id	sid	tid	token	lemma	pos
cana-5106	1	1	paper	paper	NOUN
cana-5106	1	2	title	title	NOUN
cana-5106	1	3	(	(	PUNCT
cana-5106	1	4	use	use	NOUN
cana-5106	1	5	style	style	NOUN
cana-5106	1	6	:	:	PUNCT
cana-5106	1	7	paper	paper	NOUN
cana-5106	1	8	title	title	NOUN
cana-5106	1	9	)	)	PUNCT
cana-5106	1	10	communications	communication	NOUN
cana-5106	1	11	on	on	ADP
cana-5106	1	12	applied	apply	VERB
cana-5106	1	13	nonlinear	nonlinear	ADJ
cana-5106	1	14	analysis	analysis	NOUN
cana-5106	1	15	issn	issn	NOUN
cana-5106	1	16	:	:	PUNCT
cana-5106	1	17	1074	1074	NUM
cana-5106	1	18	-	-	PUNCT
cana-5106	1	19	133x	133x	NUM
cana-5106	1	20	vol	vol	VERB
cana-5106	1	21	32	32	NUM
cana-5106	1	22	no	no	NOUN
cana-5106	1	23	.	.	PUNCT
cana-5106	2	1	10s	10	NOUN
cana-5106	2	2	(	(	PUNCT
cana-5106	2	3	2025	2025	NUM
cana-5106	2	4	)	)	PUNCT
cana-5106	2	5	1120	1120	NUM
cana-5106	2	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	2	7	smart	smart	ADJ
cana-5106	2	8	caption	caption	NOUN
cana-5106	2	9	:	:	PUNCT
cana-5106	2	10	intelligent	intelligent	ADJ
cana-5106	2	11	image	image	NOUN
cana-5106	2	12	description	description	NOUN
cana-5106	2	13	using	use	VERB
cana-5106	2	14	transformer	transformer	NOUN
cana-5106	2	15	decoder	decoder	NOUN
cana-5106	2	16	1ms.deepali	1ms.deepali	NUM
cana-5106	2	17	govindrao	govindrao	NOUN
cana-5106	2	18	navalkar	navalkar	NOUN
cana-5106	2	19	,	,	PUNCT
cana-5106	2	20	2dr	2dr	ADJ
cana-5106	2	21	,	,	PUNCT
cana-5106	2	22	rajesh	rajesh	PROPN
cana-5106	2	23	v.dahibhate	v.dahibhate	NOUN
cana-5106	2	24	,	,	PUNCT
cana-5106	2	25	3prof.(dr	3prof.(dr	NUM
cana-5106	2	26	.	.	PUNCT
cana-5106	2	27	)	)	PUNCT
cana-5106	3	1	nilesh	nilesh	PROPN
cana-5106	3	2	.r.wankhade	.r.wankhade	PROPN
cana-5106	3	3	,	,	PUNCT
cana-5106	3	4	4ms.shubhangi	4ms.shubhangi	NUM
cana-5106	3	5	g	g	PROPN
cana-5106	3	6	navalkar	navalkar	NOUN
cana-5106	3	7	1me	1me	ADJ
cana-5106	3	8	student	student	NOUN
cana-5106	3	9	,	,	PUNCT
cana-5106	3	10	computer	computer	NOUN
cana-5106	3	11	engineering	engineering	NOUN
cana-5106	3	12	,	,	PUNCT
cana-5106	4	1	late	late	ADJ
cana-5106	4	2	g.	g.	PROPN
cana-5106	4	3	n.sapkal	n.sapkal	PROPN
cana-5106	4	4	college	college	PROPN
cana-5106	4	5	of	of	ADP
cana-5106	4	6	engineering	engineering	NOUN
cana-5106	4	7	,	,	PUNCT
cana-5106	4	8	nashik	nashik	PROPN
cana-5106	4	9	,	,	PUNCT
cana-5106	4	10	india	india	PROPN
cana-5106	4	11	deepali.navalkar09@gmail.com	deepali.navalkar09@gmail.com	PROPN
cana-5106	4	12	2principal	2principal	NUM
cana-5106	4	13	,	,	PUNCT
cana-5106	4	14	shree	shree	PROPN
cana-5106	4	15	kapildhara	kapildhara	PROPN
cana-5106	4	16	college	college	PROPN
cana-5106	4	17	,	,	PUNCT
cana-5106	4	18	nashik	nashik	PROPN
cana-5106	4	19	,	,	PUNCT
cana-5106	4	20	india	india	PROPN
cana-5106	4	21	dahibhaterv@gmail.com	dahibhaterv@gmail.com	X
cana-5106	5	1	3hod	3hod	NUM
cana-5106	5	2	computer	computer	NOUN
cana-5106	5	3	engineering	engineering	NOUN
cana-5106	5	4	,	,	PUNCT
cana-5106	5	5	late	late	ADJ
cana-5106	5	6	g.	g.	PROPN
cana-5106	5	7	n.sapkal	n.sapkal	PROPN
cana-5106	5	8	college	college	PROPN
cana-5106	5	9	of	of	ADP
cana-5106	5	10	engineering	engineering	NOUN
cana-5106	5	11	,	,	PUNCT
cana-5106	5	12	nashik	nashik	PROPN
cana-5106	5	13	,	,	PUNCT
cana-5106	5	14	india	india	PROPN
cana-5106	5	15	nileshrw_2000@yahoo.com	nileshrw_2000@yahoo.com	X
cana-5106	5	16	4m.e	4m.e	NUM
cana-5106	5	17	electrical	electrical	ADJ
cana-5106	5	18	power	power	NOUN
cana-5106	5	19	system	system	NOUN
cana-5106	5	20	nawalkar.shubhangi@gmail.com	nawalkar.shubhangi@gmail.com	NOUN
cana-5106	5	21	article	article	NOUN
cana-5106	5	22	history	history	NOUN
cana-5106	5	23	:	:	PUNCT
cana-5106	5	24	received	receive	VERB
cana-5106	5	25	:	:	PUNCT
cana-5106	5	26	12	12	NUM
cana-5106	5	27	-	-	SYM
cana-5106	5	28	01	01	NUM
cana-5106	5	29	-	-	PUNCT
cana-5106	5	30	2025	2025	NUM
cana-5106	5	31	revised	revise	VERB
cana-5106	5	32	:	:	PUNCT
cana-5106	5	33	25	25	NUM
cana-5106	5	34	-	-	PUNCT
cana-5106	5	35	02	02	NUM
cana-5106	5	36	-	-	PUNCT
cana-5106	5	37	2025	2025	NUM
cana-5106	5	38	accepted	accept	VERB
cana-5106	5	39	:	:	PUNCT
cana-5106	5	40	05	05	NUM
cana-5106	5	41	-	-	PUNCT
cana-5106	5	42	03	03	NUM
cana-5106	5	43	-	-	PUNCT
cana-5106	5	44	2025	2025	NUM
cana-5106	5	45	abstract	abstract	NOUN
cana-5106	5	46	:	:	PUNCT
cana-5106	5	47	the	the	DET
cana-5106	5	48	demand	demand	NOUN
cana-5106	5	49	for	for	ADP
cana-5106	5	50	automated	automate	VERB
cana-5106	5	51	image	image	NOUN
cana-5106	5	52	description	description	NOUN
cana-5106	5	53	systems	system	NOUN
cana-5106	5	54	has	have	AUX
cana-5106	5	55	surged	surge	VERB
cana-5106	5	56	with	with	ADP
cana-5106	5	57	the	the	DET
cana-5106	5	58	exponential	exponential	ADJ
cana-5106	5	59	growth	growth	NOUN
cana-5106	5	60	of	of	ADP
cana-5106	5	61	multimedia	multimedia	NOUN
cana-5106	5	62	content	content	NOUN
cana-5106	5	63	across	across	ADP
cana-5106	5	64	various	various	ADJ
cana-5106	5	65	domains	domain	NOUN
cana-5106	5	66	,	,	PUNCT
cana-5106	5	67	including	include	VERB
cana-5106	5	68	social	social	ADJ
cana-5106	5	69	media	medium	NOUN
cana-5106	5	70	,	,	PUNCT
cana-5106	5	71	digital	digital	ADJ
cana-5106	5	72	libraries	library	NOUN
cana-5106	5	73	,	,	PUNCT
cana-5106	5	74	and	and	CCONJ
cana-5106	5	75	accessibility	accessibility	NOUN
cana-5106	5	76	technologies	technology	NOUN
cana-5106	5	77	.	.	PUNCT
cana-5106	6	1	smart	smart	ADJ
cana-5106	6	2	caption	caption	NOUN
cana-5106	6	3	introduces	introduce	VERB
cana-5106	6	4	an	an	DET
cana-5106	6	5	innovative	innovative	ADJ
cana-5106	6	6	method	method	NOUN
cana-5106	6	7	for	for	ADP
cana-5106	6	8	generating	generate	VERB
cana-5106	6	9	intelligent	intelligent	ADJ
cana-5106	6	10	,	,	PUNCT
cana-5106	6	11	context	context	NOUN
cana-5106	6	12	-	-	PUNCT
cana-5106	6	13	aware	aware	ADJ
cana-5106	6	14	image	image	NOUN
cana-5106	6	15	descriptions	description	NOUN
cana-5106	6	16	by	by	ADP
cana-5106	6	17	integrating	integrate	VERB
cana-5106	6	18	vision	vision	NOUN
cana-5106	6	19	transformers	transformer	NOUN
cana-5106	6	20	(	(	PUNCT
cana-5106	6	21	vits	vit	NOUN
cana-5106	6	22	)	)	PUNCT
cana-5106	6	23	and	and	CCONJ
cana-5106	6	24	natural	natural	ADJ
cana-5106	6	25	language	language	NOUN
cana-5106	6	26	processing	processing	NOUN
cana-5106	6	27	(	(	PUNCT
cana-5106	6	28	nlp	nlp	NOUN
cana-5106	6	29	)	)	PUNCT
cana-5106	6	30	transformer	transformer	NOUN
cana-5106	6	31	decoders.unlike	decoders.unlike	INTJ
cana-5106	6	32	traditional	traditional	ADJ
cana-5106	6	33	convolutional	convolutional	ADJ
cana-5106	6	34	approaches	approach	NOUN
cana-5106	6	35	,	,	PUNCT
cana-5106	6	36	vision	vision	NOUN
cana-5106	6	37	transformers	transformer	NOUN
cana-5106	6	38	process	process	VERB
cana-5106	6	39	images	image	NOUN
cana-5106	6	40	as	as	ADP
cana-5106	6	41	sequences	sequence	NOUN
cana-5106	6	42	of	of	ADP
cana-5106	6	43	patches	patch	NOUN
cana-5106	6	44	,	,	PUNCT
cana-5106	6	45	allowing	allow	VERB
cana-5106	6	46	for	for	ADP
cana-5106	6	47	a	a	DET
cana-5106	6	48	more	more	ADV
cana-5106	6	49	comprehensive	comprehensive	ADJ
cana-5106	6	50	capture	capture	NOUN
cana-5106	6	51	of	of	ADP
cana-5106	6	52	global	global	ADJ
cana-5106	6	53	image	image	NOUN
cana-5106	6	54	features	feature	NOUN
cana-5106	6	55	.	.	PUNCT
cana-5106	7	1	these	these	DET
cana-5106	7	2	extracted	extract	VERB
cana-5106	7	3	visual	visual	ADJ
cana-5106	7	4	representations	representation	NOUN
cana-5106	7	5	are	be	AUX
cana-5106	7	6	then	then	ADV
cana-5106	7	7	fed	feed	VERB
cana-5106	7	8	into	into	ADP
cana-5106	7	9	a	a	DET
cana-5106	7	10	transformerbased	transformerbase	VERB
cana-5106	7	11	decoder	decoder	NOUN
cana-5106	7	12	,	,	PUNCT
cana-5106	7	13	which	which	PRON
cana-5106	7	14	generates	generate	VERB
cana-5106	7	15	fluent	fluent	ADJ
cana-5106	7	16	and	and	CCONJ
cana-5106	7	17	contextually	contextually	ADV
cana-5106	7	18	accurate	accurate	ADJ
cana-5106	7	19	captions	caption	NOUN
cana-5106	7	20	,	,	PUNCT
cana-5106	7	21	effectively	effectively	ADV
cana-5106	7	22	bridging	bridge	VERB
cana-5106	7	23	the	the	DET
cana-5106	7	24	gap	gap	NOUN
cana-5106	7	25	between	between	ADP
cana-5106	7	26	image	image	NOUN
cana-5106	7	27	recognition	recognition	NOUN
cana-5106	7	28	and	and	CCONJ
cana-5106	7	29	natural	natural	ADJ
cana-5106	7	30	language	language	NOUN
cana-5106	7	31	generation.our	generation.our	NOUN
cana-5106	7	32	approach	approach	NOUN
cana-5106	7	33	capitalizes	capitalize	VERB
cana-5106	7	34	on	on	ADP
cana-5106	7	35	the	the	DET
cana-5106	7	36	attention	attention	NOUN
cana-5106	7	37	mechanisms	mechanism	NOUN
cana-5106	7	38	inherent	inherent	ADJ
cana-5106	7	39	in	in	ADP
cana-5106	7	40	both	both	DET
cana-5106	7	41	vision	vision	NOUN
cana-5106	7	42	and	and	CCONJ
cana-5106	7	43	nlp	nlp	NOUN
cana-5106	7	44	transformers	transformer	NOUN
cana-5106	7	45	to	to	PART
cana-5106	7	46	enhance	enhance	VERB
cana-5106	7	47	the	the	DET
cana-5106	7	48	quality	quality	NOUN
cana-5106	7	49	of	of	ADP
cana-5106	7	50	image	image	NOUN
cana-5106	7	51	descriptions	description	NOUN
cana-5106	7	52	.	.	PUNCT
cana-5106	8	1	the	the	DET
cana-5106	8	2	vit	vit	ADJ
cana-5106	8	3	architecture	architecture	NOUN
cana-5106	8	4	selectively	selectively	ADV
cana-5106	8	5	focuses	focus	VERB
cana-5106	8	6	on	on	ADP
cana-5106	8	7	the	the	DET
cana-5106	8	8	most	most	ADV
cana-5106	8	9	relevant	relevant	ADJ
cana-5106	8	10	regions	region	NOUN
cana-5106	8	11	within	within	ADP
cana-5106	8	12	an	an	DET
cana-5106	8	13	image	image	NOUN
cana-5106	8	14	,	,	PUNCT
cana-5106	8	15	while	while	SCONJ
cana-5106	8	16	the	the	DET
cana-5106	8	17	nlp	nlp	NOUN
cana-5106	8	18	transformer	transformer	NOUN
cana-5106	8	19	decoder	decoder	NOUN
