuracy	accuracy	NOUN
fcis-14812	213	45	rate	rate	NOUN
fcis-14812	213	46	so	so	ADV
fcis-14812	213	47	far	far	ADV
fcis-14812	213	48	on	on	ADP
fcis-14812	213	49	this	this	DET
fcis-14812	213	50	dataset	dataset	NOUN
fcis-14812	213	51	,	,	PUNCT
fcis-14812	213	52	which	which	PRON
fcis-14812	213	53	proves	prove	VERB
fcis-14812	213	54	the	the	DET
fcis-14812	213	55	effectiveness	effectiveness	NOUN
fcis-14812	213	56	and	and	CCONJ
fcis-14812	213	57	superiority	superiority	NOUN
fcis-14812	213	58	of	of	ADP
fcis-14812	213	59	the	the	DET
fcis-14812	213	60	method	method	NOUN
fcis-14812	213	61	of	of	ADP
fcis-14812	213	62	this	this	DET
fcis-14812	213	63	paper	paper	NOUN
fcis-14812	213	64	.	.	PUNCT
fcis-14812	214	1	figure	figure	NOUN
fcis-14812	214	2	4	4	NUM
fcis-14812	214	3	demonstrates	demonstrate	VERB
fcis-14812	214	4	the	the	DET
fcis-14812	214	5	prediction	prediction	NOUN
fcis-14812	214	6	effect	effect	NOUN
fcis-14812	214	7	of	of	ADP
fcis-14812	214	8	this	this	DET
fcis-14812	214	9	paper	paper	NOUN
fcis-14812	214	10	's	's	PART
fcis-14812	214	11	method	method	NOUN
fcis-14812	214	12	on	on	ADP
fcis-14812	214	13	the	the	DET
fcis-14812	214	14	fsc-147	fsc-147	NOUN
fcis-14812	214	15	dataset	dataset	NOUN
fcis-14812	214	16	.	.	PUNCT
fcis-14812	215	1	it	it	PRON
fcis-14812	215	2	can	can	AUX
fcis-14812	215	3	be	be	AUX
fcis-14812	215	4	seen	see	VERB
fcis-14812	215	5	that	that	SCONJ
fcis-14812	215	6	the	the	DET
fcis-14812	215	7	method	method	NOUN
fcis-14812	215	8	of	of	ADP
fcis-14812	215	9	this	this	DET
fcis-14812	215	10	paper	paper	NOUN
fcis-14812	215	11	can	can	AUX
fcis-14812	215	12	effectively	effectively	ADV
fcis-14812	215	13	deal	deal	VERB
fcis-14812	215	14	with	with	ADP
fcis-14812	215	15	targets	target	NOUN
fcis-14812	215	16	of	of	ADP
fcis-14812	215	17	different	different	ADJ
fcis-14812	215	18	scales	scale	NOUN
fcis-14812	215	19	and	and	CCONJ
fcis-14812	215	20	predict	predict	VERB
fcis-14812	215	21	the	the	DET
fcis-14812	215	22	number	number	NOUN
fcis-14812	215	23	of	of	ADP
fcis-14812	215	24	targets	target	NOUN
fcis-14812	215	25	more	more	ADV
fcis-14812	215	26	accurately	accurately	ADV
fcis-14812	215	27	.	.	PUNCT
fcis-14812	216	1	figure	figure	VERB
fcis-14812	216	2	4	4	NUM
fcis-14812	216	3	.	.	PUNCT
fcis-14812	216	4	prediction	prediction	NOUN
fcis-14812	216	5	effect	effect	NOUN
fcis-14812	216	6	of	of	ADP
fcis-14812	216	7	the	the	DET
fcis-14812	216	8	proposed	propose	VERB
fcis-14812	216	9	method	method	NOUN
fcis-14812	216	10	on	on	ADP
fcis-14812	216	11	fsc-147	fsc-147	NOUN
fcis-14812	216	12	data	datum	NOUN
fcis-14812	216	13	set	set	VERB
fcis-14812	216	14	5	5	NUM
fcis-14812	216	15	.	.	PUNCT
fcis-14812	216	16	conclusion	conclusion	NOUN
fcis-14812	216	17	in	in	ADP
fcis-14812	216	18	this	this	DET
fcis-14812	216	19	paper	paper	NOUN
fcis-14812	216	20	,	,	PUNCT
fcis-14812	216	21	we	we	PRON
fcis-14812	216	22	propose	propose	VERB
fcis-14812	216	23	a	a	DET
fcis-14812	216	24	class	class	NOUN
fcis-14812	216	25	agnostic	agnostic	ADJ
fcis-14812	216	26	counting	counting	NOUN
fcis-14812	216	27	network	network	NOUN
fcis-14812	216	28	structure	structure	NOUN
fcis-14812	216	29	,	,	PUNCT
fcis-14812	216	30	which	which	PRON
fcis-14812	216	31	is	be	AUX
fcis-14812	216	32	based	base	VERB
fcis-14812	216	33	on	on	ADP
fcis-14812	216	34	the	the	DET
fcis-14812	216	35	bmnet+	bmnet+	PROPN
fcis-14812	216	36	framework	framework	NOUN
fcis-14812	216	37	and	and	CCONJ
fcis-14812	216	38	is	be	AUX
fcis-14812	216	39	improved	improve	VERB
fcis-14812	216	40	for	for	ADP
fcis-14812	216	41	both	both	CCONJ
fcis-14812	216	42	feature	feature	NOUN
fcis-14812	216	43	enhancement	enhancement	NOUN
fcis-14812	216	44	and	and	CCONJ
fcis-14812	216	45	feature	feature	NOUN
fcis-14812	216	46	fusion	fusion	NOUN
fcis-14812	216	47	.	.	PUNCT
fcis-14812	217	1	we	we	PRON
fcis-14812	217	2	introduce	introduce	VERB
fcis-14812	217	3	a	a	DET
fcis-14812	217	4	feature	feature	NOUN
fcis-14812	217	5	enhancement	enhancement	NOUN
fcis-14812	217	6	module	module	NOUN
fcis-14812	217	7	based	base	VERB
fcis-14812	217	8	on	on	ADP
fcis-14812	217	9	the	the	DET
fcis-14812	217	10	conditional	conditional	ADJ
fcis-14812	217	11	random	random	ADJ
fcis-14812	217	12	field	field	NOUN
fcis-14812	217	13	principle	principle	NOUN
fcis-14812	217	14	,	,	PUNCT
fcis-14812	217	15	which	which	PRON
fcis-14812	217	16	achieves	achieve	VERB
fcis-14812	217	17	further	further	ADJ
fcis-14812	217	18	optimization	optimization	NOUN
fcis-14812	217	19	of	of	ADP
fcis-14812	217	20	the	the	DET
fcis-14812	217	21	target	target	NOUN
fcis-14812	217	22	features	feature	VERB
fcis-14812	217	23	by	by	ADP
fcis-14812	217	24	modeling	model	VERB
fcis-14812	217	25	the	the	DET
fcis-14812	217	26	probability	probability	NOUN
fcis-14812	217	27	distribution	distribution	NOUN
fcis-14812	217	28	of	of	ADP
fcis-14812	217	29	each	each	DET
fcis-14812	217	30	pixel	pixel	NOUN
fcis-14812	217	31	in	in	ADP
fcis-14812	217	32	the	the	DET
fcis-14812	217	33	image	image	NOUN
fcis-14812	217	34	.	.	PUNCT
fcis-14812	218	1	secondly	secondly	ADV
fcis-14812	218	2	,	,	PUNCT
fcis-14812	218	3	we	we	PRON
fcis-14812	218	4	added	add	VERB
fcis-14812	218	5	an	an	DET
fcis-14812	218	6	adaptive	adaptive	ADJ
fcis-14812	218	7	spatial	spatial	ADJ
fcis-14812	218	8	feature	feature	NOUN
fcis-14812	218	9	fusion	fusion	NOUN
fcis-14812	218	10	module	module	NOUN
fcis-14812	218	11	,	,	PUNCT
fcis-14812	218	12	which	which	PRON
fcis-14812	218	13	realizes	realize	VERB
fcis-14812	218	14	the	the	DET
fcis-14812	218	15	enhancement	enhancement	NOUN
fcis-14812	218	16	of	of	ADP
fcis-14812	218	17	the	the	DET
fcis-14812	218	18	feature	feature	NOUN
fcis-14812	218	19	extraction	extraction	NOUN
fcis-14812	218	20	effect	effect	NOUN
fcis-14812	218	21	for	for	ADP
fcis-14812	218	22	targets	target	NOUN
fcis-14812	218	23	of	of	ADP
fcis-14812	218	24	different	different	ADJ
fcis-14812	218	25	scales	scale	NOUN
fcis-14812	218	26	through	through	ADP
fcis-14812	218	27	the	the	DET
fcis-14812	218	28	dynamic	dynamic	ADJ
fcis-14812	218	29	weight	weight	NOUN
fcis-14812	218	30	allocation	allocation	NOUN
