id	sid	tid	token	lemma	pos
wh246q20m97	1	1	biometric	biometric	ADJ
wh246q20m97	1	2	verification	verification	NOUN
wh246q20m97	1	3	systems	system	NOUN
wh246q20m97	1	4	employing	employ	VERB
wh246q20m97	1	5	images	image	NOUN
wh246q20m97	1	6	of	of	ADP
wh246q20m97	1	7	the	the	DET
wh246q20m97	1	8	iris	iris	NOUN
wh246q20m97	1	9	are	be	AUX
wh246q20m97	1	10	claimed	claim	VERB
wh246q20m97	1	11	to	to	PART
wh246q20m97	1	12	be	be	AUX
wh246q20m97	1	13	extremely	extremely	ADV
wh246q20m97	1	14	accurate	accurate	ADJ
wh246q20m97	1	15	,	,	PUNCT
wh246q20m97	1	16	yielding	yield	VERB
wh246q20m97	1	17	no	no	DET
wh246q20m97	1	18	false	false	ADJ
wh246q20m97	1	19	accepts	accept	NOUN
wh246q20m97	1	20	at	at	ADP
wh246q20m97	1	21	any	any	DET
wh246q20m97	1	22	reasonable	reasonable	ADJ
wh246q20m97	1	23	false	false	ADJ
wh246q20m97	1	24	reject	reject	ADJ
wh246q20m97	1	25	rate	rate	NOUN
wh246q20m97	1	26	.	.	PUNCT
wh246q20m97	2	1	however	however	ADV
wh246q20m97	2	2	,	,	PUNCT
wh246q20m97	2	3	there	there	PRON
wh246q20m97	2	4	are	be	VERB
wh246q20m97	2	5	few	few	ADJ
wh246q20m97	2	6	if	if	SCONJ
wh246q20m97	2	7	any	any	DET
wh246q20m97	2	8	large	large	ADJ
wh246q20m97	2	9	scale	scale	NOUN
wh246q20m97	2	10	experimental	experimental	ADJ
wh246q20m97	2	11	evaluations	evaluation	NOUN
wh246q20m97	2	12	on	on	ADP
wh246q20m97	2	13	public	public	ADJ
wh246q20m97	2	14	iris	iris	NOUN
wh246q20m97	2	15	datasets	dataset	NOUN
wh246q20m97	2	16	reported	report	VERB
wh246q20m97	2	17	in	in	ADP
wh246q20m97	2	18	the	the	DET
wh246q20m97	2	19	literature	literature	NOUN
wh246q20m97	2	20	.	.	PUNCT
wh246q20m97	3	1	we	we	PRON
wh246q20m97	3	2	have	have	AUX
wh246q20m97	3	3	collected	collect	VERB
wh246q20m97	3	4	an	an	DET
wh246q20m97	3	5	iris	iris	PROPN
wh246q20m97	3	6	image	image	NOUN
wh246q20m97	3	7	dataset	dataset	NOUN
wh246q20m97	3	8	of	of	ADP
wh246q20m97	3	9	over	over	ADP
wh246q20m97	3	10	25,000	25,000	NUM
wh246q20m97	3	11	iris	iris	NOUN
wh246q20m97	3	12	images	image	NOUN
wh246q20m97	3	13	from	from	ADP
wh246q20m97	3	14	over	over	ADP
wh246q20m97	3	15	300	300	NUM
wh246q20m97	3	16	persons	person	NOUN
wh246q20m97	3	17	(	(	PUNCT
wh246q20m97	3	18	over	over	ADP
wh246q20m97	3	19	600	600	NUM
wh246q20m97	3	20	irises	iris	NOUN
wh246q20m97	3	21	)	)	PUNCT
wh246q20m97	3	22	.	.	PUNCT
wh246q20m97	4	1	when	when	SCONJ
wh246q20m97	4	2	collecting	collect	VERB
wh246q20m97	4	3	the	the	DET
wh246q20m97	4	4	dataset	dataset	NOUN