cana-5106	8	20	generates	generate	VERB
cana-5106	8	21	coherent	coherent	ADJ
cana-5106	8	22	and	and	CCONJ
cana-5106	8	23	detailed	detailed	ADJ
cana-5106	8	24	descriptions	description	NOUN
cana-5106	8	25	by	by	ADP
cana-5106	8	26	prioritizing	prioritize	VERB
cana-5106	8	27	key	key	ADJ
cana-5106	8	28	visual	visual	ADJ
cana-5106	8	29	features	feature	NOUN
cana-5106	8	30	.	.	PUNCT
cana-5106	9	1	this	this	DET
cana-5106	9	2	two	two	NUM
cana-5106	9	3	-	-	PUNCT
cana-5106	9	4	stage	stage	NOUN
cana-5106	9	5	process	process	NOUN
cana-5106	9	6	improves	improve	VERB
cana-5106	9	7	object	object	NOUN
cana-5106	9	8	detection	detection	NOUN
cana-5106	9	9	,	,	PUNCT
cana-5106	9	10	relationship	relationship	NOUN
cana-5106	9	11	understanding	understanding	NOUN
cana-5106	9	12	,	,	PUNCT
cana-5106	9	13	and	and	CCONJ
cana-5106	9	14	scene	scene	NOUN
cana-5106	9	15	interpretation	interpretation	NOUN
cana-5106	9	16	,	,	PUNCT
cana-5106	9	17	leading	lead	VERB
cana-5106	9	18	to	to	ADP
cana-5106	9	19	highly	highly	ADV
cana-5106	9	20	precise	precise	ADJ
cana-5106	9	21	and	and	CCONJ
cana-5106	9	22	contextually	contextually	ADV
cana-5106	9	23	relevant	relevant	ADJ
cana-5106	9	24	captions.experimental	captions.experimental	ADJ
cana-5106	9	25	results	result	NOUN
cana-5106	9	26	demonstrate	demonstrate	VERB
cana-5106	9	27	that	that	SCONJ
cana-5106	9	28	smart	smart	ADJ
cana-5106	9	29	caption	caption	NOUN
cana-5106	9	30	outperforms	outperform	VERB
cana-5106	9	31	conventional	conventional	ADJ
cana-5106	9	32	methods	method	NOUN
cana-5106	9	33	in	in	ADP
cana-5106	9	34	terms	term	NOUN
cana-5106	9	35	of	of	ADP
cana-5106	9	36	accuracy	accuracy	NOUN
cana-5106	9	37	,	,	PUNCT
cana-5106	9	38	contextual	contextual	ADJ
cana-5106	9	39	relevance	relevance	NOUN
cana-5106	9	40	,	,	PUNCT
cana-5106	9	41	and	and	CCONJ
cana-5106	9	42	fluency	fluency	NOUN
cana-5106	9	43	,	,	PUNCT
cana-5106	9	44	marking	mark	VERB
cana-5106	9	45	a	a	DET
cana-5106	9	46	significant	significant	ADJ
cana-5106	9	47	advancement	advancement	NOUN
cana-5106	9	48	in	in	ADP
cana-5106	9	49	automated	automate	VERB
cana-5106	9	50	image	image	NOUN
cana-5106	9	51	captioning	captioning	NOUN
cana-5106	9	52	.	.	PUNCT
cana-5106	10	1	this	this	DET
cana-5106	10	2	research	research	NOUN
cana-5106	10	3	aims	aim	VERB
cana-5106	10	4	to	to	PART
cana-5106	10	5	deliver	deliver	VERB
cana-5106	10	6	a	a	DET
cana-5106	10	7	scalable	scalable	ADJ
cana-5106	10	8	,	,	PUNCT
cana-5106	10	9	efficient	efficient	ADJ
cana-5106	10	10	,	,	PUNCT
cana-5106	10	11	and	and	CCONJ
cana-5106	10	12	intelligent	intelligent	ADJ
cana-5106	10	13	solution	solution	NOUN
cana-5106	10	14	for	for	ADP
cana-5106	10	15	image	image	NOUN
cana-5106	10	16	description	description	NOUN
cana-5106	10	17	,	,	PUNCT
cana-5106	10	18	with	with	ADP
cana-5106	10	19	potential	potential	ADJ
cana-5106	10	20	applications	application	NOUN
cana-5106	10	21	in	in	ADP
cana-5106	10	22	accessibility	accessibility	NOUN
cana-5106	10	23	tools	tool	NOUN
cana-5106	10	24	for	for	ADP
cana-5106	10	25	visually	visually	ADV
cana-5106	10	26	impaired	impaired	ADJ
cana-5106	10	27	users	user	NOUN
cana-5106	10	28	,	,	PUNCT
cana-5106	10	29	digital	digital	ADJ
cana-5106	10	30	content	content	NOUN
cana-5106	10	31	indexing	indexing	NOUN
cana-5106	10	32	,	,	PUNCT
cana-5106	10	33	and	and	CCONJ
cana-5106	10	34	automated	automate	VERB
cana-5106	10	35	content	content	NOUN
cana-5106	10	36	generation.our	generation.our	PROPN
cana-5106	10	37	findings	finding	NOUN
cana-5106	10	38	underscore	underscore	VERB
cana-5106	10	39	the	the	DET
cana-5106	10	40	effectiveness	effectiveness	NOUN
cana-5106	10	41	of	of	ADP
cana-5106	10	42	transformer	transformer	NOUN
cana-5106	10	43	-	-	PUNCT
cana-5106	10	44	based	base	VERB
cana-5106	10	45	models	model	NOUN
cana-5106	10	46	in	in	ADP
cana-5106	10	47	bridging	bridge	VERB
cana-5106	10	48	vision	vision	NOUN
cana-5106	10	49	and	and	CCONJ
cana-5106	10	50	language	language	NOUN
cana-5106	10	51	tasks	task	NOUN
cana-5106	10	52	,	,	PUNCT
cana-5106	10	53	offering	offer	VERB
cana-5106	10	54	promising	promising	ADJ
cana-5106	10	55	directions	direction	NOUN
cana-5106	10	56	for	for	ADP
cana-5106	10	57	future	future	ADJ
cana-5106	10	58	research	research	NOUN
cana-5106	10	59	in	in	ADP
cana-5106	10	60	multimodal	multimodal	NOUN
cana-5106	10	61	ai	ai	NOUN
cana-5106	10	62	.	.	PUNCT
cana-5106	11	1	keywords	keyword	NOUN
cana-5106	11	2	:	:	PUNCT
cana-5106	11	3	vision	vision	NOUN
cana-5106	11	4	transformers	transformer	NOUN
cana-5106	11	5	,	,	PUNCT
cana-5106	11	6	natural	natural	ADJ
cana-5106	11	7	language	language	NOUN
cana-5106	11	8	processing	processing	NOUN
cana-5106	11	9	transformers	transformer	NOUN
cana-5106	11	10	,	,	PUNCT
cana-5106	11	11	image	image	NOUN
cana-5106	11	12	captioning	captioning	NOUN
cana-5106	11	13	,	,	PUNCT
cana-5106	11	14	transformer	transformer	NOUN
cana-5106	11	15	decoder	decoder	NOUN
cana-5106	11	16	,	,	PUNCT
cana-5106	11	17	multimodal	multimodal	ADJ
cana-5106	11	18	learning	learning	NOUN
cana-5106	11	19	,	,	PUNCT
cana-5106	11	20	deep	deep	ADJ
cana-5106	11	21	learning	learning	NOUN
cana-5106	11	22	,	,	PUNCT
cana-5106	11	23	nlp	nlp	ADJ
cana-5106	11	24	,	,	PUNCT
cana-5106	11	25	visual	visual	ADJ
cana-5106	11	26	feature	feature	NOUN
cana-5106	11	27	extraction	extraction	NOUN
cana-5106	11	28	,	,	PUNCT
cana-5106	11	29	intelligent	intelligent	ADJ
cana-5106	11	30	image	image	NOUN
cana-5106	11	31	descriptions	description	NOUN
cana-5106	11	32	.	.	PUNCT
cana-5106	12	1	mailto:deepali.navalkar09@gmail.com	mailto:deepali.navalkar09@gmail.com	PROPN
cana-5106	12	2	mailto:dahibhaterv@gmail.com	mailto:dahibhaterv@gmail.com	PROPN
cana-5106	12	3	mailto:nileshrw_2000@yahoo.com	mailto:nileshrw_2000@yahoo.com	PROPN
cana-5106	12	4	mailto:nawalkar.shubhangi@gmail.com	mailto:nawalkar.shubhangi@gmail.com	NOUN
cana-5106	12	5	communications	communication	NOUN
cana-5106	12	6	on	on	ADP
cana-5106	12	7	applied	apply	VERB
cana-5106	12	8	nonlinear	nonlinear	ADJ
cana-5106	12	9	analysis	analysis	NOUN
cana-5106	12	10	issn	issn	NOUN
cana-5106	12	11	:	:	PUNCT
cana-5106	12	12	1074	1074	NUM
cana-5106	12	13	-	-	PUNCT
cana-5106	12	14	133x	133x	NUM
cana-5106	12	15	vol	vol	VERB
cana-5106	12	16	32	32	NUM
cana-5106	12	17	no	no	NOUN
cana-5106	12	18	.	.	PUNCT
cana-5106	13	1	10s	10	NOUN
cana-5106	13	2	(	(	PUNCT
cana-5106	13	3	2025	2025	NUM
cana-5106	13	4	)	)	PUNCT
cana-5106	13	5	1121	1121	NUM
cana-5106	13	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	13	7	introduction	introduction	NOUN
cana-5106	13	8	the	the	DET
cana-5106	13	9	demand	demand	NOUN
cana-5106	13	10	for	for	ADP
cana-5106	13	11	automated	automate	VERB
cana-5106	13	12	image	image	NOUN
cana-5106	13	13	captioning	captioning	NOUN
cana-5106	13	14	has	have	AUX
cana-5106	13	15	grown	grow	VERB
cana-5106	13	16	rapidly	rapidly	ADV
cana-5106	13	17	due	due	ADP
cana-5106	13	18	to	to	ADP
cana-5106	13	19	the	the	DET
cana-5106	13	20	explosion	explosion	NOUN
cana-5106	13	21	of	of	ADP
cana-5106	13	22	multimedia	multimedia	NOUN
cana-5106	13	23	content	content	NOUN
cana-5106	13	24	and	and	CCONJ
cana-5106	13	25	the	the	DET
cana-5106	13	26	need	need	NOUN
cana-5106	13	27	for	for	ADP
cana-5106	13	28	improved	improve	VERB
cana-5106	13	29	digital	digital	ADJ
cana-5106	13	30	accessibility	accessibility	NOUN
cana-5106	13	31	.	.	PUNCT
cana-5106	14	1	from	from	ADP
cana-5106	14	2	social	social	ADJ
cana-5106	14	3	media	medium	NOUN
cana-5106	14	4	and	and	CCONJ
cana-5106	14	5	digital	digital	ADJ
cana-5106	14	6	archives	archive	NOUN
cana-5106	14	7	to	to	ADP
cana-5106	14	8	assistive	assistive	ADJ
cana-5106	14	9	technologies	technology	NOUN
cana-5106	14	10	,	,	PUNCT
cana-5106	14	11	generating	generate	VERB
cana-5106	14	12	meaningful	meaningful	ADJ
cana-5106	14	13	,	,	PUNCT
cana-5106	14	14	context	context	NOUN
cana-5106	14	15	-	-	PUNCT
cana-5106	14	16	aware	aware	ADJ
cana-5106	14	17	image	image	NOUN
cana-5106	14	18	descriptions	description	NOUN
cana-5106	14	19	has	have	AUX
cana-5106	14	20	become	become	VERB
cana-5106	14	21	crucial	crucial	ADJ
cana-5106	14	22	.	.	PUNCT
cana-5106	15	1	traditional	traditional	ADJ
cana-5106	15	2	image	image	NOUN
cana-5106	15	3	captioning	captioning	NOUN
cana-5106	15	4	approaches	approach	NOUN
cana-5106	15	5	rely	rely	VERB
cana-5106	15	6	on	on	ADP
cana-5106	15	7	convolutional	convolutional	ADJ
cana-5106	15	8	neural	neural	ADJ
cana-5106	15	9	networks	network	NOUN
cana-5106	15	10	(	(	PUNCT
cana-5106	15	11	cnns	cnns	PROPN
cana-5106	15	12	)	)	PUNCT
cana-5106	15	13	for	for	ADP
cana-5106	15	14	visual	visual	ADJ
cana-5106	15	15	feature	feature	NOUN
cana-5106	15	16	extraction	extraction	NOUN
cana-5106	15	17	and	and	CCONJ
cana-5106	15	18	recurrent	recurrent	ADJ
cana-5106	15	19	neural	neural	ADJ
cana-5106	15	20	networks	network	NOUN
cana-5106	15	21	(	(	PUNCT
cana-5106	15	22	rnns	rnns	PROPN
cana-5106	15	23	)	)	PUNCT
cana-5106	15	24	for	for	ADP
cana-5106	15	25	text	text	NOUN
cana-5106	15	26	generation	generation	NOUN
cana-5106	15	27	.	.	PUNCT
cana-5106	16	1	however	however	ADV
cana-5106	16	2	,	,	PUNCT
cana-5106	16	3	these	these	DET
cana-5106	16	4	models	model	NOUN
cana-5106	16	5	often	often	ADV
cana-5106	16	6	struggle	struggle	VERB
cana-5106	16	7	with	with	ADP
cana-5106	16	8	complex	complex	ADJ
cana-5106	16	9	scenes	scene	NOUN
cana-5106	16	10	involving	involve	VERB
cana-5106	16	11	multiple	multiple	ADJ
cana-5106	16	12	objects	object	NOUN
cana-5106	16	13	and	and	CCONJ
cana-5106	16	14	interactions	interaction	NOUN
cana-5106	16	15	,	,	PUNCT
cana-5106	16	16	leading	lead	VERB
cana-5106	16	17	to	to	ADP
cana-5106	16	18	inaccurate	inaccurate	ADJ
cana-5106	16	19	or	or	CCONJ