fcis-14812	218	31	mechanism	mechanism	NOUN
fcis-14812	218	32	.	.	PUNCT
fcis-14812	219	1	through	through	ADP
fcis-14812	219	2	experimental	experimental	ADJ
fcis-14812	219	3	validation	validation	NOUN
fcis-14812	219	4	on	on	ADP
fcis-14812	219	5	the	the	DET
fcis-14812	219	6	fsc-147	fsc-147	NOUN
fcis-14812	219	7	dataset	dataset	NOUN
fcis-14812	219	8	,	,	PUNCT
fcis-14812	219	9	our	our	PRON
fcis-14812	219	10	method	method	NOUN
fcis-14812	219	11	can	can	AUX
fcis-14812	219	12	effectively	effectively	ADV
fcis-14812	219	13	realize	realize	VERB
fcis-14812	219	14	the	the	DET
fcis-14812	219	15	class	class	NOUN
fcis-14812	219	16	agnostic	agnostic	ADJ
fcis-14812	219	17	counting	counting	NOUN
fcis-14812	219	18	effect	effect	NOUN
fcis-14812	219	19	and	and	CCONJ
fcis-14812	219	20	reach	reach	VERB
fcis-14812	219	21	the	the	DET
fcis-14812	219	22	excellent	excellent	ADJ
fcis-14812	219	23	level	level	NOUN
fcis-14812	219	24	in	in	ADP
fcis-14812	219	25	the	the	DET
fcis-14812	219	26	field	field	NOUN
fcis-14812	219	27	in	in	ADP
fcis-14812	219	28	all	all	DET
fcis-14812	219	29	evaluation	evaluation	NOUN
fcis-14812	219	30	indexes	index	NOUN
fcis-14812	219	31	.	.	PUNCT
fcis-14812	220	1	there	there	PRON
fcis-14812	220	2	are	be	VERB
fcis-14812	220	3	still	still	ADV
fcis-14812	220	4	many	many	ADJ
fcis-14812	220	5	shortcomings	shortcoming	NOUN
fcis-14812	220	6	in	in	ADP
fcis-14812	220	7	this	this	DET
fcis-14812	220	8	work	work	NOUN
fcis-14812	220	9	,	,	PUNCT
fcis-14812	220	10	for	for	ADP
fcis-14812	220	11	example	example	NOUN
fcis-14812	220	12	,	,	PUNCT
fcis-14812	220	13	the	the	DET
fcis-14812	220	14	problem	problem	NOUN
fcis-14812	220	15	of	of	ADP
fcis-14812	220	16	increasing	increase	VERB
fcis-14812	220	17	computation	computation	NOUN
fcis-14812	220	18	brought	bring	VERB
fcis-14812	220	19	by	by	ADP
fcis-14812	220	20	adding	add	VERB
fcis-14812	220	21	modules	module	NOUN
fcis-14812	220	22	has	have	AUX
fcis-14812	220	23	not	not	PART
fcis-14812	220	24	been	be	AUX
fcis-14812	220	25	solved	solve	VERB
fcis-14812	220	26	,	,	PUNCT
fcis-14812	220	27	which	which	PRON
fcis-14812	220	28	may	may	AUX
fcis-14812	220	29	affect	affect	VERB
fcis-14812	220	30	the	the	DET
fcis-14812	220	31	operation	operation	NOUN
fcis-14812	220	32	speed	speed	NOUN
fcis-14812	220	33	and	and	CCONJ
fcis-14812	220	34	efficiency	efficiency	NOUN
fcis-14812	220	35	of	of	ADP
fcis-14812	220	36	the	the	DET
fcis-14812	220	37	network	network	NOUN
fcis-14812	220	38	,	,	PUNCT
fcis-14812	220	39	and	and	CCONJ
fcis-14812	220	40	we	we	PRON
fcis-14812	220	41	will	will	AUX
fcis-14812	220	42	try	try	VERB
fcis-14812	220	43	to	to	PART
fcis-14812	220	44	solve	solve	VERB
fcis-14812	220	45	this	this	DET
fcis-14812	220	46	problem	problem	NOUN
fcis-14812	220	47	in	in	ADP
fcis-14812	220	48	our	our	PRON
fcis-14812	220	49	future	future	ADJ
fcis-14812	220	50	work	work	NOUN
fcis-14812	220	51	.	.	PUNCT
fcis-14812	221	1	references	reference	NOUN
fcis-14812	221	2	[	[	X
fcis-14812	221	3	1	1	NUM
fcis-14812	221	4	]	]	PUNCT
fcis-14812	221	5	wu	wu	PROPN
fcis-14812	221	6	b	b	PROPN
fcis-14812	221	7	,	,	PUNCT
fcis-14812	221	8	nevatia	nevatia	PROPN
fcis-14812	221	9	r.detection	r.detection	NOUN
fcis-14812	221	10	and	and	CCONJ
fcis-14812	221	11	tracking	tracking	NOUN
fcis-14812	221	12	of	of	ADP
fcis-14812	221	13	multiple	multiple	ADJ
fcis-14812	221	14	,	,	PUNCT
fcis-14812	221	15	partially	partially	ADV
fcis-14812	221	16	occluded	occlude	VERB
fcis-14812	221	17	humans	human	NOUN
fcis-14812	221	18	by	by	ADP
fcis-14812	221	19	bayesian	bayesian	NOUN
fcis-14812	221	20	combination	combination	NOUN
fcis-14812	221	21	of	of	ADP
fcis-14812	221	22	edgelet	edgelet	NOUN
fcis-14812	221	23	based	base	VERB
fcis-14812	221	24	part	part	NOUN
fcis-14812	221	25	detectors[j].international	detectors[j].international	ADJ
fcis-14812	221	26	journal	journal	NOUN
fcis-14812	221	27	of	of	ADP
fcis-14812	221	28	computer	computer	NOUN
fcis-14812	221	29	vision,2007,75(2):247	vision,2007,75(2):247	NOUN
fcis-14812	221	30	-	-	PUNCT
fcis-14812	221	31	266	266	NUM
fcis-14812	221	32	.	.	PUNCT
fcis-14812	222	1	[	[	X
fcis-14812	222	2	2	2	NUM
fcis-14812	222	3	]	]	X
fcis-14812	222	4	lin	lin	PROPN
fcis-14812	222	5	s	s	PROPN
fcis-14812	222	6	f	f	PROPN
fcis-14812	222	7	,	,	PUNCT
fcis-14812	222	8	chen	chen	PROPN
fcis-14812	222	9	j	j	PROPN
fcis-14812	222	10	y	y	PROPN
fcis-14812	222	11	,	,	PUNCT
fcis-14812	222	12	chao	chao	PROPN
fcis-14812	222	13	h	h	PROPN
fcis-14812	222	14	x.estimation	x.estimation	PROPN
fcis-14812	222	15	of	of	ADP
fcis-14812	222	16	number	number	NOUN
fcis-14812	222	17	of	of	ADP
fcis-14812	222	18	people	people	NOUN
fcis-14812	222	19	in	in	ADP
fcis-14812	222	20	crowded	crowded	ADJ
fcis-14812	222	21	scenes	scene	NOUN
fcis-14812	222	22	using	use	VERB
fcis-14812	222	23	perspective	perspective	ADJ
fcis-14812	222	24	transformation	transformation	NOUN
fcis-14812	223	1	[	[	X
fcis-14812	223	2	j	j	X
fcis-14812	223	3	]	]	X
fcis-14812	223	4	.	.	PUNCT
fcis-14812	224	1	ieee	ieee	NOUN
fcis-14812	224	2	transactions	transaction	NOUN
fcis-14812	224	3	on	on	ADP
fcis-14812	224	4	systems	system	NOUN
fcis-14812	224	5	,	,	PUNCT
fcis-14812	224	6	man	man	NOUN
fcis-14812	224	7	,	,	PUNCT
fcis-14812	224	8	and	and	CCONJ
fcis-14812	224	9	cybernetics	cybernetic	NOUN
fcis-14812	224	10	,	,	PUNCT
fcis-14812	224	11	part	part	NOUN
fcis-14812	224	12	a	a	DET
fcis-14812	224	13	:	:	PUNCT
fcis-14812	224	14	systems	system	NOUN
fcis-14812	224	15	and	and	CCONJ
fcis-14812	224	16	humans,2001,31:645	humans,2001,31:645	PROPN
fcis-14812	224	17	-	-	NOUN
fcis-14812	224	18	654	654	NUM
fcis-14812	224	19	.	.	PUNCT
fcis-14812	225	1	[	[	X
fcis-14812	225	2	3	3	X
fcis-14812	225	3	]	]	SYM
fcis-14812	225	4	min	min	PROPN
fcis-14812	225	5	l	l	PROPN
fcis-14812	225	6	,	,	PUNCT
fcis-14812	225	7	zhang	zhang	PROPN
fcis-14812	225	8	z	z	PROPN
fcis-14812	225	9	,	,	PUNCT
fcis-14812	225	10	huang	huang	PROPN
fcis-14812	225	11	k	k	PROPN
fcis-14812	225	12	,	,	PUNCT
fcis-14812	225	13	et	et	NOUN
fcis-14812	225	14	al.estimating	al.estimate	VERB