wh246q20m97	4	5	,	,	PUNCT
wh246q20m97	4	6	we	we	PRON
wh246q20m97	4	7	intentionally	intentionally	ADV
wh246q20m97	4	8	allowed	allow	VERB
wh246q20m97	4	9	a	a	DET
wh246q20m97	4	10	broader	broad	ADJ
wh246q20m97	4	11	image	image	NOUN
wh246q20m97	4	12	quality	quality	NOUN
wh246q20m97	4	13	range	range	NOUN
wh246q20m97	4	14	than	than	ADP
wh246q20m97	4	15	that	that	PRON
wh246q20m97	4	16	allowed	allow	VERB
wh246q20m97	4	17	by	by	ADP
wh246q20m97	4	18	default	default	NOUN
wh246q20m97	4	19	in	in	ADP
wh246q20m97	4	20	current	current	ADJ
wh246q20m97	4	21	commercial	commercial	ADJ
wh246q20m97	4	22	iris	iris	PROPN
wh246q20m97	4	23	recognition	recognition	PROPN
wh246q20m97	4	24	systems	system	NOUN
wh246q20m97	4	25	.	.	PUNCT
wh246q20m97	5	1	the	the	DET
wh246q20m97	5	2	iris	iris	PROPN
wh246q20m97	5	3	images	image	NOUN
wh246q20m97	5	4	used	use	VERB
wh246q20m97	5	5	in	in	ADP
wh246q20m97	5	6	our	our	PRON
wh246q20m97	5	7	experiments	experiment	NOUN
wh246q20m97	5	8	have	have	AUX
wh246q20m97	5	9	been	be	AUX
wh246q20m97	5	10	,	,	PUNCT
wh246q20m97	5	11	or	or	CCONJ
wh246q20m97	5	12	will	will	AUX
wh246q20m97	5	13	be	be	AUX
wh246q20m97	5	14	,	,	PUNCT
wh246q20m97	5	15	released	release	VERB
wh246q20m97	5	16	as	as	ADP
wh246q20m97	5	17	part	part	NOUN
wh246q20m97	5	18	of	of	ADP
wh246q20m97	5	19	the	the	DET
wh246q20m97	5	20	iris	iris	PROPN
wh246q20m97	5	21	challenge	challenge	NOUN
wh246q20m97	5	22	evaluation	evaluation	NOUN
wh246q20m97	5	23	(	(	PUNCT
wh246q20m97	5	24	ice	ice	NOUN
wh246q20m97	5	25	)	)	PUNCT
wh246q20m97	5	26	.	.	PUNCT
wh246q20m97	6	1	we	we	PRON
wh246q20m97	6	2	reimplemented	reimplemente	VERB
wh246q20m97	6	3	in	in	ADP
wh246q20m97	6	4	c	c	PROPN
wh246q20m97	6	5	an	an	DET
wh246q20m97	6	6	open	open	ADJ
wh246q20m97	6	7	source	source	NOUN
wh246q20m97	6	8	iris	iris	PROPN
wh246q20m97	6	9	recognition	recognition	NOUN
wh246q20m97	6	10	system	system	NOUN
wh246q20m97	6	11	,	,	PUNCT
wh246q20m97	6	12	which	which	PRON
wh246q20m97	6	13	was	be	AUX
wh246q20m97	6	14	originally	originally	ADV
wh246q20m97	6	15	implemented	implement	VERB
wh246q20m97	6	16	in	in	ADP
wh246q20m97	6	17	matlab	matlab	NOUN
wh246q20m97	6	18	by	by	ADP
wh246q20m97	6	19	libor	libor	PROPN
wh246q20m97	6	20	masek	masek	PROPN
wh246q20m97	6	21	.	.	PUNCT
wh246q20m97	7	1	the	the	DET
wh246q20m97	7	2	ice	ice	NOUN
wh246q20m97	7	3	baseline	baseline	NOUN
wh246q20m97	7	4	is	be	AUX
wh246q20m97	7	5	a	a	DET