cana-5106	16	20	simplistic	simplistic	ADJ
cana-5106	16	21	captions	caption	NOUN
cana-5106	16	22	.	.	PUNCT
cana-5106	17	1	to	to	PART
cana-5106	17	2	overcome	overcome	VERB
cana-5106	17	3	these	these	DET
cana-5106	17	4	challenges	challenge	NOUN
cana-5106	17	5	,	,	PUNCT
cana-5106	17	6	smart	smart	ADJ
cana-5106	17	7	caption	caption	NOUN
cana-5106	17	8	leverages	leverage	NOUN
cana-5106	17	9	transformer	transformer	NOUN
cana-5106	17	10	-	-	PUNCT
cana-5106	17	11	based	base	VERB
cana-5106	17	12	architectures	architecture	NOUN
cana-5106	17	13	,	,	PUNCT
cana-5106	17	14	integrating	integrate	VERB
cana-5106	17	15	vision	vision	NOUN
cana-5106	17	16	transformers	transformer	NOUN
cana-5106	17	17	(	(	PUNCT
cana-5106	17	18	vits	vit	NOUN
cana-5106	17	19	)	)	PUNCT
cana-5106	17	20	for	for	ADP
cana-5106	17	21	visual	visual	ADJ
cana-5106	17	22	understanding	understanding	NOUN
cana-5106	17	23	and	and	CCONJ
cana-5106	17	24	nlp	nlp	NOUN
cana-5106	17	25	transformers	transformer	NOUN
cana-5106	17	26	for	for	ADP
cana-5106	17	27	natural	natural	ADJ
cana-5106	17	28	language	language	NOUN
cana-5106	17	29	generation	generation	NOUN
cana-5106	17	30	.	.	PUNCT
cana-5106	18	1	unlike	unlike	ADP
cana-5106	18	2	cnns	cnns	PROPN
cana-5106	18	3	,	,	PUNCT
cana-5106	18	4	vision	vision	NOUN
cana-5106	18	5	transformers	transformer	NOUN
cana-5106	18	6	process	process	VERB
cana-5106	18	7	images	image	NOUN
cana-5106	18	8	as	as	ADP
cana-5106	18	9	sequences	sequence	NOUN
cana-5106	18	10	of	of	ADP
cana-5106	18	11	patches	patch	NOUN
cana-5106	18	12	rather	rather	ADV
cana-5106	18	13	than	than	ADP
cana-5106	18	14	pixels	pixel	NOUN
cana-5106	18	15	,	,	PUNCT
cana-5106	18	16	allowing	allow	VERB
cana-5106	18	17	for	for	ADP
cana-5106	18	18	a	a	DET
cana-5106	18	19	more	more	ADV
cana-5106	18	20	comprehensive	comprehensive	ADJ
cana-5106	18	21	representation	representation	NOUN
cana-5106	18	22	of	of	ADP
cana-5106	18	23	global	global	ADJ
cana-5106	18	24	image	image	NOUN
cana-5106	18	25	context	context	NOUN
cana-5106	18	26	.	.	PUNCT
cana-5106	19	1	this	this	DET
cana-5106	19	2	approach	approach	NOUN
cana-5106	19	3	enables	enable	VERB
cana-5106	19	4	long	long	ADJ
cana-5106	19	5	-	-	PUNCT
cana-5106	19	6	range	range	NOUN
cana-5106	19	7	dependency	dependency	NOUN
cana-5106	19	8	modeling	modeling	NOUN
cana-5106	19	9	,	,	PUNCT
cana-5106	19	10	similar	similar	ADJ
cana-5106	19	11	to	to	ADP
cana-5106	19	12	how	how	SCONJ
cana-5106	19	13	nlp	nlp	NOUN
cana-5106	19	14	models	model	NOUN
cana-5106	19	15	handle	handle	VERB
cana-5106	19	16	sequences	sequence	NOUN
cana-5106	19	17	of	of	ADP
cana-5106	19	18	words	word	NOUN
cana-5106	19	19	,	,	PUNCT
cana-5106	19	20	resulting	result	VERB
cana-5106	19	21	in	in	ADP
cana-5106	19	22	richer	rich	ADJ
cana-5106	19	23	visual	visual	ADJ
cana-5106	19	24	representations.once	representations.once	PROPN
cana-5106	19	25	the	the	DET
cana-5106	19	26	visual	visual	ADJ
cana-5106	19	27	features	feature	NOUN
cana-5106	19	28	are	be	AUX
cana-5106	19	29	extracted	extract	VERB
cana-5106	19	30	,	,	PUNCT
cana-5106	19	31	an	an	DET
cana-5106	19	32	nlp	nlp	NOUN
cana-5106	19	33	transformer	transformer	NOUN
cana-5106	19	34	decoder	decoder	NOUN
cana-5106	19	35	processes	process	VERB
cana-5106	19	36	them	they	PRON
cana-5106	19	37	using	use	VERB
cana-5106	19	38	attention	attention	NOUN
cana-5106	19	39	mechanisms	mechanism	NOUN
cana-5106	19	40	to	to	PART
cana-5106	19	41	generate	generate	VERB
cana-5106	19	42	fluent	fluent	ADJ
cana-5106	19	43	and	and	CCONJ
cana-5106	19	44	contextually	contextually	ADV
cana-5106	19	45	relevant	relevant	ADJ
cana-5106	19	46	captions	caption	NOUN
cana-5106	19	47	.	.	PUNCT
cana-5106	20	1	a	a	DET
cana-5106	20	2	key	key	ADJ
cana-5106	20	3	advantage	advantage	NOUN
cana-5106	20	4	of	of	ADP
cana-5106	20	5	this	this	DET
cana-5106	20	6	approach	approach	NOUN
cana-5106	20	7	is	be	AUX
cana-5106	20	8	its	its	PRON
cana-5106	20	9	ability	ability	NOUN
cana-5106	20	10	to	to	PART
cana-5106	20	11	focus	focus	VERB
cana-5106	20	12	on	on	ADP
cana-5106	20	13	specific	specific	ADJ
cana-5106	20	14	image	image	NOUN
cana-5106	20	15	regions	region	NOUN
cana-5106	20	16	while	while	SCONJ
cana-5106	20	17	simultaneously	simultaneously	ADV
cana-5106	20	18	considering	consider	VERB
cana-5106	20	19	the	the	DET
cana-5106	20	20	overall	overall	ADJ
cana-5106	20	21	scene	scene	NOUN
cana-5106	20	22	.	.	PUNCT
cana-5106	21	1	this	this	DET
cana-5106	21	2	results	result	VERB
cana-5106	21	3	in	in	ADP
cana-5106	21	4	more	more	ADV
cana-5106	21	5	descriptive	descriptive	ADJ
cana-5106	21	6	,	,	PUNCT
cana-5106	21	7	human	human	ADJ
cana-5106	21	8	-	-	PUNCT
cana-5106	21	9	like	like	ADJ
cana-5106	21	10	captions	caption	NOUN
cana-5106	21	11	that	that	PRON
cana-5106	21	12	capture	capture	VERB
cana-5106	21	13	actions	action	NOUN
cana-5106	21	14	,	,	PUNCT
cana-5106	21	15	spatial	spatial	ADJ
cana-5106	21	16	relationships	relationship	NOUN
cana-5106	21	17	,	,	PUNCT
cana-5106	21	18	and	and	CCONJ
cana-5106	21	19	object	object	VERB
cana-5106	21	20	interactions	interaction	NOUN
cana-5106	21	21	—	—	PUNCT
cana-5106	21	22	moving	move	VERB
cana-5106	21	23	beyond	beyond	ADP
cana-5106	21	24	simple	simple	ADJ
cana-5106	21	25	object	object	NOUN
cana-5106	21	26	recognition	recognition	NOUN
cana-5106	21	27	to	to	PART
cana-5106	21	28	provide	provide	VERB
cana-5106	21	29	deeper	deep	ADJ
cana-5106	21	30	contextual	contextual	ADJ
cana-5106	21	31	insights.beyond	insights.beyond	X
cana-5106	21	32	improving	improve	VERB
cana-5106	21	33	caption	caption	NOUN
cana-5106	21	34	quality	quality	NOUN
cana-5106	21	35	,	,	PUNCT
cana-5106	21	36	smart	smart	ADJ
cana-5106	21	37	caption	caption	NOUN
cana-5106	21	38	also	also	ADV
cana-5106	21	39	addresses	address	VERB
cana-5106	21	40	the	the	DET
cana-5106	21	41	limitations	limitation	NOUN
cana-5106	21	42	of	of	ADP
cana-5106	21	43	conventional	conventional	ADJ
cana-5106	21	44	cnn	cnn	PROPN
cana-5106	21	45	-	-	PUNCT
cana-5106	21	46	rnn	rnn	PROPN
cana-5106	21	47	models	model	NOUN
cana-5106	21	48	,	,	PUNCT
cana-5106	21	49	which	which	PRON
cana-5106	21	50	often	often	ADV
cana-5106	21	51	require	require	VERB
cana-5106	21	52	extensive	extensive	ADJ
cana-5106	21	53	manual	manual	ADJ
cana-5106	21	54	tuning	tuning	NOUN
cana-5106	21	55	and	and	CCONJ
cana-5106	21	56	struggle	struggle	VERB
cana-5106	21	57	with	with	ADP
cana-5106	21	58	generalization	generalization	NOUN
cana-5106	21	59	across	across	ADP
cana-5106	21	60	diverse	diverse	ADJ
cana-5106	21	61	datasets	dataset	NOUN
cana-5106	21	62	.	.	PUNCT
cana-5106	22	1	in	in	ADP
cana-5106	22	2	contrast	contrast	NOUN
cana-5106	22	3	,	,	PUNCT
cana-5106	22	4	transformer	transformer	NOUN
cana-5106	22	5	-	-	PUNCT
cana-5106	22	6	based	base	VERB
cana-5106	22	7	architectures	architecture	NOUN
cana-5106	22	8	are	be	AUX
cana-5106	22	9	highly	highly	ADV
cana-5106	22	10	scalable	scalable	ADJ
cana-5106	22	11	and	and	CCONJ
cana-5106	22	12	adaptable	adaptable	ADJ
cana-5106	22	13	,	,	PUNCT
cana-5106	22	14	making	make	VERB
cana-5106	22	15	them	they	PRON
cana-5106	22	16	effective	effective	ADJ
cana-5106	22	17	across	across	ADP
cana-5106	22	18	various	various	ADJ
cana-5106	22	19	image	image	NOUN
cana-5106	22	20	types	type	NOUN
cana-5106	22	21	and	and	CCONJ
cana-5106	22	22	complexities	complexity	NOUN
cana-5106	22	23	.	.	PUNCT
cana-5106	23	1	vision	vision	NOUN
cana-5106	23	2	transformers	transformer	NOUN
cana-5106	23	3	have	have	AUX
cana-5106	23	4	already	already	ADV
cana-5106	23	5	demonstrated	demonstrate	VERB
cana-5106	23	6	state	state	NOUN
cana-5106	23	7	-	-	PUNCT
cana-5106	23	8	of	of	ADP
cana-5106	23	9	-	-	PUNCT
cana-5106	23	10	the	the	DET
cana-5106	23	11	-	-	PUNCT
cana-5106	23	12	art	art	NOUN
cana-5106	23	13	performance	performance	NOUN
cana-5106	23	14	in	in	ADP
cana-5106	23	15	image	image	NOUN
cana-5106	23	16	recognition	recognition	NOUN
cana-5106	23	17	,	,	PUNCT
cana-5106	23	18	while	while	SCONJ
cana-5106	23	19	nlp	nlp	ADJ
cana-5106	23	20	transformers	transformer	NOUN
cana-5106	23	21	,	,	PUNCT
cana-5106	23	22	through	through	ADP
cana-5106	23	23	models	model	NOUN
cana-5106	23	24	like	like	ADP
cana-5106	23	25	bert	bert	NOUN
cana-5106	23	26	and	and	CCONJ
cana-5106	23	27	gpt	gpt	NOUN
cana-5106	23	28	,	,	PUNCT
cana-5106	23	29	have	have	AUX
cana-5106	23	30	revolutionized	revolutionize	VERB
cana-5106	23	31	natural	natural	ADJ
cana-5106	23	32	language	language	NOUN
cana-5106	23	33	processing	processing	NOUN
cana-5106	23	34	.	.	PUNCT
cana-5106	24	1	by	by	ADP
cana-5106	24	2	combining	combine	VERB
cana-5106	24	3	these	these	DET
cana-5106	24	4	advanced	advanced	ADJ
cana-5106	24	5	techniques	technique	NOUN
cana-5106	24	6	,	,	PUNCT
cana-5106	24	7	smart	smart	ADJ
cana-5106	24	8	caption	caption	NOUN
cana-5106	24	9	establishes	establish	VERB
cana-5106	24	10	a	a	DET
cana-5106	24	11	new	new	ADJ
cana-5106	24	12	benchmark	benchmark	NOUN
cana-5106	24	13	for	for	ADP
cana-5106	24	14	automated	automate	VERB
cana-5106	24	15	image	image	NOUN
cana-5106	24	16	captioning.this	captioning.this	PRON
cana-5106	24	17	paper	paper	NOUN
cana-5106	24	18	presents	present	VERB
cana-5106	24	19	the	the	DET
cana-5106	24	20	design	design	NOUN
cana-5106	24	21	,	,	PUNCT
cana-5106	24	22	implementation	implementation	NOUN
cana-5106	24	23	,	,	PUNCT
cana-5106	24	24	and	and	CCONJ
cana-5106	24	25	evaluation	evaluation	NOUN
cana-5106	24	26	of	of	ADP
cana-5106	24	27	smart	smart	ADJ
cana-5106	24	28	caption	caption	NOUN
cana-5106	24	29	,	,	PUNCT
cana-5106	24	30	detailing	detail	VERB
cana-5106	24	31	how	how	SCONJ
cana-5106	24	32	integrating	integrating	ADJ
cana-5106	24	33	vision	vision	NOUN