fcis-14812	225	15	the	the	DET
fcis-14812	225	16	number	number	NOUN
fcis-14812	225	17	of	of	ADP
fcis-14812	225	18	people	people	NOUN
fcis-14812	225	19	in	in	ADP
fcis-14812	225	20	crowded	crowded	ADJ
fcis-14812	225	21	scenes	scene	NOUN
fcis-14812	225	22	by	by	ADP
fcis-14812	225	23	mid	mid	ADJ
fcis-14812	225	24	based	base	VERB
fcis-14812	225	25	foreground	foreground	NOUN
fcis-14812	225	26	segmentation	segmentation	NOUN
fcis-14812	225	27	and	and	CCONJ
fcis-14812	225	28	head	head	NOUN
fcis-14812	225	29	-	-	PUNCT
fcis-14812	225	30	shoulder	shoulder	NOUN
fcis-14812	225	31	detection[c]//	detection[c]//	PROPN
fcis-14812	225	32	2008	2008	NUM
fcis-14812	225	33	19th	19th	ADJ
fcis-14812	225	34	international	international	ADJ
fcis-14812	225	35	conference	conference	NOUN
fcis-14812	225	36	on	on	ADP
fcis-14812	225	37	pattern	pattern	NOUN
fcis-14812	225	38	recognition,2009	recognition,2009	NOUN
fcis-14812	225	39	.	.	PUNCT
fcis-14812	226	1	[	[	X
fcis-14812	226	2	4	4	NUM
fcis-14812	226	3	]	]	PUNCT
fcis-14812	226	4	xu	xu	PROPN
fcis-14812	226	5	t	t	PROPN
fcis-14812	226	6	,	,	PUNCT
fcis-14812	226	7	chen	chen	PROPN
fcis-14812	226	8	x	x	PROPN
fcis-14812	226	9	,	,	PUNCT
fcis-14812	226	10	wei	wei	PROPN
fcis-14812	226	11	g	g	PROPN
fcis-14812	226	12	,	,	PUNCT
fcis-14812	226	13	et	et	PROPN
fcis-14812	226	14	al.crowd	al.crowd	ADP
fcis-14812	226	15	counting	count	VERB
fcis-14812	226	16	using	use	VERB
fcis-14812	226	17	accumulated	accumulate	VERB
fcis-14812	226	18	hog[c]//	hog[c]//	ADJ
fcis-14812	226	19	2016	2016	NUM
fcis-14812	226	20	12th	12th	NOUN
fcis-14812	226	21	international	international	ADJ
fcis-14812	226	22	conference	conference	NOUN
fcis-14812	226	23	on	on	ADP
fcis-14812	226	24	natural	natural	ADJ
fcis-14812	226	25	computation	computation	NOUN
fcis-14812	226	26	,	,	PUNCT
fcis-14812	226	27	fuzzy	fuzzy	ADJ
fcis-14812	226	28	systems	system	NOUN
fcis-14812	226	29	and	and	CCONJ
fcis-14812	226	30	knowledge	knowledge	NOUN
fcis-14812	226	31	discovery(icnc	discovery(icnc	PROPN
fcis-14812	226	32	-	-	PUNCT
fcis-14812	226	33	fskd),2016:1877	fskd),2016:1877	NOUN
fcis-14812	226	34	-	-	PUNCT
fcis-14812	226	35	1881	1881	NUM
fcis-14812	226	36	.	.	PUNCT
fcis-14812	227	1	[	[	X
fcis-14812	227	2	5	5	X
fcis-14812	227	3	]	]	PUNCT
fcis-14812	227	4	chan	chan	NOUN
fcis-14812	227	5	a	a	DET
fcis-14812	227	6	b	b	PROPN
fcis-14812	227	7	,	,	PUNCT
fcis-14812	227	8	liang	liang	PROPN
fcis-14812	227	9	z	z	PROPN
fcis-14812	227	10	s	s	PROPN
fcis-14812	227	11	j	j	PROPN
fcis-14812	227	12	,	,	PUNCT
fcis-14812	227	13	vasconcelos	vasconcelos	PROPN
fcis-14812	227	14	n.privacy	n.privacy	NOUN
fcis-14812	227	15	preserving	preserve	VERB
fcis-14812	227	16	crowd	crowd	NOUN
fcis-14812	227	17	monitoring	monitoring	NOUN
fcis-14812	227	18	:	:	PUNCT
fcis-14812	227	19	counting	count	VERB
fcis-14812	227	20	people	people	NOUN
fcis-14812	227	21	without	without	ADP
fcis-14812	227	22	people	people	NOUN
fcis-14812	227	23	models	model	NOUN
fcis-14812	227	24	or	or	CCONJ
fcis-14812	227	25	tracking[c]//	tracking[c]//	PROPN
fcis-14812	227	26	2008	2008	NUM
fcis-14812	227	27	ieee	ieee	NOUN
fcis-14812	227	28	conference	conference	NOUN
fcis-14812	227	29	on	on	ADP
fcis-14812	227	30	computer	computer	NOUN
fcis-14812	227	31	vision	vision	NOUN
fcis-14812	227	32	and	and	CCONJ
fcis-14812	227	33	pattern	pattern	NOUN
fcis-14812	227	34	recognition,2008:1	recognition,2008:1	NOUN
fcis-14812	227	35	-	-	SYM
fcis-14812	227	36	7	7	NUM
fcis-14812	227	37	.	.	PUNCT
fcis-14812	228	1	[	[	X
fcis-14812	228	2	6	6	NUM
fcis-14812	228	3	]	]	SYM
fcis-14812	228	4	ryan	ryan	PROPN
fcis-14812	228	5	d	d	PROPN
fcis-14812	228	6	,	,	PUNCT
fcis-14812	228	7	denman	denman	PROPN
fcis-14812	228	8	s	s	PROPN
fcis-14812	228	9	,	,	PUNCT
fcis-14812	228	10	fookes	fooke	VERB
fcis-14812	228	11	c	c	NOUN
fcis-14812	228	12	,	,	PUNCT
fcis-14812	228	13	et	et	PROPN
fcis-14812	228	14	al.crowd	al.crowd	NOUN
fcis-14812	228	15	counting	count	VERB
fcis-14812	228	16	using	use	VERB
fcis-14812	228	17	multiple	multiple	ADJ
fcis-14812	228	18	local	local	ADJ
fcis-14812	228	19	features[c]//	features[c]//	PROPN
fcis-14812	228	20	2009	2009	NUM
fcis-14812	228	21	digital	digital	ADJ
fcis-14812	228	22	image	image	NOUN
fcis-14812	228	23	computing	computing	NOUN
fcis-14812	228	24	:	:	PUNCT
fcis-14812	228	25	techniques	technique	NOUN
fcis-14812	228	26	and	and	CCONJ
fcis-14812	228	27	applications,2009:81	applications,2009:81	PROPN
fcis-14812	228	28	-	-	PUNCT
fcis-14812	228	29	88	88	NUM
fcis-14812	228	30	.	.	PUNCT
fcis-14812	229	1	[	[	X
fcis-14812	229	2	7	7	X
fcis-14812	229	3	]	]	X
fcis-14812	229	4	ke	ke	NOUN
fcis-14812	229	5	c	c	PROPN
fcis-14812	229	6	,	,	PUNCT
fcis-14812	229	7	chen	chen	PROPN
fcis-14812	229	8	c	c	PROPN
fcis-14812	229	9	l	l	PROPN
fcis-14812	229	10	,	,	PUNCT
fcis-14812	229	11	gong	gong	PROPN
fcis-14812	229	12	s	s	PROPN
fcis-14812	229	13	,	,	PUNCT
fcis-14812	229	14	et	et	PROPN
fcis-14812	229	15	al.feature	al.feature	NOUN
fcis-14812	229	16	mining	mining	NOUN
fcis-14812	229	17	for	for	ADP
fcis-14812	229	18	localised	localised	ADJ
fcis-14812	229	19	crowd	crowd	NOUN
fcis-14812	229	20	counting[c]//	counting[c]//	PROPN
fcis-14812	229	21	british	british	ADJ
fcis-14812	229	22	machine	machine	NOUN
fcis-14812	229	23	vision	vision	PROPN
fcis-14812	229	24	conference,2012	conference,2012	PROPN
fcis-14812	229	25	.	.	PUNCT
fcis-14812	230	1	[	[	X
fcis-14812	230	2	8	8	NUM
fcis-14812	230	3	]	]	PUNCT
fcis-14812	230	4	paragios	paragio	NOUN
fcis-14812	230	5	n	n	CCONJ
fcis-14812	230	6	,	,	PUNCT
fcis-14812	230	7	ramesh	ramesh	PROPN
fcis-14812	230	8	v.a	v.a	PROPN
fcis-14812	230	9	mrf	mrf	PROPN
fcis-14812	230	10	-	-	PUNCT
fcis-14812	230	11	based	base	VERB
fcis-14812	230	12	approach	approach	NOUN
fcis-14812	230	13	for	for	ADP
fcis-14812	230	14	realtime	realtime	NOUN
fcis-14812	230	15	subway	subway	NOUN
fcis-14812	230	16	monitoring[c]//	monitoring[c]//	NOUN
fcis-14812	230	17	proceedings	proceeding	NOUN
fcis-14812	230	18	of	of	ADP
fcis-14812	230	19	the	the	DET
fcis-14812	230	20	2001	2001	NUM
fcis-14812	230	21	ieee	ieee	NOUN