wh246q20m97	7	6	c++	c++	NOUN
wh246q20m97	7	7	translation	translation	NOUN
wh246q20m97	7	8	of	of	ADP
wh246q20m97	7	9	our	our	PRON
wh246q20m97	7	10	c	c	NOUN
wh246q20m97	7	11	re	re	NOUN
wh246q20m97	7	12	-	-	NOUN
wh246q20m97	7	13	implementation	implementation	NOUN
wh246q20m97	7	14	with	with	ADP
wh246q20m97	7	15	modifications	modification	NOUN
wh246q20m97	7	16	for	for	ADP
wh246q20m97	7	17	optimization	optimization	NOUN
wh246q20m97	7	18	in	in	ADP
wh246q20m97	7	19	speed	speed	NOUN
wh246q20m97	7	20	and	and	CCONJ
wh246q20m97	7	21	memory	memory	NOUN
wh246q20m97	7	22	usage	usage	NOUN
wh246q20m97	7	23	.	.	PUNCT
wh246q20m97	8	1	we	we	PRON
wh246q20m97	8	2	evaluated	evaluate	VERB
wh246q20m97	8	3	the	the	DET
wh246q20m97	8	4	effects	effect	NOUN
wh246q20m97	8	5	of	of	ADP
wh246q20m97	8	6	iris	iris	PROPN
wh246q20m97	8	7	image	image	NOUN
wh246q20m97	8	8	quality	quality	NOUN
wh246q20m97	8	9	by	by	ADP
wh246q20m97	8	10	using	use	VERB
wh246q20m97	8	11	the	the	DET
wh246q20m97	8	12	ice	ice	NOUN
wh246q20m97	8	13	baseline	baseline	NOUN
wh246q20m97	8	14	system	system	NOUN
wh246q20m97	8	15	on	on	ADP
wh246q20m97	8	16	our	our	PRON
wh246q20m97	8	17	iris	iris	PROPN
wh246q20m97	8	18	dataset	dataset	NOUN
wh246q20m97	8	19	.	.	PUNCT
wh246q20m97	9	1	we	we	PRON
wh246q20m97	9	2	have	have	AUX
wh246q20m97	9	3	implemented	implement	VERB
wh246q20m97	9	4	an	an	DET
wh246q20m97	9	5	improved	improved	ADJ
wh246q20m97	9	6	iris	iris	NOUN
wh246q20m97	9	7	segmentation	segmentation	NOUN
wh246q20m97	9	8	and	and	CCONJ
wh246q20m97	9	9	eyelid	eyelid	ADJ
wh246q20m97	9	10	detection	detection	NOUN
wh246q20m97	9	11	stage	stage	NOUN
wh246q20m97	9	12	compared	compare	VERB
wh246q20m97	9	13	to	to	ADP
wh246q20m97	9	14	the	the	DET
wh246q20m97	9	15	ice	ice	NOUN
wh246q20m97	9	16	baseline	baseline	NOUN
wh246q20m97	9	17	code	code	NOUN
wh246q20m97	9	18	,	,	PUNCT
wh246q20m97	9	19	and	and	CCONJ
wh246q20m97	9	20	experimentally	experimentally	ADV
wh246q20m97	9	21	verified	verify	VERB
wh246q20m97	9	22	an	an	DET
wh246q20m97	9	23	improvement	improvement	NOUN
wh246q20m97	9	24	in	in	ADP
wh246q20m97	9	25	both	both	DET
wh246q20m97	9	26	the	the	DET
wh246q20m97	9	27	verification	verification	NOUN
wh246q20m97	9	28	and	and	CCONJ
wh246q20m97	9	29	identification	identification	NOUN
wh246q20m97	9	30	contexts	context	NOUN
wh246q20m97	9	31	.	.	PUNCT
wh246q20m97	10	1	replacing	replace	VERB
wh246q20m97	10	2	the	the	DET
wh246q20m97	10	3	ice	ice	NOUN