cana-5106	24	34	and	and	CCONJ
cana-5106	24	35	nlp	nlp	NOUN
cana-5106	24	36	transformers	transformer	NOUN
cana-5106	24	37	enhances	enhance	VERB
cana-5106	24	38	visual	visual	ADJ
cana-5106	24	39	feature	feature	NOUN
cana-5106	24	40	extraction	extraction	NOUN
cana-5106	24	41	and	and	CCONJ
cana-5106	24	42	language	language	NOUN
cana-5106	24	43	generation	generation	NOUN
cana-5106	24	44	.	.	PUNCT
cana-5106	25	1	through	through	ADP
cana-5106	25	2	extensive	extensive	ADJ
cana-5106	25	3	experiments	experiment	NOUN
cana-5106	25	4	,	,	PUNCT
cana-5106	25	5	we	we	PRON
cana-5106	25	6	demonstrate	demonstrate	VERB
cana-5106	25	7	that	that	SCONJ
cana-5106	25	8	smart	smart	ADJ
cana-5106	25	9	caption	caption	NOUN
cana-5106	25	10	significantly	significantly	ADV
cana-5106	25	11	outperforms	outperform	VERB
cana-5106	25	12	traditional	traditional	ADJ
cana-5106	25	13	models	model	NOUN
cana-5106	25	14	in	in	ADP
cana-5106	25	15	accuracy	accuracy	NOUN
cana-5106	25	16	,	,	PUNCT
cana-5106	25	17	fluency	fluency	NOUN
cana-5106	25	18	,	,	PUNCT
cana-5106	25	19	and	and	CCONJ
cana-5106	25	20	contextual	contextual	ADJ
cana-5106	25	21	relevance	relevance	NOUN
cana-5106	25	22	.	.	PUNCT
cana-5106	26	1	additionally	additionally	ADV
cana-5106	26	2	,	,	PUNCT
cana-5106	26	3	we	we	PRON
cana-5106	26	4	explore	explore	VERB
cana-5106	26	5	its	its	PRON
cana-5106	26	6	applications	application	NOUN
cana-5106	26	7	across	across	ADP
cana-5106	26	8	domains	domain	NOUN
cana-5106	26	9	such	such	ADJ
cana-5106	26	10	as	as	ADP
cana-5106	26	11	assistive	assistive	ADJ
cana-5106	26	12	technology	technology	NOUN
cana-5106	26	13	for	for	ADP
cana-5106	26	14	visually	visually	ADV
cana-5106	26	15	impaired	impaired	ADJ
cana-5106	26	16	users	user	NOUN
cana-5106	26	17	,	,	PUNCT
cana-5106	26	18	digital	digital	ADJ
cana-5106	26	19	content	content	NOUN
cana-5106	26	20	indexing	indexing	NOUN
cana-5106	26	21	,	,	PUNCT
cana-5106	26	22	and	and	CCONJ
cana-5106	26	23	automated	automated	ADJ
cana-5106	26	24	multimedia	multimedia	NOUN
cana-5106	26	25	analysis	analysis	NOUN
cana-5106	26	26	.	.	PUNCT
cana-5106	27	1	finally	finally	ADV
cana-5106	27	2	,	,	PUNCT
cana-5106	27	3	we	we	PRON
cana-5106	27	4	discuss	discuss	VERB
cana-5106	27	5	future	future	ADJ
cana-5106	27	6	directions	direction	NOUN
cana-5106	27	7	in	in	ADP
cana-5106	27	8	multimodal	multimodal	ADJ
cana-5106	27	9	learning	learning	NOUN
cana-5106	27	10	,	,	PUNCT
cana-5106	27	11	paving	pave	VERB
cana-5106	27	12	the	the	DET
cana-5106	27	13	way	way	NOUN
cana-5106	27	14	for	for	ADP
cana-5106	27	15	more	more	ADV
cana-5106	27	16	intelligent	intelligent	ADJ
cana-5106	27	17	,	,	PUNCT
cana-5106	27	18	context	context	NOUN
cana-5106	27	19	-	-	PUNCT
cana-5106	27	20	aware	aware	ADJ
cana-5106	27	21	ai	ai	PROPN
cana-5106	27	22	systems	system	NOUN
cana-5106	27	23	.	.	PUNCT
cana-5106	28	1	i.	i.	PROPN
cana-5106	28	2	related	relate	VERB
cana-5106	28	3	work+	work+	PROPN
cana-5106	28	4	1	1	NUM
cana-5106	28	5	)	)	PUNCT
cana-5106	28	6	in	in	ADP
cana-5106	28	7	their	their	PRON
cana-5106	28	8	paper	paper	NOUN
cana-5106	28	9	“	"	PUNCT
cana-5106	28	10	automatic	automatic	ADJ
cana-5106	28	11	indonesian	indonesian	ADJ
cana-5106	28	12	image	image	NOUN
cana-5106	28	13	captioning	captioning	NOUN
cana-5106	28	14	using	use	VERB
cana-5106	28	15	cnn	cnn	PROPN
cana-5106	28	16	and	and	CCONJ
cana-5106	28	17	transformerbased	transformerbase	VERB
cana-5106	28	18	model	model	NOUN
cana-5106	28	19	approach	approach	NOUN
cana-5106	28	20	,	,	PUNCT
cana-5106	28	21	”	"	PUNCT
cana-5106	28	22	r.	r.	PROPN
cana-5106	28	23	mulyawan	mulyawan	PROPN
cana-5106	28	24	,	,	PUNCT
cana-5106	28	25	a.	a.	NOUN
cana-5106	28	26	sunyoto	sunyoto	PROPN
cana-5106	28	27	,	,	PUNCT
cana-5106	28	28	and	and	CCONJ
cana-5106	28	29	a.	a.	PROPN
cana-5106	28	30	h.	h.	PROPN
cana-5106	28	31	muhammad	muhammad	PROPN
cana-5106	28	32	propose	propose	VERB
cana-5106	28	33	a	a	DET
cana-5106	28	34	novel	novel	ADJ
cana-5106	28	35	hybrid	hybrid	NOUN
cana-5106	28	36	approach	approach	NOUN
cana-5106	28	37	combining	combine	VERB
cana-5106	28	38	convolutional	convolutional	ADJ
cana-5106	28	39	neural	neural	ADJ
cana-5106	28	40	networks	network	NOUN
cana-5106	28	41	(	(	PUNCT
cana-5106	28	42	cnns	cnns	PROPN
cana-5106	28	43	)	)	PUNCT
cana-5106	28	44	and	and	CCONJ
cana-5106	28	45	transformer	transformer	NOUN
cana-5106	28	46	models	model	NOUN
cana-5106	28	47	to	to	PART
cana-5106	28	48	improve	improve	VERB
cana-5106	28	49	image	image	NOUN
cana-5106	28	50	captioning	captioning	NOUN
cana-5106	28	51	for	for	ADP
cana-5106	28	52	the	the	DET
cana-5106	28	53	indonesian	indonesian	ADJ
cana-5106	28	54	language	language	NOUN
cana-5106	28	55	.	.	PUNCT
cana-5106	29	1	the	the	DET
cana-5106	29	2	study	study	NOUN
cana-5106	29	3	aims	aim	VERB
cana-5106	29	4	to	to	PART
cana-5106	29	5	enhance	enhance	VERB
cana-5106	29	6	communications	communication	NOUN
cana-5106	29	7	on	on	ADP
cana-5106	29	8	applied	apply	VERB
cana-5106	29	9	nonlinear	nonlinear	ADJ
cana-5106	29	10	analysis	analysis	NOUN
cana-5106	29	11	issn	issn	NOUN
cana-5106	29	12	:	:	PUNCT
cana-5106	29	13	1074	1074	NUM
cana-5106	29	14	-	-	PUNCT
cana-5106	29	15	133x	133x	NUM
cana-5106	29	16	vol	vol	VERB
cana-5106	29	17	32	32	NUM
cana-5106	29	18	no	no	NOUN
cana-5106	29	19	.	.	PUNCT
cana-5106	30	1	10s	10	NOUN
cana-5106	30	2	(	(	PUNCT
cana-5106	30	3	2025	2025	NUM
cana-5106	30	4	)	)	PUNCT
cana-5106	30	5	1122	1122	NUM
cana-5106	30	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	30	7	the	the	DET
cana-5106	30	8	quality	quality	NOUN
cana-5106	30	9	of	of	ADP
cana-5106	30	10	automatic	automatic	ADJ
cana-5106	30	11	image	image	NOUN
cana-5106	30	12	descriptions	description	NOUN
cana-5106	30	13	by	by	ADP
cana-5106	30	14	leveraging	leverage	VERB
cana-5106	30	15	the	the	DET
cana-5106	30	16	strengths	strength	NOUN
cana-5106	30	17	of	of	ADP
cana-5106	30	18	both	both	DET
cana-5106	30	19	architectures	architecture	NOUN
cana-5106	30	20	,	,	PUNCT
cana-5106	30	21	resulting	result	VERB
cana-5106	30	22	in	in	ADP
cana-5106	30	23	better	well	ADJ
cana-5106	30	24	contextual	contextual	ADJ
cana-5106	30	25	understanding	understanding	NOUN
cana-5106	30	26	and	and	CCONJ
cana-5106	30	27	linguistic	linguistic	ADJ
cana-5106	30	28	fluency	fluency	NOUN
cana-5106	30	29	.	.	PUNCT
cana-5106	31	1	the	the	DET
cana-5106	31	2	key	key	ADJ
cana-5106	31	3	merit	merit	NOUN
cana-5106	31	4	of	of	ADP
cana-5106	31	5	this	this	DET
cana-5106	31	6	approach	approach	NOUN
cana-5106	31	7	lies	lie	VERB
cana-5106	31	8	in	in	ADP
cana-5106	31	9	its	its	PRON
cana-5106	31	10	ability	ability	NOUN
cana-5106	31	11	to	to	PART
cana-5106	31	12	effectively	effectively	ADV
cana-5106	31	13	capture	capture	VERB
cana-5106	31	14	visual	visual	ADJ
cana-5106	31	15	features	feature	NOUN
cana-5106	31	16	while	while	SCONJ
cana-5106	31	17	generating	generate	VERB
cana-5106	31	18	coherent	coherent	ADJ
cana-5106	31	19	textual	textual	ADJ
cana-5106	31	20	descriptions	description	NOUN
cana-5106	31	21	.	.	PUNCT
cana-5106	32	1	we	we	PRON
cana-5106	32	2	plan	plan	VERB
cana-5106	32	3	to	to	PART
cana-5106	32	4	adapt	adapt	VERB
cana-5106	32	5	this	this	DET
cana-5106	32	6	hybrid	hybrid	ADJ
cana-5106	32	7	model	model	NOUN
cana-5106	32	8	in	in	ADP
cana-5106	32	9	our	our	PRON
cana-5106	32	10	project	project	NOUN
cana-5106	32	11	by	by	ADP
cana-5106	32	12	incorporating	incorporate	VERB
cana-5106	32	13	vision	vision	NOUN
cana-5106	32	14	transformers	transformer	NOUN
cana-5106	32	15	and	and	CCONJ
cana-5106	32	16	nlp	nlp	NOUN
cana-5106	32	17	transformers	transformer	NOUN
cana-5106	32	18	for	for	ADP
cana-5106	32	19	improved	improved	ADJ
cana-5106	32	20	image	image	NOUN
cana-5106	32	21	-	-	PUNCT
cana-5106	32	22	text	text	NOUN
cana-5106	32	23	integration	integration	NOUN
cana-5106	32	24	.	.	PUNCT
cana-5106	33	1	however	however	ADV
cana-5106	33	2	,	,	PUNCT
cana-5106	33	3	the	the	DET
cana-5106	33	4	paper	paper	NOUN
cana-5106	33	5	highlights	highlight	NOUN
cana-5106	33	6	limitations	limitation	NOUN
cana-5106	33	7	such	such	ADJ
cana-5106	33	8	as	as	ADP
cana-5106	33	9	dataset	dataset	ADJ
cana-5106	33	10	specificity	specificity	NOUN
cana-5106	33	11	and	and	CCONJ
cana-5106	33	12	potential	potential	ADJ
cana-5106	33	13	biases	bias	NOUN
cana-5106	33	14	in	in	ADP
cana-5106	33	15	training	training	NOUN
cana-5106	33	16	data	datum	NOUN
cana-5106	33	17	,	,	PUNCT
cana-5106	33	18	which	which	PRON
cana-5106	33	19	may	may	AUX
cana-5106	33	20	hinder	hinder	VERB
cana-5106	33	21	generalization	generalization	NOUN
cana-5106	33	22	.	.	PUNCT
cana-5106	34	1	to	to	PART
cana-5106	34	2	address	address	VERB
cana-5106	34	3	these	these	DET
cana-5106	34	4	issues	issue	NOUN
cana-5106	34	5	,	,	PUNCT
cana-5106	34	6	our	our	PRON
cana-5106	34	7	project	project	NOUN
cana-5106	34	8	will	will	AUX
cana-5106	34	9	utilize	utilize	VERB
cana-5106	34	10	a	a	DET
cana-5106	34	11	more	more	ADV
cana-5106	34	12	diverse	diverse	ADJ
cana-5106	34	13	dataset	dataset	NOUN
cana-5106	34	14	and	and	CCONJ
cana-5106	34	15	implement	implement	VERB
cana-5106	34	16	bias	bias	NOUN
cana-5106	34	17	mitigation	mitigation	NOUN
cana-5106	34	18	strategies	strategy	NOUN
cana-5106	34	19	,	,	PUNCT
cana-5106	34	20	aiming	aim	VERB
cana-5106	34	21	for	for	ADP
cana-5106	34	22	a	a	DET
cana-5106	34	23	more	more	ADV
cana-5106	34	24	robust	robust	ADJ
cana-5106	34	25	and	and	CCONJ
cana-5106	34	26	adaptable	adaptable	ADJ
cana-5106	34	27	image	image	NOUN
cana-5106	34	28	captioning	caption	VERB
cana-5106	34	29	system.[1	system.[1	NOUN
cana-5106	34	30	]	]	PUNCT
cana-5106	34	31	2	2	X
cana-5106	34	32	)	)	PUNCT
cana-5106	34	33	the	the	DET
cana-5106	34	34	visual	visual	ADJ