fcis-14812	230	22	computer	computer	NOUN
fcis-14812	230	23	society	society	PROPN
fcis-14812	230	24	conference	conference	NOUN
fcis-14812	230	25	on	on	ADP
fcis-14812	230	26	computer	computer	NOUN
fcis-14812	230	27	vision	vision	NOUN
fcis-14812	230	28	and	and	CCONJ
fcis-14812	230	29	pattern	pattern	NOUN
fcis-14812	230	30	recognition,2001	recognition,2001	NOUN
fcis-14812	230	31	.	.	PUNCT
fcis-14812	231	1	[	[	X
fcis-14812	231	2	9	9	NUM
fcis-14812	231	3	]	]	X
fcis-14812	231	4	mcdonald	mcdonald	NOUN
fcis-14812	231	5	g	g	PROPN
fcis-14812	231	6	c.ridge	c.ridge	NOUN
fcis-14812	231	7	regression[j].wiley	regression[j].wiley	ADJ
fcis-14812	231	8	interdisciplinary	interdisciplinary	ADJ
fcis-14812	231	9	reviews	review	NOUN
fcis-14812	231	10	:	:	PUNCT
fcis-14812	231	11	computational	computational	ADJ
fcis-14812	231	12	statistics,2009,1(1):93	statistics,2009,1(1):93	X
fcis-14812	231	13	-	-	PUNCT
fcis-14812	231	14	100	100	NUM
fcis-14812	231	15	.	.	PUNCT
fcis-14812	232	1	[	[	X
fcis-14812	232	2	10	10	NUM
fcis-14812	232	3	]	]	X
fcis-14812	232	4	n.	n.	PROPN
fcis-14812	232	5	dalal	dalal	PROPN
fcis-14812	232	6	and	and	CCONJ
fcis-14812	232	7	b.	b.	PROPN
fcis-14812	232	8	triggs	triggs	PROPN
fcis-14812	232	9	.	.	PUNCT
fcis-14812	233	1	histograms	histogram	NOUN
fcis-14812	233	2	of	of	ADP
fcis-14812	233	3	oriented	orient	VERB
fcis-14812	233	4	gradients	gradient	NOUN
fcis-14812	233	5	for	for	ADP
fcis-14812	233	6	human	human	ADJ
fcis-14812	233	7	detection[c]//2005	detection[c]//2005	PROPN
fcis-14812	233	8	ieee	ieee	PROPN
fcis-14812	233	9	computer	computer	NOUN
fcis-14812	233	10	society	society	PROPN
fcis-14812	233	11	conference	conference	NOUN
fcis-14812	233	12	on	on	ADP
fcis-14812	233	13	computer	computer	NOUN
fcis-14812	233	14	vision	vision	NOUN
fcis-14812	233	15	and	and	CCONJ
fcis-14812	233	16	pattern	pattern	NOUN
fcis-14812	233	17	recognition	recognition	NOUN
fcis-14812	233	18	(	(	PUNCT
fcis-14812	233	19	cvpr'05	cvpr'05	PROPN
fcis-14812	233	20	)	)	PUNCT
fcis-14812	233	21	,	,	PUNCT
fcis-14812	233	22	2005	2005	NUM
fcis-14812	233	23	,	,	PUNCT
fcis-14812	233	24	pp	pp	PROPN
fcis-14812	233	25	.	.	PUNCT
fcis-14812	233	26	886	886	NUM
fcis-14812	233	27	-	-	SYM
fcis-14812	233	28	893	893	NUM
fcis-14812	233	29	vol	vol	NOUN
fcis-14812	233	30	.	.	NOUN
fcis-14812	233	31	1	1	NUM
fcis-14812	233	32	,	,	PUNCT
fcis-14812	233	33	doi	doi	NOUN
fcis-14812	233	34	:	:	PUNCT
fcis-14812	233	35	10.1109	10.1109	NUM
fcis-14812	233	36	/	/	SYM
fcis-14812	233	37	cvpr.2005.177	cvpr.2005.177	PROPN
fcis-14812	233	38	.	.	PUNCT
fcis-14812	234	1	[	[	X
fcis-14812	234	2	11	11	NUM
fcis-14812	234	3	]	]	X
fcis-14812	234	4	girshick	girshick	NOUN
fcis-14812	234	5	r	r	NOUN
fcis-14812	234	6	,	,	PUNCT
fcis-14812	234	7	donahue	donahue	PROPN
fcis-14812	234	8	j	j	PROPN
fcis-14812	234	9	,	,	PUNCT
fcis-14812	234	10	darrell	darrell	PROPN
fcis-14812	234	11	t	t	PROPN
fcis-14812	234	12	,	,	PUNCT
fcis-14812	234	13	et	et	PROPN
fcis-14812	234	14	al.rich	al.rich	ADP
fcis-14812	234	15	feature	feature	NOUN
fcis-14812	234	16	hierarchies	hierarchy	NOUN
fcis-14812	234	17	for	for	ADP
fcis-14812	234	18	accurate	accurate	ADJ
fcis-14812	234	19	object	object	NOUN
fcis-14812	234	20	detection	detection	NOUN
fcis-14812	234	21	and	and	CCONJ
fcis-14812	234	22	semantic	semantic	ADJ
fcis-14812	234	23	segmentation	segmentation	NOUN
fcis-14812	234	24	tech	tech	NOUN
fcis-14812	234	25	report	report	NOUN
fcis-14812	234	26	(	(	PUNCT
fcis-14812	234	27	v5)[j	v5)[j	NOUN
fcis-14812	234	28	]	]	PUNCT
fcis-14812	234	29	.	.	PUNCT
fcis-14812	235	1	2017.doi:10	2017.doi:10	NUM
fcis-14812	235	2	.	.	PUNCT
fcis-14812	236	1	1109/	1109/	NUM
fcis-14812	236	2	cvpr	cvpr	NOUN
fcis-14812	236	3	.	.	PUNCT
fcis-14812	237	1	2014.81	2014.81	NUM
fcis-14812	237	2	.	.	PUNCT
fcis-14812	238	1	[	[	X
fcis-14812	238	2	12	12	NUM
fcis-14812	238	3	]	]	X
fcis-14812	238	4	redmon	redmon	PROPN
fcis-14812	238	5	j	j	PROPN
fcis-14812	238	6	,	,	PUNCT
fcis-14812	238	7	divvala	divvala	PROPN
fcis-14812	238	8	s	s	PROPN
fcis-14812	238	9	,	,	PUNCT
fcis-14812	238	10	girshick	girshick	ADJ
fcis-14812	238	11	r	r	NOUN
fcis-14812	238	12	,	,	PUNCT
fcis-14812	238	13	et	et	PROPN
fcis-14812	238	14	al	al	PROPN
fcis-14812	238	15	.	.	PUNCT
fcis-14812	239	1	you	you	PRON
fcis-14812	239	2	only	only	ADV
fcis-14812	239	3	look	look	VERB
fcis-14812	239	4	once	once	ADV
fcis-14812	239	5	:	:	PUNCT
fcis-14812	239	6	unified	unified	ADJ
fcis-14812	239	7	,	,	PUNCT
fcis-14812	239	8	real	real	ADJ
fcis-14812	239	9	-	-	PUNCT
fcis-14812	239	10	time	time	NOUN
fcis-14812	239	11	object	object	NOUN
fcis-14812	239	12	detection[c]//proceedings	detection[c]//proceeding	NOUN
fcis-14812	239	13	of	of	ADP
fcis-14812	239	14	the	the	DET
fcis-14812	239	15	ieee	ieee	NOUN
fcis-14812	239	16	conference	conference	NOUN
fcis-14812	239	17	on	on	ADP
fcis-14812	239	18	computer	computer	NOUN
fcis-14812	239	19	vision	vision	NOUN
fcis-14812	239	20	and	and	CCONJ
fcis-14812	239	21	pattern	pattern	NOUN
fcis-14812	239	22	recognition	recognition	NOUN
fcis-14812	239	23	.	.	PUNCT
fcis-14812	240	1	2016	2016	NUM
fcis-14812	240	2	:	:	PUNCT
fcis-14812	240	3	779	779	NUM
fcis-14812	240	4	-	-	SYM
fcis-14812	240	5	788	788	NUM
fcis-14812	240	6	.	.	PUNCT
fcis-14812	241	1	[	[	X
fcis-14812	241	2	13	13	NUM
fcis-14812	241	3	]	]	X
fcis-14812	241	4	erika	erika	PROPN
fcis-14812	241	5	lu	lu	PROPN
fcis-14812	241	6	,	,	PUNCT
fcis-14812	241	7	weidi	weidi	PROPN
fcis-14812	241	8	xie	xie	PROPN
fcis-14812	241	9	,	,	PUNCT
fcis-14812	241	10	andrew	andrew	PROPN
fcis-14812	241	11	zisserman	zisserman	PROPN
fcis-14812	241	12	.	.	PUNCT
fcis-14812	242	1	class	class	NOUN
fcis-14812	242	2	-	-	PUNCT
fcis-14812	242	3	agnostic	agnostic	ADJ
fcis-14812	242	4	counting	counting	NOUN
fcis-14812	242	5	[	[	X
fcis-14812	242	6	c]//asian	c]//asian	ADJ
fcis-14812	242	7	conference	conference	NOUN
fcis-14812	242	8	on	on	ADP
fcis-14812	242	9	computer	computer	NOUN
fcis-14812	242	10	vision	vision	NOUN
fcis-14812	242	11	(	(	PUNCT
fcis-14812	242	12	accv),2018	accv),2018	NOUN
fcis-14812	242	13	.	.	NOUN
fcis-14812	242	14	arxiv:1811.00472	arxiv:1811.00472	NOUN