wh246q20m97	10	4	baseline	baseline	NOUN
wh246q20m97	10	5	segmentation	segmentation	NOUN
wh246q20m97	10	6	with	with	ADP
wh246q20m97	10	7	our	our	PRON
wh246q20m97	10	8	improved	improved	ADJ
wh246q20m97	10	9	segmentation	segmentation	NOUN
wh246q20m97	10	10	algorithm	algorithm	NOUN
wh246q20m97	10	11	,	,	PUNCT
wh246q20m97	10	12	and	and	CCONJ
wh246q20m97	10	13	keeping	keep	VERB
wh246q20m97	10	14	other	other	ADJ
wh246q20m97	10	15	modules	module	NOUN
wh246q20m97	10	16	of	of	ADP
wh246q20m97	10	17	the	the	DET
wh246q20m97	10	18	ice	ice	NOUN
wh246q20m97	10	19	baseline	baseline	NOUN
wh246q20m97	10	20	the	the	DET
wh246q20m97	10	21	same	same	ADJ
wh246q20m97	10	22	,	,	PUNCT
wh246q20m97	10	23	leads	lead	VERB
wh246q20m97	10	24	to	to	ADP
wh246q20m97	10	25	an	an	DET
wh246q20m97	10	26	increase	increase	NOUN
wh246q20m97	10	27	of	of	ADP
wh246q20m97	10	28	over	over	ADP
wh246q20m97	10	29	6	6	NUM
wh246q20m97	10	30	%	%	NOUN
wh246q20m97	10	31	in	in	ADP
wh246q20m97	10	32	the	the	DET
wh246q20m97	10	33	rank	rank	NOUN
wh246q20m97	10	34	-	-	PUNCT
wh246q20m97	10	35	one	one	NUM
wh246q20m97	10	36	recognition	recognition	NOUN
wh246q20m97	10	37	rate	rate	NOUN
wh246q20m97	10	38	and	and	CCONJ
wh246q20m97	10	39	a	a	DET
wh246q20m97	10	40	decrease	decrease	NOUN
wh246q20m97	10	41	of	of	ADP
wh246q20m97	10	42	over	over	ADP
wh246q20m97	10	43	4	4	NUM
wh246q20m97	10	44	%	%	NOUN
wh246q20m97	10	45	in	in	ADP
wh246q20m97	10	46	the	the	DET
wh246q20m97	10	47	equal	equal	ADJ
wh246q20m97	10	48	error	error	NOUN
wh246q20m97	10	49	rate	rate	NOUN
wh246q20m97	10	50	.	.	PUNCT
wh246q20m97	11	1	we	we	PRON
wh246q20m97	11	2	utilized	utilize	VERB
wh246q20m97	11	3	an	an	DET
wh246q20m97	11	4	active	active	ADJ
wh246q20m97	11	5	contour	contour	NOUN
wh246q20m97	11	6	model	model	NOUN
wh246q20m97	11	7	to	to	PART
wh246q20m97	11	8	refine	refine	VERB
wh246q20m97	11	9	the	the	DET
wh246q20m97	11	10	noise	noise	NOUN
wh246q20m97	11	11	detection	detection	NOUN
wh246q20m97	11	12	results	result	NOUN
wh246q20m97	11	13	and	and	CCONJ
wh246q20m97	11	14	optimized	optimize	VERB
wh246q20m97	11	15	the	the	DET
wh246q20m97	11	16	matching	matching	NOUN
wh246q20m97	11	17	stage	stage	NOUN
wh246q20m97	11	18	to	to	PART
wh246q20m97	11	19	compensate	compensate	VERB
wh246q20m97	11	20	for	for	ADP
wh246q20m97	11	21	the	the	DET
wh246q20m97	11	22	possible	possible	ADJ
wh246q20m97	11	23	inaccuracy	inaccuracy	NOUN
wh246q20m97	11	24	in	in	ADP