cana-5106	34	35	features	feature	NOUN
cana-5106	34	36	of	of	ADP
cana-5106	34	37	images	image	NOUN
cana-5106	34	38	are	be	AUX
cana-5106	34	39	pre	pre	X
cana-5106	34	40	computed	compute	VERB
cana-5106	34	41	offline	offline	ADJ
cana-5106	34	42	and	and	CCONJ
cana-5106	34	43	stored	store	VERB
cana-5106	34	44	by	by	ADP
cana-5106	34	45	the	the	DET
cana-5106	34	46	search	search	NOUN
cana-5106	34	47	engine	engine	NOUN
cana-5106	34	48	.	.	PUNCT
cana-5106	35	1	the	the	DET
cana-5106	35	2	main	main	ADJ
cana-5106	35	3	online	online	ADJ
cana-5106	35	4	computational	computational	ADJ
cana-5106	35	5	cost	cost	NOUN
cana-5106	35	6	of	of	ADP
cana-5106	35	7	image	image	NOUN
cana-5106	35	8	re	re	VERB
cana-5106	35	9	-	-	VERB
cana-5106	35	10	ranking	ranking	ADJ
cana-5106	35	11	is	be	AUX
cana-5106	35	12	on	on	ADP
cana-5106	35	13	comparing	compare	VERB
cana-5106	35	14	visual	visual	ADJ
cana-5106	35	15	features	feature	NOUN
cana-5106	35	16	.	.	PUNCT
cana-5106	36	1	in	in	ADP
cana-5106	36	2	order	order	NOUN
cana-5106	36	3	to	to	PART
cana-5106	36	4	achieve	achieve	VERB
cana-5106	36	5	high	high	ADJ
cana-5106	36	6	efficiency	efficiency	NOUN
cana-5106	36	7	,	,	PUNCT
cana-5106	36	8	the	the	DET
cana-5106	36	9	visual	visual	ADJ
cana-5106	36	10	feature	feature	NOUN
cana-5106	36	11	vectors	vector	NOUN
cana-5106	36	12	need	need	VERB
cana-5106	36	13	to	to	PART
cana-5106	36	14	be	be	AUX
cana-5106	36	15	short	short	ADJ
cana-5106	36	16	and	and	CCONJ
cana-5106	36	17	their	their	PRON
cana-5106	36	18	matching	matching	NOUN
cana-5106	36	19	needs	need	NOUN
cana-5106	36	20	to	to	PART
cana-5106	36	21	be	be	AUX
cana-5106	36	22	fast	fast	ADJ
cana-5106	36	23	.	.	PUNCT
cana-5106	37	1	another	another	DET
cana-5106	37	2	major	major	ADJ
cana-5106	37	3	challenge	challenge	NOUN
cana-5106	37	4	is	be	AUX
cana-5106	37	5	that	that	SCONJ
cana-5106	37	6	the	the	DET
cana-5106	37	7	similarities	similarity	NOUN
cana-5106	37	8	of	of	ADP
cana-5106	37	9	low	low	ADJ
cana-5106	37	10	-	-	PUNCT
cana-5106	37	11	level	level	NOUN
cana-5106	37	12	visual	visual	ADJ
cana-5106	37	13	features	feature	NOUN
cana-5106	37	14	may	may	AUX
cana-5106	37	15	not	not	PART
cana-5106	37	16	well	well	ADV
cana-5106	37	17	correlate	correlate	VERB
cana-5106	37	18	with	with	ADP
cana-5106	37	19	images	image	NOUN
cana-5106	37	20	„	„	PUNCT
cana-5106	37	21	high	high	ADJ
cana-5106	37	22	-	-	PUNCT
cana-5106	37	23	level	level	NOUN
cana-5106	37	24	semantic	semantic	ADJ
cana-5106	37	25	meanings	meaning	NOUN
cana-5106	37	26	which	which	PRON
cana-5106	37	27	interpret	interpret	VERB
cana-5106	37	28	users‟	users‟	PROPN
cana-5106	37	29	search	search	NOUN
cana-5106	37	30	intention	intention	NOUN
cana-5106	37	31	.	.	PUNCT
cana-5106	38	1	to	to	PART
cana-5106	38	2	narrow	narrow	VERB
cana-5106	38	3	down	down	ADP
cana-5106	38	4	this	this	DET
cana-5106	38	5	semantic	semantic	ADJ
cana-5106	38	6	gap	gap	NOUN
cana-5106	38	7	,	,	PUNCT
cana-5106	38	8	for	for	ADP
cana-5106	38	9	offline	offline	ADJ
cana-5106	38	10	image	image	NOUN
cana-5106	38	11	recognition	recognition	NOUN
cana-5106	38	12	and	and	CCONJ
cana-5106	38	13	retrieval	retrieval	NOUN
cana-5106	38	14	,	,	PUNCT
cana-5106	38	15	there	there	PRON
cana-5106	38	16	have	have	AUX
cana-5106	38	17	been	be	AUX
cana-5106	38	18	a	a	DET
cana-5106	38	19	number	number	NOUN
cana-5106	38	20	of	of	ADP
cana-5106	38	21	studies	study	NOUN
cana-5106	38	22	to	to	PART
cana-5106	38	23	map	map	VERB
cana-5106	38	24	visual	visual	ADJ
cana-5106	38	25	features	feature	NOUN
cana-5106	38	26	to	to	ADP
cana-5106	38	27	a	a	DET
cana-5106	38	28	set	set	NOUN
cana-5106	38	29	of	of	ADP
cana-5106	38	30	predefined	predefine	VERB
cana-5106	38	31	concepts	concept	NOUN
cana-5106	38	32	or	or	CCONJ
cana-5106	38	33	attributes	attribute	NOUN
cana-5106	38	34	as	as	ADP
cana-5106	38	35	semantic	semantic	ADJ
cana-5106	38	36	signature	signature	NOUN
cana-5106	38	37	however	however	ADV
cana-5106	38	38	,	,	PUNCT
cana-5106	38	39	these	these	DET
cana-5106	38	40	approaches	approach	NOUN
cana-5106	38	41	are	be	AUX
cana-5106	38	42	only	only	ADV
cana-5106	38	43	applicable	applicable	ADJ
cana-5106	38	44	to	to	ADP
cana-5106	38	45	closed	close	VERB
cana-5106	38	46	image	image	NOUN
cana-5106	38	47	sets	set	NOUN
cana-5106	38	48	of	of	ADP
cana-5106	38	49	relatively	relatively	ADV
cana-5106	38	50	small	small	ADJ
cana-5106	38	51	sizes	size	NOUN
cana-5106	38	52	.	.	PUNCT
cana-5106	39	1	they	they	PRON
cana-5106	39	2	are	be	AUX
cana-5106	39	3	not	not	PART
cana-5106	39	4	suitable	suitable	ADJ
cana-5106	39	5	for	for	ADP
cana-5106	39	6	online	online	ADJ
cana-5106	39	7	webbased	webbase	VERB
cana-5106	39	8	image	image	NOUN
cana-5106	39	9	re	re	VERB
cana-5106	39	10	-	-	NOUN
cana-5106	39	11	ranking	ranking	ADJ
cana-5106	39	12	.	.	PUNCT
cana-5106	40	1	according	accord	VERB
cana-5106	40	2	to	to	ADP
cana-5106	40	3	our	our	PRON
cana-5106	40	4	empirical	empirical	ADJ
cana-5106	40	5	study	study	NOUN
cana-5106	40	6	,	,	PUNCT
cana-5106	40	7	images	image	NOUN
cana-5106	40	8	retrieved	retrieve	VERB
cana-5106	40	9	by	by	ADP
cana-5106	40	10	120	120	NUM
cana-5106	40	11	query	query	NOUN
cana-5106	40	12	keywords	keyword	NOUN
cana-5106	40	13	alone	alone	ADV
cana-5106	40	14	include	include	VERB
cana-5106	40	15	more	more	ADJ
cana-5106	40	16	than	than	ADP
cana-5106	40	17	1500	1500	NUM
cana-5106	40	18	concepts	concept	NOUN
cana-5106	40	19	.	.	PUNCT
cana-5106	41	1	therefore	therefore	ADV
cana-5106	41	2	,	,	PUNCT
cana-5106	41	3	it	it	PRON
cana-5106	41	4	is	be	AUX
cana-5106	41	5	difficult	difficult	ADJ
cana-5106	41	6	and	and	CCONJ
cana-5106	41	7	inefficient	inefficient	ADJ
cana-5106	41	8	to	to	PART
cana-5106	41	9	design	design	VERB
cana-5106	41	10	a	a	DET
cana-5106	41	11	huge	huge	ADJ
cana-5106	41	12	concept	concept	NOUN
cana-5106	41	13	dictionary	dictionary	NOUN
cana-5106	41	14	to	to	PART
cana-5106	41	15	characterize	characterize	VERB
cana-5106	41	16	highly	highly	ADV
cana-5106	41	17	diverse	diverse	ADJ
cana-5106	41	18	web	web	NOUN
cana-5106	41	19	images.[2	images.[2	NOUN
cana-5106	41	20	]	]	X
cana-5106	41	21	3	3	NUM
cana-5106	41	22	)	)	PUNCT
cana-5106	41	23	in	in	ADP
cana-5106	41	24	“	"	PUNCT
cana-5106	41	25	dynamic	dynamic	ADJ
cana-5106	41	26	head	head	NOUN
cana-5106	41	27	:	:	PUNCT
cana-5106	41	28	unifying	unify	VERB
cana-5106	41	29	object	object	NOUN
cana-5106	41	30	detection	detection	NOUN
cana-5106	41	31	heads	head	NOUN
cana-5106	41	32	with	with	ADP
cana-5106	41	33	attentions	attention	NOUN
cana-5106	41	34	,	,	PUNCT
cana-5106	41	35	”	"	PUNCT
cana-5106	41	36	xiyang	xiyang	PROPN
cana-5106	41	37	dai	dai	PROPN
cana-5106	41	38	et	et	PROPN
cana-5106	41	39	al	al	PROPN
cana-5106	41	40	.	.	PROPN
cana-5106	41	41	introduce	introduce	VERB
cana-5106	41	42	a	a	DET
cana-5106	41	43	novel	novel	ADJ
cana-5106	41	44	method	method	NOUN
cana-5106	41	45	to	to	PART
cana-5106	41	46	unify	unify	VERB
cana-5106	41	47	object	object	NOUN
cana-5106	41	48	detection	detection	NOUN
cana-5106	41	49	heads	head	NOUN
cana-5106	41	50	using	use	VERB
cana-5106	41	51	attention	attention	NOUN
cana-5106	41	52	mechanisms	mechanism	NOUN
cana-5106	41	53	,	,	PUNCT
cana-5106	41	54	aiming	aim	VERB
cana-5106	41	55	to	to	PART
cana-5106	41	56	enhance	enhance	VERB
cana-5106	41	57	object	object	NOUN
cana-5106	41	58	detection	detection	NOUN
cana-5106	41	59	performance	performance	NOUN
cana-5106	41	60	.	.	PUNCT
cana-5106	42	1	the	the	DET
cana-5106	42	2	study	study	NOUN
cana-5106	42	3	focuses	focus	VERB
cana-5106	42	4	on	on	ADP
cana-5106	42	5	dynamically	dynamically	ADV
cana-5106	42	6	adjusting	adjust	VERB
cana-5106	42	7	attention	attention	NOUN
cana-5106	42	8	across	across	ADP
cana-5106	42	9	different	different	ADJ
cana-5106	42	10	heads	head	NOUN
cana-5106	42	11	based	base	VERB
cana-5106	42	12	on	on	ADP
cana-5106	42	13	input	input	NOUN
cana-5106	42	14	data	datum	NOUN
cana-5106	42	15	,	,	PUNCT
cana-5106	42	16	improving	improve	VERB
cana-5106	42	17	feature	feature	NOUN
cana-5106	42	18	extraction	extraction	NOUN
cana-5106	42	19	and	and	CCONJ
cana-5106	42	20	representation	representation	NOUN
cana-5106	42	21	.	.	PUNCT
cana-5106	43	1	the	the	DET
cana-5106	43	2	key	key	ADJ
cana-5106	43	3	strength	strength	NOUN
cana-5106	43	4	of	of	ADP
cana-5106	43	5	this	this	DET
cana-5106	43	6	approach	approach	NOUN
cana-5106	43	7	is	be	AUX
cana-5106	43	8	its	its	PRON
cana-5106	43	9	innovative	innovative	ADJ
cana-5106	43	10	use	use	NOUN
cana-5106	43	11	of	of	ADP
cana-5106	43	12	attention	attention	NOUN
cana-5106	43	13	mechanisms	mechanism	NOUN
cana-5106	43	14	,	,	PUNCT
cana-5106	43	15	which	which	PRON
cana-5106	43	16	we	we	PRON
cana-5106	43	17	plan	plan	VERB
cana-5106	43	18	to	to	PART
cana-5106	43	19	integrate	integrate	VERB
cana-5106	43	20	into	into	ADP
cana-5106	43	21	our	our	PRON
cana-5106	43	22	project	project	NOUN
cana-5106	43	23	to	to	PART
cana-5106	43	24	enhance	enhance	VERB
cana-5106	43	25	the	the	DET
cana-5106	43	26	vision	vision	NOUN
cana-5106	43	27	transformer	transformer	PROPN
cana-5106	43	28	’s	’s	PART
cana-5106	43	29	visual	visual	ADJ
cana-5106	43	30	feature	feature	NOUN
cana-5106	43	31	extraction	extraction	NOUN
cana-5106	43	32	capabilities	capability	NOUN
cana-5106	43	33	.	.	PUNCT
cana-5106	44	1	however	however	ADV
cana-5106	44	2	,	,	PUNCT
cana-5106	44	3	the	the	DET
cana-5106	44	4	increased	increase	VERB
cana-5106	44	5	complexity	complexity	NOUN
cana-5106	44	6	of	of	ADP
cana-5106	44	7	dynamic	dynamic	ADJ
cana-5106	44	8	head	head	NOUN
cana-5106	44	9	configurations	configuration	NOUN
cana-5106	44	10	may	may	AUX
cana-5106	44	11	demand	demand	VERB
cana-5106	44	12	significant	significant	ADJ