fcis-14812	242	15	.	.	PUNCT
fcis-14812	243	1	[	[	X
fcis-14812	243	2	14	14	NUM
fcis-14812	243	3	]	]	X
fcis-14812	243	4	viresh	viresh	PROPN
fcis-14812	243	5	ranjan	ranjan	PROPN
fcis-14812	243	6	,	,	PUNCT
fcis-14812	243	7	udbhav	udbhav	PROPN
fcis-14812	243	8	sharma	sharma	PROPN
fcis-14812	243	9	,	,	PUNCT
fcis-14812	243	10	thu	thu	PROPN
fcis-14812	243	11	nguyen	nguyen	NOUN
fcis-14812	243	12	,	,	PUNCT
fcis-14812	243	13	minh	minh	X
fcis-14812	243	14	hoai	hoai	X
fcis-14812	243	15	.	.	PUNCT
fcis-14812	244	1	learning	learn	VERB
fcis-14812	244	2	to	to	PART
fcis-14812	244	3	count	count	VERB
fcis-14812	244	4	everything[c]//computer	everything[c]//comput	ADJ
fcis-14812	244	5	vision	vision	NOUN
fcis-14812	244	6	and	and	CCONJ
fcis-14812	244	7	pattern	pattern	NOUN
fcis-14812	244	8	recognition(cvpr	recognition(cvpr	NOUN
fcis-14812	244	9	)	)	PUNCT
fcis-14812	244	10	,	,	PUNCT
fcis-14812	244	11	2021	2021	NUM
fcis-14812	244	12	.	.	PUNCT
fcis-14812	245	1	arxiv:2104.08391	arxiv:2104.08391	NOUN
fcis-14812	245	2	.	.	PUNCT
fcis-14812	246	1	[	[	X
fcis-14812	246	2	15	15	NUM
fcis-14812	246	3	]	]	SYM
fcis-14812	246	4	min	min	PROPN
fcis-14812	246	5	shi	shi	PROPN
fcis-14812	246	6	,	,	PUNCT
fcis-14812	246	7	hao	hao	PROPN
fcis-14812	246	8	lu	lu	PROPN
fcis-14812	246	9	,	,	PUNCT
fcis-14812	246	10	chen	chen	PROPN
fcis-14812	246	11	feng	feng	PROPN
fcis-14812	246	12	,	,	PUNCT
fcis-14812	246	13	chengxin	chengxin	PROPN
fcis-14812	246	14	liu	liu	PROPN
fcis-14812	246	15	,	,	PUNCT
fcis-14812	246	16	zhiguo	zhiguo	PROPN
fcis-14812	246	17	cao	cao	PROPN
fcis-14812	246	18	.	.	PUNCT
fcis-14812	247	1	represent	represent	PROPN
fcis-14812	247	2	,	,	PUNCT
fcis-14812	247	3	compare	compare	NOUN
fcis-14812	247	4	,	,	PUNCT
fcis-14812	247	5	and	and	CCONJ
fcis-14812	247	6	learn	learn	VERB
fcis-14812	247	7	:	:	PUNCT
fcis-14812	247	8	a	a	DET
fcis-14812	247	9	similarity	similarity	NOUN
fcis-14812	247	10	-	-	PUNCT
fcis-14812	247	11	aware	aware	ADJ
fcis-14812	247	12	framework	framework	NOUN
fcis-14812	247	13	for	for	ADP
fcis-14812	247	14	class	class	NOUN
fcis-14812	247	15	-	-	PUNCT
fcis-14812	247	16	agnostic	agnostic	ADJ
fcis-14812	247	17	counting[c]//	counting[c]//	ADJ
fcis-14812	247	18	computer	computer	NOUN
fcis-14812	247	19	vision	vision	NOUN
fcis-14812	247	20	and	and	CCONJ
fcis-14812	247	21	pattern	pattern	NOUN
fcis-14812	247	22	recognition(cvpr	recognition(cvpr	NOUN
fcis-14812	247	23	)	)	PUNCT
fcis-14812	247	24	,	,	PUNCT
fcis-14812	247	25	2022	2022	NUM
fcis-14812	247	26	.	.	PUNCT
fcis-14812	248	1	arxiv	arxiv	NOUN
fcis-14812	248	2	:	:	PUNCT
fcis-14812	248	3	2203	2203	NUM
fcis-14812	248	4	.	.	PUNCT
fcis-14812	248	5	08354	08354	NUM
fcis-14812	248	6	.	.	PUNCT
fcis-14812	249	1	[	[	X
fcis-14812	249	2	16	16	NUM
fcis-14812	249	3	]	]	X
fcis-14812	249	4	zheng	zheng	PROPN
fcis-14812	249	5	s	s	PROPN
fcis-14812	249	6	,	,	PUNCT
fcis-14812	249	7	jayasumana	jayasumana	PROPN
fcis-14812	249	8	s	s	PART
fcis-14812	249	9	,	,	PUNCT
fcis-14812	249	10	romera	romera	NOUN
fcis-14812	249	11	-	-	PUNCT
fcis-14812	249	12	paredes	paredes	PROPN
fcis-14812	249	13	b	b	NOUN
fcis-14812	249	14	,	,	PUNCT
fcis-14812	249	15	et	et	PROPN
fcis-14812	249	16	al	al	PROPN
fcis-14812	249	17	.	.	PUNCT
fcis-14812	249	18	conditional	conditional	ADJ
fcis-14812	249	19	random	random	ADJ
fcis-14812	249	20	fields	field	NOUN
fcis-14812	249	21	as	as	ADP
fcis-14812	249	22	recurrent	recurrent	ADJ
fcis-14812	249	23	neural	neural	ADJ
fcis-14812	249	24	networks[c]//proceedings	networks[c]//proceeding	NOUN
fcis-14812	249	25	of	of	ADP
fcis-14812	249	26	the	the	DET
fcis-14812	249	27	ieee	ieee	NOUN
fcis-14812	249	28	international	international	PROPN
fcis-14812	249	29	conference	conference	NOUN
fcis-14812	249	30	on	on	ADP
fcis-14812	249	31	computer	computer	NOUN
fcis-14812	249	32	vision	vision	NOUN
fcis-14812	249	33	.	.	PUNCT
fcis-14812	250	1	2015	2015	NUM
fcis-14812	250	2	:	:	PUNCT
fcis-14812	250	3	1529	1529	NUM
fcis-14812	250	4	-	-	SYM
fcis-14812	250	5	1537	1537	NUM
fcis-14812	250	6	.	.	PUNCT
fcis-14812	251	1	[	[	X
fcis-14812	251	2	17	17	NUM
fcis-14812	251	3	]	]	X
fcis-14812	251	4	dan	dan	PROPN
fcis-14812	251	5	xu	xu	PROPN
fcis-14812	251	6	,	,	PUNCT
fcis-14812	251	7	wanli	wanli	PROPN
fcis-14812	251	8	ouyang	ouyang	PROPN
fcis-14812	251	9	,	,	PUNCT
fcis-14812	251	10	xavier	xavier	PROPN
fcis-14812	251	11	alameda	alameda	PROPN
fcis-14812	251	12	-	-	PUNCT
fcis-14812	251	13	pineda	pineda	PROPN
fcis-14812	251	14	,	,	PUNCT
fcis-14812	251	15	elisa	elisa	PROPN
fcis-14812	251	16	ricci	ricci	PROPN
fcis-14812	251	17	,	,	PUNCT
fcis-14812	251	18	xiaogang	xiaogang	PROPN
fcis-14812	251	19	wang	wang	PROPN
fcis-14812	251	20	,	,	PUNCT
fcis-14812	251	21	and	and	CCONJ
fcis-14812	251	22	nicu	nicu	PROPN
fcis-14812	251	23	sebe	sebe	PROPN
fcis-14812	251	24	.	.	PUNCT
fcis-14812	252	1	learning	learn	VERB
fcis-14812	252	2	deep	deep	ADJ
fcis-14812	252	3	structured	structured	ADJ
fcis-14812	252	4	multi	multi	ADJ
fcis-14812	252	5	-	-	ADJ
fcis-14812	252	6	scale	scale	ADJ
fcis-14812	252	7	features	feature	NOUN
fcis-14812	252	8	using	use	VERB
fcis-14812	252	9	attention	attention	NOUN
fcis-14812	252	10	-	-	PUNCT
fcis-14812	252	11	gated	gate	VERB
fcis-14812	252	12	crfs	crf	NOUN
fcis-14812	252	13	for	for	ADP
fcis-14812	252	14	contour	contour	NOUN
fcis-14812	252	15	prediction	prediction	NOUN
fcis-14812	252	16	.	.	PUNCT
fcis-14812	253	1	in	in	ADP
fcis-14812	253	2	nips	nip	NOUN
fcis-14812	253	3	,	,	PUNCT
fcis-14812	253	4	pages	page	NOUN
fcis-14812	253	5	3961–3970	3961–3970	NUM
fcis-14812	253	6	,	,	PUNCT
fcis-14812	253	7	2017	2017	NUM
fcis-14812	253	8	.	.	PUNCT
fcis-14812	254	1	[	[	X
fcis-14812	254	2	18	18	NUM
fcis-14812	254	3	]	]	X
fcis-14812	254	4	wang	wang	PROPN
fcis-14812	254	5	d	d	PROPN
fcis-14812	254	6	,	,	PUNCT
fcis-14812	254	7	ouyang	ouyang	PROPN
fcis-14812	254	8	w	w	PROPN
fcis-14812	254	9	,	,	PUNCT
fcis-14812	254	10	li	li	PROPN
fcis-14812	254	11	w	w	PROPN
fcis-14812	254	12	,	,	PUNCT
fcis-14812	254	13	et	et	PROPN
fcis-14812	254	14	al	al	PROPN
fcis-14812	254	15	.	.	PROPN