wh246q20m97	11	25	iris	iris	NOUN
wh246q20m97	11	26	segmentation	segmentation	NOUN
wh246q20m97	11	27	and	and	CCONJ
wh246q20m97	11	28	noise	noise	NOUN
wh246q20m97	11	29	detection	detection	NOUN
wh246q20m97	11	30	,	,	PUNCT
wh246q20m97	11	31	which	which	PRON
wh246q20m97	11	32	leads	lead	VERB
wh246q20m97	11	33	to	to	ADP
wh246q20m97	11	34	another	another	DET
wh246q20m97	11	35	0.95	0.95	NUM
wh246q20m97	11	36	%	%	NOUN
wh246q20m97	11	37	increase	increase	NOUN
wh246q20m97	11	38	in	in	ADP
wh246q20m97	11	39	the	the	DET
wh246q20m97	11	40	rank	rank	NOUN
wh246q20m97	11	41	one	one	NUM
wh246q20m97	11	42	recognition	recognition	NOUN
wh246q20m97	11	43	rate	rate	NOUN
wh246q20m97	11	44	and	and	CCONJ
wh246q20m97	11	45	0.85	0.85	NUM
wh246q20m97	11	46	%	%	NOUN
wh246q20m97	11	47	decrease	decrease	NOUN
wh246q20m97	11	48	in	in	ADP
wh246q20m97	11	49	the	the	DET
wh246q20m97	11	50	equal	equal	ADJ
wh246q20m97	11	51	error	error	NOUN
wh246q20m97	11	52	rate	rate	NOUN
wh246q20m97	11	53	.	.	PUNCT
wh246q20m97	12	1	this	this	DET
wh246q20m97	12	2	research	research	NOUN
wh246q20m97	12	3	demonstrates	demonstrate	VERB
wh246q20m97	12	4	that	that	SCONJ
wh246q20m97	12	5	a	a	DET
wh246q20m97	12	6	more	more	ADV
wh246q20m97	12	7	accurate	accurate	ADJ
wh246q20m97	12	8	iris	iris	NOUN
wh246q20m97	12	9	segmentation	segmentation	NOUN
wh246q20m97	12	10	helps	help	VERB
wh246q20m97	12	11	to	to	PART
wh246q20m97	12	12	improve	improve	VERB
wh246q20m97	12	13	the	the	DET
wh246q20m97	12	14	overall	overall	ADJ
wh246q20m97	12	15	system	system	NOUN
wh246q20m97	12	16	performance	performance	NOUN
wh246q20m97	12	17	,	,	PUNCT
wh246q20m97	12	18	and	and	CCONJ
wh246q20m97	12	19	that	that	SCONJ
wh246q20m97	12	20	the	the	DET
wh246q20m97	12	21	inaccuracy	inaccuracy	NOUN
wh246q20m97	12	22	of	of	ADP
wh246q20m97	12	23	iris	iris	PROPN
wh246q20m97	12	24	segmentation	segmentation	NOUN
wh246q20m97	12	25	and	and	CCONJ
wh246q20m97	12	26	noise	noise	NOUN
wh246q20m97	12	27	detection	detection	NOUN
wh246q20m97	12	28	could	could	AUX
wh246q20m97	12	29	be	be	AUX
wh246q20m97	12	30	partly	partly	ADV
wh246q20m97	12	31	compensated	compensate	VERB
wh246q20m97	12	32	for	for	ADP
wh246q20m97	12	33	with	with	ADP
wh246q20m97	12	34	optimizations	optimization	NOUN
wh246q20m97	12	35	in	in	ADP
wh246q20m97	12	36	the	the	DET
wh246q20m97	12	37	matching	matching	NOUN
wh246q20m97	12	38	stage	stage	NOUN
wh246q20m97	12	39	.	.	PUNCT