cana-5106	44	13	computational	computational	ADJ
cana-5106	44	14	resources	resource	NOUN
cana-5106	44	15	and	and	CCONJ
cana-5106	44	16	tuning	tuning	NOUN
cana-5106	44	17	.	.	PUNCT
cana-5106	45	1	to	to	PART
cana-5106	45	2	mitigate	mitigate	VERB
cana-5106	45	3	this	this	PRON
cana-5106	45	4	,	,	PUNCT
cana-5106	45	5	our	our	PRON
cana-5106	45	6	project	project	NOUN
cana-5106	45	7	will	will	AUX
cana-5106	45	8	focus	focus	VERB
cana-5106	45	9	on	on	ADP
cana-5106	45	10	streamlining	streamline	VERB
cana-5106	45	11	the	the	DET
cana-5106	45	12	architecture	architecture	NOUN
cana-5106	45	13	to	to	PART
cana-5106	45	14	maintain	maintain	VERB
cana-5106	45	15	effectiveness	effectiveness	NOUN
cana-5106	45	16	while	while	SCONJ
cana-5106	45	17	reducing	reduce	VERB
cana-5106	45	18	computational	computational	ADJ
cana-5106	45	19	demands	demand	NOUN
cana-5106	45	20	,	,	PUNCT
cana-5106	45	21	ensuring	ensure	VERB
cana-5106	45	22	efficient	efficient	ADJ
cana-5106	45	23	image	image	NOUN
cana-5106	45	24	caption	caption	NOUN
cana-5106	45	25	generation.[3	generation.[3	NOUN
cana-5106	45	26	]	]	PUNCT
cana-5106	45	27	4	4	X
cana-5106	45	28	)	)	PUNCT
cana-5106	45	29	yuxin	yuxin	PROPN
cana-5106	45	30	fang	fang	X
cana-5106	45	31	et	et	PROPN
cana-5106	45	32	al	al	PROPN
cana-5106	45	33	.	.	PROPN
cana-5106	45	34	,	,	PUNCT
cana-5106	45	35	in	in	ADP
cana-5106	45	36	“	"	PUNCT
cana-5106	45	37	you	you	PRON
cana-5106	45	38	only	only	ADV
cana-5106	45	39	look	look	VERB
cana-5106	45	40	at	at	ADP
cana-5106	45	41	one	one	NUM
cana-5106	45	42	sequence	sequence	NOUN
cana-5106	45	43	:	:	PUNCT
cana-5106	45	44	rethinking	rethink	VERB
cana-5106	45	45	transformer	transformer	NOUN
cana-5106	45	46	in	in	ADP
cana-5106	45	47	vision	vision	NOUN
cana-5106	45	48	through	through	ADP
cana-5106	45	49	object	object	NOUN
cana-5106	45	50	detection	detection	NOUN
cana-5106	45	51	,	,	PUNCT
cana-5106	45	52	”	"	PUNCT
cana-5106	45	53	propose	propose	VERB
cana-5106	45	54	a	a	DET
cana-5106	45	55	simplified	simplified	ADJ
cana-5106	45	56	approach	approach	NOUN
cana-5106	45	57	to	to	ADP
cana-5106	45	58	transformer	transformer	NOUN
cana-5106	45	59	-	-	PUNCT
cana-5106	45	60	based	base	VERB
cana-5106	45	61	object	object	NOUN
cana-5106	45	62	detection	detection	NOUN
cana-5106	45	63	by	by	ADP
cana-5106	45	64	reducing	reduce	VERB
cana-5106	45	65	input	input	NOUN
cana-5106	45	66	complexity	complexity	NOUN
cana-5106	45	67	to	to	ADP
cana-5106	45	68	a	a	DET
cana-5106	45	69	single	single	ADJ
cana-5106	45	70	sequence	sequence	NOUN
cana-5106	45	71	.	.	PUNCT
cana-5106	46	1	the	the	DET
cana-5106	46	2	study	study	NOUN
cana-5106	46	3	demonstrates	demonstrate	VERB
cana-5106	46	4	that	that	SCONJ
cana-5106	46	5	this	this	DET
cana-5106	46	6	method	method	NOUN
cana-5106	46	7	can	can	AUX
cana-5106	46	8	achieve	achieve	VERB
cana-5106	46	9	high	high	ADJ
cana-5106	46	10	-	-	PUNCT
cana-5106	46	11	quality	quality	NOUN
cana-5106	46	12	detection	detection	NOUN
cana-5106	46	13	results	result	NOUN
cana-5106	46	14	while	while	SCONJ
cana-5106	46	15	improving	improve	VERB
cana-5106	46	16	efficiency	efficiency	NOUN
cana-5106	46	17	.	.	PUNCT
cana-5106	47	1	the	the	DET
cana-5106	47	2	approach	approach	NOUN
cana-5106	47	3	’s	’s	PART
cana-5106	47	4	merits	merit	NOUN
cana-5106	47	5	include	include	VERB
cana-5106	47	6	reduced	reduce	VERB
cana-5106	47	7	computational	computational	ADJ
cana-5106	47	8	overhead	overhead	NOUN
cana-5106	47	9	and	and	CCONJ
cana-5106	47	10	faster	fast	ADJ
cana-5106	47	11	processing	processing	NOUN
cana-5106	47	12	,	,	PUNCT
cana-5106	47	13	aligning	align	VERB
cana-5106	47	14	with	with	ADP
cana-5106	47	15	our	our	PRON
cana-5106	47	16	project	project	NOUN
cana-5106	47	17	’s	’s	PART
cana-5106	47	18	goals	goal	NOUN
cana-5106	47	19	.	.	PUNCT
cana-5106	48	1	we	we	PRON
cana-5106	48	2	aim	aim	VERB
cana-5106	48	3	to	to	PART
cana-5106	48	4	adopt	adopt	VERB
cana-5106	48	5	this	this	DET
cana-5106	48	6	idea	idea	NOUN
cana-5106	48	7	of	of	ADP
cana-5106	48	8	simplified	simplified	ADJ
cana-5106	48	9	input	input	NOUN
cana-5106	48	10	representations	representation	NOUN
cana-5106	48	11	to	to	PART
cana-5106	48	12	streamline	streamline	VERB
cana-5106	48	13	our	our	PRON
cana-5106	48	14	model	model	NOUN
cana-5106	48	15	.	.	PUNCT
cana-5106	49	1	however	however	ADV
cana-5106	49	2	,	,	PUNCT
cana-5106	49	3	the	the	DET
cana-5106	49	4	sequential	sequential	ADJ
cana-5106	49	5	nature	nature	NOUN
cana-5106	49	6	of	of	ADP
cana-5106	49	7	the	the	DET
cana-5106	49	8	model	model	NOUN
cana-5106	49	9	may	may	AUX
cana-5106	49	10	limit	limit	VERB
cana-5106	49	11	its	its	PRON
cana-5106	49	12	ability	ability	NOUN
cana-5106	49	13	to	to	PART
cana-5106	49	14	capture	capture	VERB
cana-5106	49	15	intricate	intricate	ADJ
cana-5106	49	16	spatial	spatial	ADJ
cana-5106	49	17	relationships	relationship	NOUN
cana-5106	49	18	,	,	PUNCT
cana-5106	49	19	potentially	potentially	ADV
cana-5106	49	20	oversimplifying	oversimplify	VERB
cana-5106	49	21	complex	complex	ADJ
cana-5106	49	22	scenes	scene	NOUN
cana-5106	49	23	.	.	PUNCT
cana-5106	50	1	to	to	PART
cana-5106	50	2	address	address	VERB
cana-5106	50	3	this	this	PRON
cana-5106	50	4	,	,	PUNCT
cana-5106	50	5	our	our	PRON
cana-5106	50	6	project	project	NOUN
cana-5106	50	7	will	will	AUX
cana-5106	50	8	integrate	integrate	VERB
cana-5106	50	9	advanced	advanced	ADJ
cana-5106	50	10	attention	attention	NOUN
cana-5106	50	11	mechanisms	mechanism	NOUN
cana-5106	50	12	to	to	PART
cana-5106	50	13	preserve	preserve	VERB
cana-5106	50	14	spatial	spatial	ADJ
cana-5106	50	15	communications	communication	NOUN
cana-5106	50	16	on	on	ADP
cana-5106	50	17	applied	apply	VERB
cana-5106	50	18	nonlinear	nonlinear	ADJ
cana-5106	50	19	analysis	analysis	NOUN
cana-5106	50	20	issn	issn	NOUN
cana-5106	50	21	:	:	PUNCT
cana-5106	50	22	1074	1074	NUM
cana-5106	50	23	-	-	PUNCT
cana-5106	50	24	133x	133x	NUM
cana-5106	50	25	vol	vol	VERB
cana-5106	50	26	32	32	NUM
cana-5106	50	27	no	no	NOUN
cana-5106	50	28	.	.	PUNCT
cana-5106	50	29	10s	10	NOUN
cana-5106	50	30	(	(	PUNCT
cana-5106	50	31	2025	2025	NUM
cana-5106	50	32	)	)	PUNCT
cana-5106	50	33	1123	1123	NUM
cana-5106	50	34	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	50	35	information	information	NOUN
cana-5106	50	36	while	while	SCONJ
cana-5106	50	37	maintaining	maintain	VERB
cana-5106	50	38	efficiency	efficiency	NOUN
cana-5106	50	39	,	,	PUNCT
cana-5106	50	40	striking	strike	VERB
cana-5106	50	41	a	a	DET
cana-5106	50	42	balance	balance	NOUN
cana-5106	50	43	between	between	ADP
cana-5106	50	44	performance	performance	NOUN
cana-5106	50	45	and	and	CCONJ
cana-5106	50	46	complexity.[4	complexity.[4	NUM
cana-5106	50	47	]	]	X
cana-5106	50	48	5	5	NUM
cana-5106	50	49	)	)	PUNCT
cana-5106	50	50	in	in	ADP
cana-5106	50	51	“	"	PUNCT
cana-5106	50	52	tood	tood	PROPN
cana-5106	50	53	:	:	PUNCT
cana-5106	50	54	task	task	NOUN
cana-5106	50	55	-	-	PUNCT
cana-5106	50	56	aligned	align	VERB
cana-5106	50	57	one	one	NUM
cana-5106	50	58	-	-	PUNCT
cana-5106	50	59	stage	stage	NOUN
cana-5106	50	60	object	object	NOUN
cana-5106	50	61	detection	detection	NOUN
cana-5106	50	62	,	,	PUNCT
cana-5106	50	63	”	"	PUNCT
cana-5106	50	64	chengjian	chengjian	PROPN
cana-5106	50	65	feng	feng	PROPN
cana-5106	50	66	et	et	PROPN
cana-5106	50	67	al	al	PROPN
cana-5106	50	68	.	.	PROPN
cana-5106	51	1	present	present	VERB
cana-5106	51	2	a	a	DET
cana-5106	51	3	one	one	NUM
cana-5106	51	4	-	-	PUNCT
cana-5106	51	5	stage	stage	NOUN
cana-5106	51	6	object	object	NOUN
cana-5106	51	7	detection	detection	NOUN
cana-5106	51	8	framework	framework	NOUN
cana-5106	51	9	that	that	PRON
cana-5106	51	10	aligns	align	VERB
cana-5106	51	11	detection	detection	NOUN
cana-5106	51	12	tasks	task	NOUN
cana-5106	51	13	with	with	ADP
cana-5106	51	14	learning	learn	VERB
cana-5106	51	15	objectives	objective	NOUN
cana-5106	51	16	,	,	PUNCT
cana-5106	51	17	simplifying	simplify	VERB
cana-5106	51	18	the	the	DET
cana-5106	51	19	process	process	NOUN
cana-5106	51	20	while	while	SCONJ
cana-5106	51	21	improving	improve	VERB
cana-5106	51	22	accuracy	accuracy	NOUN
cana-5106	51	23	.	.	PUNCT
cana-5106	52	1	the	the	DET
cana-5106	52	2	approach	approach	NOUN
cana-5106	52	3	reduces	reduce	VERB
cana-5106	52	4	the	the	DET
cana-5106	52	5	complexity	complexity	NOUN
cana-5106	52	6	of	of	ADP
cana-5106	52	7	multi	multi	ADJ
cana-5106	52	8	-	-	ADJ
cana-5106	52	9	stage	stage	NOUN
cana-5106	52	10	systems	system	NOUN
cana-5106	52	11	and	and	CCONJ
cana-5106	52	12	enhances	enhance	VERB
cana-5106	52	13	inference	inference	NOUN
cana-5106	52	14	speed	speed	NOUN
cana-5106	52	15	,	,	PUNCT
cana-5106	52	16	making	make	VERB
cana-5106	52	17	it	it	PRON
cana-5106	52	18	suitable	suitable	ADJ
cana-5106	52	19	for	for	ADP
cana-5106	52	20	realtime	realtime	NOUN
cana-5106	52	21	applications	application	NOUN
cana-5106	52	22	.	.	PUNCT
cana-5106	53	1	we	we	PRON
cana-5106	53	2	plan	plan	VERB
cana-5106	53	3	to	to	PART
cana-5106	53	4	leverage	leverage	VERB
cana-5106	53	5	task	task	NOUN
cana-5106	53	6	alignment	alignment	NOUN
cana-5106	53	7	principles	principle	NOUN
cana-5106	53	8	in	in	ADP
cana-5106	53	9	our	our	PRON
cana-5106	53	10	project	project	NOUN
cana-5106	53	11	to	to	PART
cana-5106	53	12	better	well	ADV
cana-5106	53	13	integrate	integrate	VERB
cana-5106	53	14	visual	visual	ADJ
cana-5106	53	15	and	and	CCONJ
cana-5106	53	16	textual	textual	ADJ
cana-5106	53	17	information	information	NOUN
cana-5106	53	18	,	,	PUNCT
cana-5106	53	19	ensuring	ensure	VERB
cana-5106	53	20	captions	caption	NOUN
cana-5106	53	21	are	be	AUX
cana-5106	53	22	closely	closely	ADV
cana-5106	53	23	tied	tie	VERB
cana-5106	53	24	to	to	ADP