fcis-14812	254	16	dividing	divide	VERB
fcis-14812	254	17	and	and	CCONJ
fcis-14812	254	18	aggregating	aggregate	VERB
fcis-14812	254	19	network	network	NOUN
fcis-14812	254	20	for	for	ADP
fcis-14812	254	21	multi	multi	ADJ
fcis-14812	254	22	-	-	ADJ
fcis-14812	254	23	view	view	ADJ
fcis-14812	254	24	action	action	NOUN
fcis-14812	254	25	recognition[c]//proceedings	recognition[c]//proceeding	NOUN
fcis-14812	254	26	of	of	ADP
fcis-14812	254	27	the	the	DET
fcis-14812	254	28	european	european	ADJ
fcis-14812	254	29	conference	conference	PROPN
fcis-14812	254	30	on	on	ADP
fcis-14812	254	31	computer	computer	NOUN
fcis-14812	254	32	vision	vision	NOUN
fcis-14812	254	33	(	(	PUNCT
fcis-14812	254	34	eccv	eccv	ADV
fcis-14812	254	35	)	)	PUNCT
fcis-14812	254	36	.	.	PUNCT
fcis-14812	255	1	2018	2018	NUM
fcis-14812	255	2	:	:	PUNCT
fcis-14812	255	3	451	451	NUM
fcis-14812	255	4	-	-	SYM
fcis-14812	255	5	467	467	NUM
fcis-14812	255	6	.	.	PUNCT
fcis-14812	256	1	[	[	X
fcis-14812	256	2	19	19	NUM
fcis-14812	256	3	]	]	X
fcis-14812	256	4	liu	liu	PROPN
fcis-14812	256	5	l	l	PROPN
fcis-14812	256	6	,	,	PUNCT
fcis-14812	256	7	qiu	qiu	PROPN
fcis-14812	256	8	z	z	PROPN
fcis-14812	256	9	,	,	PUNCT
fcis-14812	256	10	li	li	PROPN
fcis-14812	256	11	g	g	PROPN
fcis-14812	256	12	,	,	PUNCT
fcis-14812	256	13	et	et	PROPN
fcis-14812	256	14	al	al	PROPN
fcis-14812	256	15	.	.	PROPN
fcis-14812	256	16	crowd	crowd	PROPN
fcis-14812	256	17	counting	count	VERB
fcis-14812	256	18	with	with	ADP
fcis-14812	256	19	deep	deep	ADJ
fcis-14812	256	20	structured	structured	ADJ
fcis-14812	256	21	scale	scale	NOUN
fcis-14812	256	22	integration	integration	NOUN
fcis-14812	256	23	network[c]//proceedings	network[c]//proceeding	NOUN
fcis-14812	256	24	of	of	ADP
fcis-14812	256	25	the	the	DET
fcis-14812	256	26	ieee	ieee	NOUN
fcis-14812	256	27	/	/	SYM
fcis-14812	256	28	cvf	cvf	NOUN
fcis-14812	256	29	international	international	ADJ
fcis-14812	256	30	conference	conference	NOUN
fcis-14812	256	31	on	on	ADP
fcis-14812	256	32	computer	computer	NOUN
fcis-14812	256	33	vision	vision	NOUN
fcis-14812	256	34	.	.	PUNCT
fcis-14812	257	1	2019	2019	NUM
fcis-14812	257	2	:	:	PUNCT
fcis-14812	257	3	1774	1774	NUM
fcis-14812	257	4	-	-	SYM
fcis-14812	257	5	1783	1783	NUM
fcis-14812	257	6	.	.	PUNCT
fcis-14812	258	1	[	[	X
fcis-14812	258	2	20	20	NUM
fcis-14812	258	3	]	]	X
fcis-14812	258	4	liu	liu	PROPN
fcis-14812	258	5	s	s	PROPN
fcis-14812	258	6	,	,	PUNCT
fcis-14812	258	7	huang	huang	PROPN
fcis-14812	258	8	d	d	PROPN
fcis-14812	258	9	,	,	PUNCT
fcis-14812	258	10	wang	wang	PROPN
fcis-14812	258	11	y.	y.	PROPN
fcis-14812	258	12	learning	learn	VERB
fcis-14812	258	13	spatial	spatial	ADJ
fcis-14812	258	14	fusion	fusion	NOUN
fcis-14812	258	15	for	for	ADP
fcis-14812	258	16	singleshot	singleshot	ADJ
fcis-14812	258	17	object	object	NOUN
fcis-14812	258	18	detection[j	detection[j	PROPN
fcis-14812	258	19	]	]	PUNCT
fcis-14812	258	20	.	.	PUNCT
fcis-14812	259	1	arxiv	arxiv	PROPN
fcis-14812	259	2	preprint	preprint	VERB
fcis-14812	259	3	arxiv:1911.09516	arxiv:1911.09516	ADJ
fcis-14812	259	4	,	,	PUNCT
fcis-14812	259	5	2019	2019	NUM
fcis-14812	259	6	.	.	PUNCT
fcis-14812	260	1	[	[	X
fcis-14812	260	2	21	21	NUM
fcis-14812	260	3	]	]	X
fcis-14812	260	4	yang	yang	PROPN
fcis-14812	260	5	s	s	PROPN
fcis-14812	260	6	d	d	PROPN
fcis-14812	260	7	,	,	PUNCT
fcis-14812	260	8	su	su	PROPN
fcis-14812	260	9	h	h	PROPN
fcis-14812	260	10	t	t	PROPN
fcis-14812	260	11	,	,	PUNCT
fcis-14812	260	12	hsu	hsu	PROPN
fcis-14812	260	13	w	w	PROPN
fcis-14812	260	14	h	h	PROPN
fcis-14812	260	15	,	,	PUNCT
fcis-14812	260	16	et	et	PROPN
fcis-14812	260	17	al	al	PROPN
fcis-14812	260	18	.	.	PUNCT
fcis-14812	260	19	class	class	NOUN
fcis-14812	260	20	-	-	PUNCT
fcis-14812	260	21	agnostic	agnostic	ADJ
fcis-14812	260	22	few	few	ADJ
fcis-14812	260	23	-	-	PUNCT
fcis-14812	260	24	shot	shot	NOUN
fcis-14812	260	25	object	object	NOUN
fcis-14812	260	26	counting[c]//proceedings	counting[c]//proceeding	NOUN
fcis-14812	260	27	of	of	ADP
fcis-14812	260	28	the	the	DET
fcis-14812	260	29	ieee	ieee	NOUN
fcis-14812	260	30	/	/	SYM
fcis-14812	260	31	cvf	cvf	NOUN
fcis-14812	260	32	winter	winter	NOUN
fcis-14812	260	33	conference	conference	NOUN
fcis-14812	260	34	on	on	ADP
fcis-14812	260	35	applications	application	NOUN
fcis-14812	260	36	of	of	ADP
fcis-14812	260	37	computer	computer	NOUN
fcis-14812	260	38	vision	vision	NOUN
fcis-14812	260	39	.	.	PUNCT
fcis-14812	261	1	2021	2021	NUM
fcis-14812	261	2	:	:	PUNCT
fcis-14812	261	3	870878	870878	NUM
fcis-14812	261	4	.	.	PUNCT
fcis-14812	262	1	76	76	NUM
fcis-14812	263	1	[	[	X
fcis-14812	263	2	22	22	NUM
fcis-14812	263	3	]	]	X
fcis-14812	263	4	musgrave	musgrave	PROPN
fcis-14812	263	5	k	k	PROPN
fcis-14812	263	6	,	,	PUNCT
fcis-14812	263	7	belongie	belongie	PROPN
fcis-14812	263	8	s	s	PROPN
fcis-14812	263	9	,	,	PUNCT
fcis-14812	263	10	lim	lim	PROPN
fcis-14812	263	11	s	s	PART
fcis-14812	263	12	n.	n.	NOUN
fcis-14812	263	13	a	a	DET
fcis-14812	263	14	metric	metric	ADJ
fcis-14812	263	15	learning	learning	NOUN
fcis-14812	263	16	reality	reality	NOUN
fcis-14812	263	17	check	check	NOUN
fcis-14812	263	18	[	[	X
fcis-14812	263	19	c]//computer	c]//computer	NOUN
fcis-14812	263	20	vision	vision	NOUN
fcis-14812	263	21	–	–	PUNCT
fcis-14812	263	22	eccv	eccv	ADJ
fcis-14812	263	23	2020	2020	NUM
fcis-14812	263	24	:	:	PUNCT
fcis-14812	263	25	16th	16th	ADJ
fcis-14812	263	26	european	european	ADJ
fcis-14812	263	27	conference	conference	PROPN
fcis-14812	263	28	,	,	PUNCT
fcis-14812	263	29	glasgow	glasgow	PROPN
fcis-14812	263	30	,	,	PUNCT
fcis-14812	263	31	uk	uk	PROPN
fcis-14812	263	32	,	,	PUNCT
fcis-14812	263	33	august	august	PROPN
fcis-14812	263	34	23–28	23–28	NUM
fcis-14812	263	35	,	,	PUNCT
fcis-14812	263	36	2020	2020	NUM
fcis-14812	263	37	,	,	PUNCT
fcis-14812	263	38	proceedings	proceeding	NOUN
fcis-14812	263	39	,	,	PUNCT
fcis-14812	263	40	part	part	NOUN
fcis-14812	263	41	xxv	xxv	PROPN
fcis-14812	263	42	16	16	NUM
fcis-14812	263	43	.	.	PUNCT
fcis-14812	264	1	springer	springer	NOUN
fcis-14812	264	2	international	international	ADJ
fcis-14812	264	3	publishing	publishing	NOUN
fcis-14812	264	4	,	,	PUNCT
fcis-14812	264	5	2020	2020	NUM
fcis-14812	264	6	:	:	PUNCT
fcis-14812	264	7	681	681	NUM
fcis-14812	264	8	-	-	SYM
fcis-14812	264	9	699	699	NUM
fcis-14812	264	10	.	.	PUNCT
fcis-14812	265	1	[	[	X
fcis-14812	265	2	23	23	NUM
fcis-14812	265	3	]	]	PUNCT
fcis-14812	265	4	atallah	atallah	PROPN
fcis-14812	265	5	m	m	PROPN
fcis-14812	265	6	j.	j.	PROPN
fcis-14812	265	7	faster	fast	ADJ
fcis-14812	265	8	image	image	NOUN
fcis-14812	265	9	template	template	NOUN
fcis-14812	265	10	matching	match	VERB
fcis-14812	265	11	in	in	ADP
fcis-14812	265	12	the	the	DET
fcis-14812	265	13	sum	sum	NOUN
fcis-14812	265	14	of	of	ADP
fcis-14812	265	15	the	the	DET
fcis-14812	265	16	absolute	absolute	ADJ
fcis-14812	265	17	value	value	NOUN
fcis-14812	265	18	of	of	ADP
fcis-14812	265	19	differences	difference	NOUN
fcis-14812	265	20	measure[j	measure[j	NOUN
fcis-14812	265	21	]	]	PUNCT
fcis-14812	265	22	.	.	PUNCT
fcis-14812	266	1	ieee	ieee	NOUN
fcis-14812	266	2	transactions	transaction	NOUN
fcis-14812	266	3	on	on	ADP
fcis-14812	266	4	image	image	NOUN
fcis-14812	266	5	processing	processing	NOUN
fcis-14812	266	6	,	,	PUNCT
fcis-14812	266	7	2001	2001	NUM
fcis-14812	266	8	,	,	PUNCT
fcis-14812	266	9	10(4	10(4	NUM
fcis-14812	266	10	):	):	PUNCT
fcis-14812	266	11	659	659	NUM
fcis-14812	266	12	-	-	SYM
fcis-14812	266	13	663	663	NUM
fcis-14812	266	14	.	.	PUNCT
fcis-14812	267	1	[	[	X
fcis-14812	267	2	24	24	NUM
fcis-14812	267	3	]	]	PUNCT
fcis-14812	267	4	lewis	lewis	PROPN
fcis-14812	267	5	j	j	PROPN
fcis-14812	267	6	p.	p.	PROPN
fcis-14812	267	7	fast	fast	PROPN
fcis-14812	267	8	template	template	NOUN
fcis-14812	267	9	matching[c]//vision	matching[c]//vision	NOUN
fcis-14812	267	10	interface	interface	NOUN
fcis-14812	267	11	.	.	PUNCT
fcis-14812	268	1	1995	1995	NUM
fcis-14812	268	2	,	,	PUNCT
fcis-14812	268	3	95(120123	95(120123	NUM
fcis-14812	268	4	):	):	PUNCT
fcis-14812	268	5	15	15	NUM
fcis-14812	268	6	-	-	SYM
fcis-14812	268	7	19	19	NUM
fcis-14812	268	8	.	.	PUNCT
fcis-14812	269	1	[	[	X
fcis-14812	269	2	25	25	NUM
fcis-14812	269	3	]	]	X
fcis-14812	269	4	sun	sun	PROPN
fcis-14812	269	5	y	y	PROPN
fcis-14812	269	6	,	,	PUNCT
fcis-14812	269	7	cheng	cheng	PROPN
fcis-14812	269	8	c	c	PROPN
fcis-14812	269	9	,	,	PUNCT
fcis-14812	269	10	zhang	zhang	PROPN
fcis-14812	269	11	y	y	PROPN
fcis-14812	269	12	,	,	PUNCT
fcis-14812	269	13	et	et	PROPN
fcis-14812	269	14	al	al	PROPN
fcis-14812	269	15	.	.	PROPN
fcis-14812	269	16	circle	circle	PROPN
fcis-14812	269	17	loss	loss	PROPN
fcis-14812	269	18	:	:	PUNCT
fcis-14812	269	19	a	a	DET
fcis-14812	269	20	unified	unified	ADJ
fcis-14812	269	21	perspective	perspective	NOUN
fcis-14812	269	22	of	of	ADP
fcis-14812	269	23	pair	pair	NOUN
fcis-14812	269	24	similarity	similarity	NOUN
fcis-14812	269	25	optimization[c]//proceedings	optimization[c]//proceeding	NOUN
fcis-14812	269	26	of	of	ADP
fcis-14812	269	27	the	the	DET
fcis-14812	269	28	ieee	ieee	NOUN
fcis-14812	269	29	/	/	SYM
fcis-14812	269	30	cvf	cvf	NOUN
fcis-14812	269	31	conference	conference	NOUN
fcis-14812	269	32	on	on	ADP
fcis-14812	269	33	computer	computer	NOUN
fcis-14812	269	34	vision	vision	NOUN
fcis-14812	269	35	and	and	CCONJ
fcis-14812	269	36	pattern	pattern	NOUN
fcis-14812	269	37	recognition	recognition	NOUN
fcis-14812	269	38	.	.	PUNCT
fcis-14812	270	1	2020	2020	NUM
fcis-14812	270	2	:	:	PUNCT
fcis-14812	270	3	6398	6398	NUM
fcis-14812	270	4	-	-	SYM
fcis-14812	270	5	6407	6407	NUM
fcis-14812	270	6	.	.	PUNCT
fcis-14812	271	1	[	[	X
fcis-14812	271	2	26	26	NUM
fcis-14812	271	3	]	]	X
fcis-14812	271	4	wang	wang	PROPN
fcis-14812	271	5	h	h	PROPN
fcis-14812	271	6	,	,	PUNCT
fcis-14812	271	7	wang	wang	PROPN
fcis-14812	271	8	y	y	PROPN
fcis-14812	271	9	,	,	PUNCT
fcis-14812	271	10	zhou	zhou	PROPN
fcis-14812	271	11	z	z	PROPN
fcis-14812	271	12	,	,	PUNCT
fcis-14812	271	13	et	et	PROPN
fcis-14812	271	14	al	al	PROPN
fcis-14812	271	15	.	.	PROPN
fcis-14812	271	16	cosface	cosface	PROPN
fcis-14812	271	17	:	:	PUNCT
fcis-14812	271	18	large	large	ADJ
fcis-14812	271	19	margin	margin	NOUN
fcis-14812	271	20	cosine	cosine	NOUN
fcis-14812	271	21	loss	loss	NOUN
fcis-14812	271	22	for	for	ADP
fcis-14812	271	23	deep	deep	ADJ
fcis-14812	271	24	face	face	NOUN
fcis-14812	271	25	recognition[c]//proceedings	recognition[c]//proceeding	NOUN
fcis-14812	271	26	of	of	ADP
fcis-14812	271	27	the	the	DET
fcis-14812	271	28	ieee	ieee	NOUN
fcis-14812	271	29	conference	conference	NOUN
fcis-14812	271	30	on	on	ADP
fcis-14812	271	31	computer	computer	NOUN
fcis-14812	271	32	vision	vision	NOUN
fcis-14812	271	33	and	and	CCONJ
fcis-14812	271	34	pattern	pattern	NOUN
fcis-14812	271	35	recognition	recognition	NOUN
fcis-14812	271	36	.	.	PUNCT
fcis-14812	272	1	2018	2018	NUM
fcis-14812	272	2	:	:	PUNCT
fcis-14812	272	3	5265	5265	NUM
fcis-14812	272	4	-	-	SYM
fcis-14812	272	5	5274	5274	NUM
fcis-14812	272	6	.	.	PUNCT
fcis-14812	273	1	[	[	X
fcis-14812	273	2	27	27	NUM
fcis-14812	273	3	]	]	X
fcis-14812	273	4	hadsell	hadsell	PROPN
fcis-14812	273	5	r	r	NOUN
fcis-14812	273	6	,	,	PUNCT
fcis-14812	273	7	chopra	chopra	PROPN
fcis-14812	273	8	s	s	PART
fcis-14812	273	9	,	,	PUNCT
fcis-14812	273	10	lecun	lecun	ADJ
fcis-14812	273	11	y.	y.	NOUN
fcis-14812	273	12	dimensionality	dimensionality	NOUN
fcis-14812	273	13	reduction	reduction	NOUN
fcis-14812	273	14	by	by	ADP
fcis-14812	273	15	learning	learn	VERB
fcis-14812	273	16	an	an	DET
fcis-14812	273	17	invariant	invariant	ADJ
fcis-14812	273	18	mapping[c]//2006	mapping[c]//2006	NUM
fcis-14812	273	19	ieee	ieee	NOUN
fcis-14812	273	20	computer	computer	NOUN
fcis-14812	273	21	society	society	PROPN
fcis-14812	273	22	conference	conference	NOUN
fcis-14812	273	23	on	on	ADP
fcis-14812	273	24	computer	computer	NOUN
fcis-14812	273	25	vision	vision	NOUN
fcis-14812	273	26	and	and	CCONJ
fcis-14812	273	27	pattern	pattern	NOUN
fcis-14812	273	28	recognition	recognition	NOUN
fcis-14812	273	29	(	(	PUNCT
fcis-14812	273	30	cvpr'06	cvpr'06	PROPN
fcis-14812	273	31	)	)	PUNCT
fcis-14812	273	32	.	.	PUNCT
fcis-14812	274	1	ieee	ieee	PROPN
fcis-14812	274	2	,	,	PUNCT