cana-5106	53	25	detected	detect	VERB
cana-5106	53	26	objects	object	NOUN
cana-5106	53	27	.	.	PUNCT
cana-5106	54	1	however	however	ADV
cana-5106	54	2	,	,	PUNCT
cana-5106	54	3	one	one	NUM
cana-5106	54	4	-	-	PUNCT
cana-5106	54	5	stage	stage	NOUN
cana-5106	54	6	models	model	NOUN
cana-5106	54	7	may	may	AUX
cana-5106	54	8	struggle	struggle	VERB
cana-5106	54	9	with	with	ADP
cana-5106	54	10	accuracy	accuracy	NOUN
cana-5106	54	11	in	in	ADP
cana-5106	54	12	complex	complex	ADJ
cana-5106	54	13	scenes	scene	NOUN
cana-5106	54	14	involving	involve	VERB
cana-5106	54	15	overlapping	overlap	VERB
cana-5106	54	16	objects	object	NOUN
cana-5106	54	17	.	.	PUNCT
cana-5106	55	1	to	to	PART
cana-5106	55	2	overcome	overcome	VERB
cana-5106	55	3	this	this	PRON
cana-5106	55	4	,	,	PUNCT
cana-5106	55	5	our	our	PRON
cana-5106	55	6	project	project	NOUN
cana-5106	55	7	will	will	AUX
cana-5106	55	8	incorporate	incorporate	VERB
cana-5106	55	9	transformerbased	transformerbase	VERB
cana-5106	55	10	attention	attention	NOUN
cana-5106	55	11	mechanisms	mechanism	NOUN
cana-5106	55	12	to	to	PART
cana-5106	55	13	handle	handle	VERB
cana-5106	55	14	complex	complex	ADJ
cana-5106	55	15	interactions	interaction	NOUN
cana-5106	55	16	,	,	PUNCT
cana-5106	55	17	improving	improve	VERB
cana-5106	55	18	caption	caption	NOUN
cana-5106	55	19	quality	quality	NOUN
cana-5106	55	20	while	while	SCONJ
cana-5106	55	21	retaining	retain	VERB
cana-5106	55	22	the	the	DET
cana-5106	55	23	benefits	benefit	NOUN
cana-5106	55	24	of	of	ADP
cana-5106	55	25	a	a	DET
cana-5106	55	26	simplified	simplified	ADJ
cana-5106	55	27	architecture.[5	architecture.[5	NOUN
cana-5106	55	28	]	]	X
cana-5106	55	29	6	6	NUM
cana-5106	55	30	)	)	PUNCT
cana-5106	55	31	peng	peng	PROPN
cana-5106	55	32	gao	gao	PROPN
cana-5106	55	33	et	et	PROPN
cana-5106	55	34	al	al	PROPN
cana-5106	55	35	.	.	PROPN
cana-5106	55	36	,	,	PUNCT
cana-5106	55	37	in	in	ADP
cana-5106	55	38	“	"	PUNCT
cana-5106	55	39	fast	fast	ADJ
cana-5106	55	40	convergence	convergence	NOUN
cana-5106	55	41	of	of	ADP
cana-5106	55	42	detr	detr	NOUN
cana-5106	55	43	with	with	ADP
cana-5106	55	44	spatially	spatially	ADV
cana-5106	55	45	modulated	modulate	VERB
cana-5106	55	46	coattention	coattention	NOUN
cana-5106	55	47	,	,	PUNCT
cana-5106	55	48	”	"	PUNCT
cana-5106	55	49	address	address	NOUN
cana-5106	55	50	the	the	DET
cana-5106	55	51	slow	slow	ADJ
cana-5106	55	52	convergence	convergence	NOUN
cana-5106	55	53	of	of	ADP
cana-5106	55	54	the	the	DET
cana-5106	55	55	detection	detection	NOUN
cana-5106	55	56	transformer	transformer	NOUN
cana-5106	55	57	(	(	PUNCT
cana-5106	55	58	detr	detr	NOUN
cana-5106	55	59	)	)	PUNCT
cana-5106	55	60	by	by	ADP
cana-5106	55	61	introducing	introduce	VERB
cana-5106	55	62	a	a	DET
cana-5106	55	63	spatially	spatially	ADV
cana-5106	55	64	modulated	modulate	VERB
cana-5106	55	65	co	co	ADJ
cana-5106	55	66	-	-	NOUN
cana-5106	55	67	attention	attention	ADJ
cana-5106	55	68	mechanism	mechanism	NOUN
cana-5106	55	69	.	.	PUNCT
cana-5106	56	1	this	this	DET
cana-5106	56	2	innovation	innovation	NOUN
cana-5106	56	3	improves	improve	VERB
cana-5106	56	4	feature	feature	NOUN
cana-5106	56	5	alignment	alignment	NOUN
cana-5106	56	6	and	and	CCONJ
cana-5106	56	7	accelerates	accelerate	VERB
cana-5106	56	8	training	training	NOUN
cana-5106	56	9	,	,	PUNCT
cana-5106	56	10	making	make	VERB
cana-5106	56	11	it	it	PRON
cana-5106	56	12	suitable	suitable	ADJ
cana-5106	56	13	for	for	ADP
cana-5106	56	14	real	real	ADJ
cana-5106	56	15	-	-	PUNCT
cana-5106	56	16	time	time	NOUN
cana-5106	56	17	applications	application	NOUN
cana-5106	56	18	.	.	PUNCT
cana-5106	57	1	we	we	PRON
cana-5106	57	2	plan	plan	VERB
cana-5106	57	3	to	to	PART
cana-5106	57	4	use	use	VERB
cana-5106	57	5	co	co	NOUN
cana-5106	57	6	-	-	NOUN
cana-5106	57	7	attention	attention	NOUN
cana-5106	57	8	in	in	ADP
cana-5106	57	9	our	our	PRON
cana-5106	57	10	project	project	NOUN
cana-5106	57	11	to	to	PART
cana-5106	57	12	enhance	enhance	VERB
cana-5106	57	13	the	the	DET
cana-5106	57	14	interaction	interaction	NOUN
cana-5106	57	15	between	between	ADP
cana-5106	57	16	visual	visual	ADJ
cana-5106	57	17	and	and	CCONJ
cana-5106	57	18	textual	textual	ADJ
cana-5106	57	19	modalities	modality	NOUN
cana-5106	57	20	,	,	PUNCT
cana-5106	57	21	improving	improve	VERB
cana-5106	57	22	caption	caption	PRON
cana-5106	57	23	accuracy	accuracy	NOUN
cana-5106	57	24	.	.	PUNCT
cana-5106	58	1	however	however	ADV
cana-5106	58	2	,	,	PUNCT
cana-5106	58	3	the	the	DET
cana-5106	58	4	added	add	VERB
cana-5106	58	5	complexity	complexity	NOUN
cana-5106	58	6	of	of	ADP
cana-5106	58	7	co	co	NOUN
cana-5106	58	8	-	-	NOUN
cana-5106	58	9	attention	attention	NOUN
cana-5106	58	10	may	may	AUX
cana-5106	58	11	increase	increase	VERB
cana-5106	58	12	computational	computational	ADJ
cana-5106	58	13	requirements	requirement	NOUN
cana-5106	58	14	.	.	PUNCT
cana-5106	59	1	to	to	PART
cana-5106	59	2	address	address	VERB
cana-5106	59	3	this	this	PRON
cana-5106	59	4	,	,	PUNCT
cana-5106	59	5	our	our	PRON
cana-5106	59	6	project	project	NOUN
cana-5106	59	7	will	will	AUX
cana-5106	59	8	focus	focus	VERB
cana-5106	59	9	on	on	ADP
cana-5106	59	10	a	a	DET
cana-5106	59	11	streamlined	streamlined	ADJ
cana-5106	59	12	design	design	NOUN
cana-5106	59	13	that	that	PRON
cana-5106	59	14	balances	balance	VERB
cana-5106	59	15	convergence	convergence	NOUN
cana-5106	59	16	speed	speed	NOUN
cana-5106	59	17	with	with	ADP
cana-5106	59	18	computational	computational	ADJ
cana-5106	59	19	efficiency	efficiency	NOUN
cana-5106	59	20	,	,	PUNCT
cana-5106	59	21	ensuring	ensure	VERB
cana-5106	59	22	robust	robust	ADJ
cana-5106	59	23	image	image	NOUN
cana-5106	59	24	captioning.[6	captioning.[6	NOUN
cana-5106	59	25	]	]	PUNCT
cana-5106	59	26	7	7	NUM
cana-5106	59	27	)	)	PUNCT
cana-5106	59	28	in	in	ADP
cana-5106	59	29	the	the	DET
cana-5106	59	30	paper	paper	NOUN
cana-5106	59	31	"	"	PUNCT
cana-5106	59	32	exploring	explore	VERB
cana-5106	59	33	region	region	NOUN
cana-5106	59	34	relationships	relationship	NOUN
cana-5106	59	35	implicitly	implicitly	ADV
cana-5106	59	36	:	:	PUNCT
cana-5106	59	37	image	image	NOUN
cana-5106	59	38	captioning	captioning	NOUN
cana-5106	59	39	with	with	ADP
cana-5106	59	40	visual	visual	ADJ
cana-5106	59	41	relationship	relationship	NOUN
cana-5106	59	42	attention	attention	NOUN
cana-5106	59	43	,	,	PUNCT
cana-5106	59	44	"	"	PUNCT
cana-5106	59	45	z.	z.	PROPN
cana-5106	59	46	zhang	zhang	PROPN
cana-5106	59	47	et	et	PROPN
cana-5106	59	48	al	al	PROPN
cana-5106	59	49	.	.	PROPN
cana-5106	60	1	introduce	introduce	VERB
cana-5106	60	2	a	a	DET
cana-5106	60	3	novel	novel	ADJ
cana-5106	60	4	approach	approach	NOUN
cana-5106	60	5	to	to	ADP
cana-5106	60	6	image	image	NOUN
cana-5106	60	7	captioning	captioning	NOUN
cana-5106	60	8	that	that	PRON
cana-5106	60	9	leverages	leverage	VERB
cana-5106	60	10	visual	visual	ADJ
cana-5106	60	11	relationship	relationship	NOUN
cana-5106	60	12	attention	attention	NOUN
cana-5106	60	13	to	to	PART
cana-5106	60	14	enhance	enhance	VERB
cana-5106	60	15	the	the	DET
cana-5106	60	16	understanding	understanding	NOUN
cana-5106	60	17	of	of	ADP
cana-5106	60	18	spatial	spatial	ADJ
cana-5106	60	19	and	and	CCONJ
cana-5106	60	20	contextual	contextual	ADJ
cana-5106	60	21	relationships	relationship	NOUN
cana-5106	60	22	between	between	ADP
cana-5106	60	23	objects	object	NOUN
cana-5106	60	24	in	in	ADP
cana-5106	60	25	an	an	DET
cana-5106	60	26	image.the	image.the	DET
cana-5106	60	27	approach	approach	NOUN
cana-5106	60	28	generates	generate	VERB
cana-5106	60	29	contextually	contextually	ADV
cana-5106	60	30	rich	rich	ADJ
cana-5106	60	31	captions	caption	NOUN
cana-5106	60	32	by	by	ADP
cana-5106	60	33	emphasizing	emphasize	VERB
cana-5106	60	34	object	object	NOUN
cana-5106	60	35	relationships	relationship	NOUN
cana-5106	60	36	,	,	PUNCT
cana-5106	60	37	aligning	align	VERB
cana-5106	60	38	with	with	ADP
cana-5106	60	39	our	our	PRON
cana-5106	60	40	project	project	NOUN
cana-5106	60	41	’s	’s	PART
cana-5106	60	42	objectives	objective	NOUN
cana-5106	60	43	.	.	PUNCT
cana-5106	61	1	we	we	PRON
cana-5106	61	2	plan	plan	VERB
cana-5106	61	3	to	to	PART
cana-5106	61	4	integrate	integrate	VERB
cana-5106	61	5	similar	similar	ADJ
cana-5106	61	6	attention	attention	NOUN
cana-5106	61	7	mechanisms	mechanism	NOUN
cana-5106	61	8	to	to	PART
cana-5106	61	9	enhance	enhance	VERB
cana-5106	61	10	caption	caption	NOUN
cana-5106	61	11	accuracy	accuracy	NOUN
cana-5106	61	12	and	and	CCONJ
cana-5106	61	13	depth	depth	NOUN
cana-5106	61	14	.	.	PUNCT
cana-5106	62	1	however	however	ADV
cana-5106	62	2	,	,	PUNCT
cana-5106	62	3	the	the	DET
cana-5106	62	4	added	add	VERB
cana-5106	62	5	complexity	complexity	NOUN
cana-5106	62	6	of	of	ADP
cana-5106	62	7	modeling	model	VERB
cana-5106	62	8	relationships	relationship	NOUN
cana-5106	62	9	may	may	AUX
cana-5106	62	10	increase	increase	VERB
cana-5106	62	11	computational	computational	ADJ
cana-5106	62	12	demands	demand	NOUN
cana-5106	62	13	.	.	PUNCT
cana-5106	63	1	our	our	PRON
cana-5106	63	2	project	project	NOUN
cana-5106	63	3	will	will	AUX
cana-5106	63	4	optimize	optimize	VERB
cana-5106	63	5	these	these	DET
cana-5106	63	6	mechanisms	mechanism	NOUN
cana-5106	63	7	for	for	ADP
cana-5106	63	8	efficiency	efficiency	NOUN
cana-5106	63	9	,	,	PUNCT
cana-5106	63	10	ensuring	ensure	VERB
cana-5106	63	11	that	that	SCONJ
cana-5106	63	12	relationship	relationship	NOUN
cana-5106	63	13	modeling	modeling	NOUN
cana-5106	63	14	does	do	AUX
cana-5106	63	15	not	not	PART
cana-5106	63	16	compromise	compromise	VERB
cana-5106	63	17	system	system	NOUN
cana-5106	63	18	performance.[6	performance.[6	NOUN
cana-5106	63	19	]	]	PUNCT