fcis-14812	274	3	2006	2006	NUM
fcis-14812	274	4	,	,	PUNCT
fcis-14812	274	5	2	2	NUM
fcis-14812	274	6	:	:	SYM
fcis-14812	274	7	1735	1735	NUM
fcis-14812	274	8	-	-	SYM
fcis-14812	274	9	1742	1742	NUM
fcis-14812	274	10	.	.	PUNCT
fcis-14812	275	1	[	[	X
fcis-14812	275	2	28	28	NUM
fcis-14812	275	3	]	]	X
fcis-14812	275	4	weinberger	weinberger	PROPN
fcis-14812	275	5	k	k	PROPN
fcis-14812	275	6	q	q	PROPN
fcis-14812	275	7	,	,	PUNCT
fcis-14812	275	8	saul	saul	PROPN
fcis-14812	275	9	l	l	PROPN
fcis-14812	275	10	k.	k.	PROPN
fcis-14812	275	11	distance	distance	PROPN
fcis-14812	275	12	metric	metric	ADJ
fcis-14812	275	13	learning	learning	NOUN
fcis-14812	275	14	for	for	ADP
fcis-14812	275	15	large	large	ADJ
fcis-14812	275	16	margin	margin	NOUN
fcis-14812	275	17	nearest	near	ADJ
fcis-14812	275	18	neighbor	neighbor	NOUN
fcis-14812	275	19	classification[j	classification[j	PROPN
fcis-14812	275	20	]	]	PUNCT
fcis-14812	275	21	.	.	PUNCT
fcis-14812	276	1	journal	journal	PROPN
fcis-14812	276	2	of	of	ADP
fcis-14812	276	3	machine	machine	NOUN
fcis-14812	276	4	learning	learn	VERB
fcis-14812	276	5	research	research	NOUN
fcis-14812	276	6	,	,	PUNCT
fcis-14812	276	7	2009	2009	NUM
fcis-14812	276	8	,	,	PUNCT
fcis-14812	276	9	10(2	10(2	NUM
fcis-14812	276	10	)	)	PUNCT
fcis-14812	276	11	.	.	PUNCT
fcis-14812	277	1	[	[	X
fcis-14812	277	2	29	29	NUM
fcis-14812	277	3	]	]	X
fcis-14812	277	4	sohn	sohn	PROPN
fcis-14812	277	5	k.	k.	PROPN
fcis-14812	277	6	improved	improve	VERB
fcis-14812	277	7	deep	deep	ADJ
fcis-14812	277	8	metric	metric	ADJ
fcis-14812	277	9	learning	learning	NOUN
fcis-14812	277	10	with	with	ADP
fcis-14812	277	11	multi	multi	ADJ
fcis-14812	277	12	-	-	ADJ
fcis-14812	277	13	class	class	ADJ
fcis-14812	277	14	n	n	CCONJ
fcis-14812	277	15	-	-	PUNCT
fcis-14812	277	16	pair	pair	NOUN
fcis-14812	277	17	loss	loss	NOUN
fcis-14812	277	18	objective[j	objective[j	NOUN
fcis-14812	277	19	]	]	PUNCT
fcis-14812	277	20	.	.	PUNCT
fcis-14812	278	1	advances	advance	NOUN
fcis-14812	278	2	in	in	ADP
fcis-14812	278	3	neural	neural	ADJ
fcis-14812	278	4	information	information	NOUN
fcis-14812	278	5	processing	processing	NOUN
fcis-14812	278	6	systems	system	NOUN
fcis-14812	278	7	,	,	PUNCT
fcis-14812	278	8	2016	2016	NUM
fcis-14812	278	9	,	,	PUNCT
fcis-14812	278	10	29	29	NUM
fcis-14812	278	11	.	.	PUNCT
fcis-14812	279	1	[	[	X
fcis-14812	279	2	30	30	NUM
fcis-14812	279	3	]	]	X
fcis-14812	279	4	oord	oord	PROPN
fcis-14812	279	5	a	a	PROPN
fcis-14812	279	6	,	,	PUNCT
fcis-14812	279	7	li	li	PROPN
fcis-14812	279	8	y	y	PROPN
fcis-14812	279	9	,	,	PUNCT
fcis-14812	279	10	vinyals	vinyal	NOUN
fcis-14812	279	11	o.	o.	NOUN
fcis-14812	279	12	representation	representation	NOUN
fcis-14812	279	13	learning	learn	VERB
fcis-14812	279	14	with	with	ADP
fcis-14812	279	15	contrastive	contrastive	ADJ
fcis-14812	279	16	predictive	predictive	ADJ
fcis-14812	279	17	coding[j	coding[j	NOUN
fcis-14812	279	18	]	]	PUNCT
fcis-14812	279	19	.	.	PUNCT
fcis-14812	280	1	arxiv	arxiv	PROPN
fcis-14812	280	2	preprint	preprint	PROPN
fcis-14812	280	3	arxiv	arxiv	PROPN
fcis-14812	280	4	:	:	PUNCT
fcis-14812	280	5	1807	1807	NUM
fcis-14812	280	6	.	.	PUNCT
fcis-14812	280	7	03748	03748	NUM
fcis-14812	280	8	,	,	PUNCT
fcis-14812	280	9	2018	2018	NUM
fcis-14812	280	10	.	.	PUNCT
fcis-14812	281	1	[	[	X
fcis-14812	281	2	31	31	NUM
fcis-14812	281	3	]	]	PUNCT
fcis-14812	281	4	yuan	yuan	PROPN
fcis-14812	281	5	t	t	PROPN
fcis-14812	281	6	,	,	PUNCT
fcis-14812	281	7	deng	deng	PROPN
fcis-14812	281	8	w	w	PROPN
fcis-14812	281	9	,	,	PUNCT
fcis-14812	281	10	tang	tang	PROPN
fcis-14812	281	11	j	j	PROPN
fcis-14812	281	12	,	,	PUNCT
fcis-14812	281	13	et	et	PROPN
fcis-14812	281	14	al	al	PROPN
fcis-14812	281	15	.	.	PROPN
fcis-14812	281	16	signal	signal	PROPN
fcis-14812	281	17	-	-	PUNCT
fcis-14812	281	18	to	to	ADP
fcis-14812	281	19	-	-	PUNCT
fcis-14812	281	20	noise	noise	NOUN
fcis-14812	281	21	ratio	ratio	NOUN
fcis-14812	281	22	:	:	PUNCT
fcis-14812	281	23	a	a	DET
fcis-14812	281	24	robust	robust	ADJ
fcis-14812	281	25	distance	distance	NOUN
fcis-14812	281	26	metric	metric	ADJ
fcis-14812	281	27	for	for	ADP
fcis-14812	281	28	deep	deep	ADJ
fcis-14812	281	29	metric	metric	ADJ
fcis-14812	281	30	learning[c]//proceedings	learning[c]//proceeding	NOUN
fcis-14812	281	31	of	of	ADP
fcis-14812	281	32	the	the	DET
fcis-14812	281	33	ieee	ieee	NOUN
fcis-14812	281	34	/	/	SYM
fcis-14812	281	35	cvf	cvf	NOUN
fcis-14812	281	36	conference	conference	NOUN
fcis-14812	281	37	on	on	ADP
fcis-14812	281	38	computer	computer	NOUN
fcis-14812	281	39	vision	vision	NOUN
fcis-14812	281	40	and	and	CCONJ
fcis-14812	281	41	pattern	pattern	NOUN
fcis-14812	281	42	recognition	recognition	NOUN
fcis-14812	281	43	.	.	PUNCT
fcis-14812	282	1	2019	2019	NUM
fcis-14812	282	2	:	:	PUNCT
fcis-14812	282	3	4815	4815	NUM
fcis-14812	282	4	-	-	SYM
fcis-14812	282	5	4824	4824	NUM
fcis-14812	282	6	.	.	PUNCT
fcis-14812	283	1	[	[	X
fcis-14812	283	2	32	32	NUM
fcis-14812	283	3	]	]	SYM
fcis-14812	283	4	hu	hu	PROPN
fcis-14812	283	5	j	j	PROPN
fcis-14812	283	6	,	,	PUNCT
fcis-14812	283	7	shen	shen	PROPN
fcis-14812	283	8	l	l	PROPN
fcis-14812	283	9	,	,	PUNCT
fcis-14812	283	10	sun	sun	PROPN
fcis-14812	283	11	g.	g.	PROPN
fcis-14812	283	12	squeeze	squeeze	PROPN
fcis-14812	283	13	-	-	PUNCT
fcis-14812	283	14	and	and	CCONJ
fcis-14812	283	15	-	-	PUNCT
fcis-14812	283	16	excitation	excitation	NOUN
fcis-14812	283	17	networks	network	NOUN
fcis-14812	283	18	[	[	PUNCT
fcis-14812	283	19	c]//	c]//	ADJ
fcis-14812	283	20	proceedings	proceeding	NOUN
fcis-14812	283	21	of	of	ADP
fcis-14812	283	22	the	the	DET
fcis-14812	283	23	ieee	ieee	NOUN
fcis-14812	283	24	conference	conference	NOUN
fcis-14812	283	25	on	on	ADP
fcis-14812	283	26	computer	computer	NOUN
fcis-14812	283	27	vision	vision	NOUN
fcis-14812	283	28	and	and	CCONJ
fcis-14812	283	29	pattern	pattern	NOUN
fcis-14812	283	30	recognition	recognition	NOUN
fcis-14812	283	31	.	.	PUNCT
fcis-14812	284	1	2018	2018	NUM
fcis-14812	284	2	:	:	PUNCT
fcis-14812	284	3	7132	7132	NUM
fcis-14812	284	4	-	-	SYM
fcis-14812	284	5	7141	7141	NUM
fcis-14812	284	6	.	.	PUNCT