cana-5106	63	20	8)	8)	NUM
cana-5106	63	21	ziteng	ziteng	PROPN
cana-5106	63	22	gao	gao	PROPN
cana-5106	63	23	et	et	PROPN
cana-5106	63	24	al	al	PROPN
cana-5106	63	25	.	.	PROPN
cana-5106	63	26	,	,	PUNCT
cana-5106	63	27	in	in	ADP
cana-5106	63	28	“	"	PUNCT
cana-5106	63	29	mutual	mutual	ADJ
cana-5106	63	30	supervision	supervision	NOUN
cana-5106	63	31	for	for	ADP
cana-5106	63	32	dense	dense	ADJ
cana-5106	63	33	object	object	NOUN
cana-5106	63	34	detection	detection	NOUN
cana-5106	63	35	,	,	PUNCT
cana-5106	63	36	”	"	PUNCT
cana-5106	63	37	introduce	introduce	VERB
cana-5106	63	38	a	a	DET
cana-5106	63	39	framework	framework	NOUN
cana-5106	63	40	that	that	PRON
cana-5106	63	41	uses	use	VERB
cana-5106	63	42	mutual	mutual	ADJ
cana-5106	63	43	supervision	supervision	NOUN
cana-5106	63	44	to	to	PART
cana-5106	63	45	improve	improve	VERB
cana-5106	63	46	dense	dense	ADJ
cana-5106	63	47	object	object	NOUN
cana-5106	63	48	detection	detection	NOUN
cana-5106	63	49	.	.	PUNCT
cana-5106	64	1	the	the	DET
cana-5106	64	2	approach	approach	NOUN
cana-5106	64	3	enhances	enhance	VERB
cana-5106	64	4	accuracy	accuracy	NOUN
cana-5106	64	5	and	and	CCONJ
cana-5106	64	6	robustness	robustness	NOUN
cana-5106	64	7	by	by	ADP
cana-5106	64	8	integrating	integrate	VERB
cana-5106	64	9	multiple	multiple	ADJ
cana-5106	64	10	supervisory	supervisory	ADJ
cana-5106	64	11	signals	signal	NOUN
cana-5106	64	12	,	,	PUNCT
cana-5106	64	13	reducing	reduce	VERB
cana-5106	64	14	reliance	reliance	NOUN
cana-5106	64	15	on	on	ADP
cana-5106	64	16	extensive	extensive	ADJ
cana-5106	64	17	labeled	label	VERB
cana-5106	64	18	data	datum	NOUN
cana-5106	64	19	.	.	PUNCT
cana-5106	65	1	we	we	PRON
cana-5106	65	2	plan	plan	VERB
cana-5106	65	3	to	to	PART
cana-5106	65	4	adopt	adopt	VERB
cana-5106	65	5	mutual	mutual	ADJ
cana-5106	65	6	supervision	supervision	NOUN
cana-5106	65	7	principles	principle	NOUN
cana-5106	65	8	to	to	PART
cana-5106	65	9	strengthen	strengthen	VERB
cana-5106	65	10	the	the	DET
cana-5106	65	11	connection	connection	NOUN
cana-5106	65	12	between	between	ADP
cana-5106	65	13	visual	visual	ADJ
cana-5106	65	14	feature	feature	NOUN
cana-5106	65	15	extraction	extraction	NOUN
cana-5106	65	16	and	and	CCONJ
cana-5106	65	17	caption	caption	NOUN
cana-5106	65	18	generation	generation	NOUN
cana-5106	65	19	.	.	PUNCT
cana-5106	66	1	however	however	ADV
cana-5106	66	2	,	,	PUNCT
cana-5106	66	3	the	the	DET
cana-5106	66	4	framework	framework	NOUN
cana-5106	66	5	’s	’s	PART
cana-5106	66	6	complexity	complexity	NOUN
cana-5106	66	7	may	may	AUX
cana-5106	66	8	increase	increase	VERB
cana-5106	66	9	training	training	NOUN
cana-5106	66	10	time	time	NOUN
cana-5106	66	11	.	.	PUNCT
cana-5106	67	1	our	our	PRON
cana-5106	67	2	project	project	NOUN
cana-5106	67	3	will	will	AUX
cana-5106	67	4	optimize	optimize	VERB
cana-5106	67	5	the	the	DET
cana-5106	67	6	training	training	NOUN
cana-5106	67	7	process	process	NOUN
cana-5106	67	8	to	to	PART
cana-5106	67	9	balance	balance	VERB
cana-5106	67	10	mutual	mutual	ADJ
cana-5106	67	11	supervision	supervision	NOUN
cana-5106	67	12	benefits	benefit	NOUN
cana-5106	67	13	with	with	ADP
cana-5106	67	14	computational	computational	ADJ
cana-5106	67	15	efficiency	efficiency	NOUN
cana-5106	67	16	,	,	PUNCT
cana-5106	67	17	improving	improve	VERB
cana-5106	67	18	overall	overall	ADJ
cana-5106	67	19	captioning	captioning	NOUN
cana-5106	67	20	effectiveness.[8	effectiveness.[8	NOUN
cana-5106	67	21	]	]	X
cana-5106	67	22	9	9	NUM
cana-5106	67	23	)	)	PUNCT
cana-5106	67	24	in	in	ADP
cana-5106	67	25	“	"	PUNCT
cana-5106	67	26	unifying	unify	VERB
cana-5106	67	27	vision	vision	NOUN
cana-5106	67	28	-	-	PUNCT
cana-5106	67	29	and	and	CCONJ
cana-5106	67	30	-	-	PUNCT
cana-5106	67	31	language	language	NOUN
cana-5106	67	32	tasks	task	NOUN
cana-5106	67	33	via	via	ADP
cana-5106	67	34	text	text	NOUN
cana-5106	67	35	generation	generation	NOUN
cana-5106	67	36	,	,	PUNCT
cana-5106	67	37	”	"	PUNCT
cana-5106	67	38	jaemin	jaemin	PROPN
cana-5106	67	39	cho	cho	PROPN
cana-5106	67	40	et	et	PROPN
cana-5106	67	41	al	al	PROPN
cana-5106	67	42	.	.	PUNCT
cana-5106	68	1	propose	propose	VERB
cana-5106	68	2	a	a	DET
cana-5106	68	3	framework	framework	NOUN
cana-5106	68	4	that	that	PRON
cana-5106	68	5	unifies	unify	VERB
cana-5106	68	6	vision	vision	NOUN
cana-5106	68	7	-	-	PUNCT
cana-5106	68	8	and	and	CCONJ
cana-5106	68	9	-	-	PUNCT
cana-5106	68	10	language	language	NOUN
cana-5106	68	11	tasks	task	NOUN
cana-5106	68	12	through	through	ADP
cana-5106	68	13	a	a	DET
cana-5106	68	14	text	text	NOUN
cana-5106	68	15	generation	generation	NOUN
cana-5106	68	16	approach	approach	NOUN
cana-5106	68	17	.	.	PUNCT
cana-5106	69	1	the	the	DET
cana-5106	69	2	method	method	NOUN
cana-5106	69	3	simplifies	simplifie	NOUN
cana-5106	69	4	model	model	NOUN
cana-5106	69	5	architecture	architecture	NOUN
cana-5106	69	6	and	and	CCONJ
cana-5106	69	7	enhances	enhance	VERB
cana-5106	69	8	transferability	transferability	NOUN
cana-5106	69	9	across	across	ADP
cana-5106	69	10	communications	communication	NOUN
cana-5106	69	11	on	on	ADP
cana-5106	69	12	applied	apply	VERB
cana-5106	69	13	nonlinear	nonlinear	ADJ
cana-5106	69	14	analysis	analysis	NOUN
cana-5106	69	15	issn	issn	NOUN
cana-5106	69	16	:	:	PUNCT
cana-5106	69	17	1074	1074	NUM
cana-5106	69	18	-	-	PUNCT
cana-5106	69	19	133x	133x	NUM
cana-5106	69	20	vol	vol	VERB
cana-5106	69	21	32	32	NUM
cana-5106	69	22	no	no	NOUN
cana-5106	69	23	.	.	PUNCT
cana-5106	70	1	10s	10	NOUN
cana-5106	70	2	(	(	PUNCT
cana-5106	70	3	2025	2025	NUM
cana-5106	70	4	)	)	PUNCT
cana-5106	70	5	1124	1124	NUM
cana-5106	70	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	70	7	tasks	task	NOUN
cana-5106	70	8	,	,	PUNCT
cana-5106	70	9	making	make	VERB
cana-5106	70	10	it	it	PRON
cana-5106	70	11	highly	highly	ADV
cana-5106	70	12	applicable	applicable	ADJ
cana-5106	70	13	to	to	ADP
cana-5106	70	14	our	our	PRON
cana-5106	70	15	project	project	NOUN
cana-5106	70	16	.	.	PUNCT
cana-5106	71	1	we	we	PRON
cana-5106	71	2	plan	plan	VERB
cana-5106	71	3	to	to	PART
cana-5106	71	4	use	use	VERB
cana-5106	71	5	this	this	DET
cana-5106	71	6	unified	unified	ADJ
cana-5106	71	7	approach	approach	NOUN
cana-5106	71	8	to	to	PART
cana-5106	71	9	generate	generate	VERB
cana-5106	71	10	more	more	ADV
cana-5106	71	11	coherent	coherent	ADJ
cana-5106	71	12	and	and	CCONJ
cana-5106	71	13	contextually	contextually	ADV
cana-5106	71	14	relevant	relevant	ADJ
cana-5106	71	15	captions	caption	NOUN
cana-5106	71	16	.	.	PUNCT
cana-5106	72	1	however	however	ADV
cana-5106	72	2	,	,	PUNCT
cana-5106	72	3	optimizing	optimize	VERB
cana-5106	72	4	for	for	ADP
cana-5106	72	5	multiple	multiple	ADJ
cana-5106	72	6	tasks	task	NOUN
cana-5106	72	7	within	within	ADP
cana-5106	72	8	a	a	DET
cana-5106	72	9	single	single	ADJ
cana-5106	72	10	framework	framework	NOUN
cana-5106	72	11	may	may	AUX
cana-5106	72	12	impact	impact	VERB
cana-5106	72	13	performance	performance	NOUN
cana-5106	72	14	.	.	PUNCT
cana-5106	73	1	our	our	PRON
cana-5106	73	2	project	project	NOUN
cana-5106	73	3	will	will	AUX
cana-5106	73	4	carefully	carefully	ADV
cana-5106	73	5	design	design	VERB
cana-5106	73	6	the	the	DET
cana-5106	73	7	architecture	architecture	NOUN
cana-5106	73	8	to	to	PART
cana-5106	73	9	balance	balance	VERB
cana-5106	73	10	task	task	NOUN
cana-5106	73	11	performance	performance	NOUN
cana-5106	73	12	while	while	SCONJ
cana-5106	73	13	leveraging	leverage	VERB
cana-5106	73	14	unification	unification	NOUN
cana-5106	73	15	benefits	benefit	NOUN
cana-5106	73	16	,	,	PUNCT
cana-5106	73	17	ensuring	ensure	VERB
cana-5106	73	18	high	high	ADJ
cana-5106	73	19	-	-	PUNCT
cana-5106	73	20	quality	quality	NOUN
cana-5106	73	21	caption	caption	NOUN
cana-5106	73	22	generation.[9	generation.[9	NOUN
cana-5106	73	23	]	]	X
cana-5106	73	24	10	10	NUM
cana-5106	73	25	)	)	PUNCT
cana-5106	73	26	salman	salman	PROPN
cana-5106	73	27	khan	khan	PROPN
cana-5106	73	28	et	et	PROPN
cana-5106	73	29	al	al	PROPN
cana-5106	73	30	.	.	PROPN
cana-5106	73	31	,	,	PUNCT
cana-5106	73	32	in	in	ADP
cana-5106	73	33	“	"	PUNCT
cana-5106	73	34	transformers	transformer	NOUN
cana-5106	73	35	in	in	ADP
cana-5106	73	36	vision	vision	NOUN
cana-5106	73	37	:	:	PUNCT
cana-5106	73	38	a	a	DET
cana-5106	73	39	survey	survey	NOUN
cana-5106	73	40	,	,	PUNCT
cana-5106	73	41	”	"	PUNCT
cana-5106	73	42	provide	provide	VERB
cana-5106	73	43	a	a	DET
cana-5106	73	44	comprehensive	comprehensive	ADJ
cana-5106	73	45	review	review	NOUN
cana-5106	73	46	of	of	ADP
cana-5106	73	47	transformer	transformer	NOUN
cana-5106	73	48	applications	application	NOUN
cana-5106	73	49	in	in	ADP
cana-5106	73	50	computer	computer	NOUN
cana-5106	73	51	vision	vision	NOUN
cana-5106	73	52	.	.	PUNCT
cana-5106	74	1	the	the	DET
cana-5106	74	2	survey	survey	NOUN
cana-5106	74	3	highlights	highlight	NOUN
cana-5106	74	4	transformers	transformer	NOUN
cana-5106	74	5	’	'	PUNCT
cana-5106	74	6	strengths	strength	NOUN
cana-5106	74	7	in	in	ADP
cana-5106	74	8	capturing	capture	VERB
cana-5106	74	9	long	long	ADJ
cana-5106	74	10	-	-	PUNCT
cana-5106	74	11	range	range	NOUN
cana-5106	74	12	dependencies	dependency	NOUN
cana-5106	74	13	and	and	CCONJ
cana-5106	74	14	facilitating	facilitate	VERB
cana-5106	74	15	multi	multi	ADJ
cana-5106	74	16	-	-	ADJ
cana-5106	74	17	modal	modal	ADJ
cana-5106	74	18	learning	learning	NOUN
cana-5106	74	19	,	,	PUNCT
cana-5106	74	20	insights	insight	NOUN
cana-5106	74	21	crucial	crucial	ADJ
cana-5106	74	22	to	to	ADP
cana-5106	74	23	our	our	PRON
cana-5106	74	24	project	project	NOUN
cana-5106	74	25	.	.	PUNCT
cana-5106	75	1	we	we	PRON
cana-5106	75	2	plan	plan	VERB
cana-5106	75	3	to	to	PART
cana-5106	75	4	leverage	leverage	VERB
cana-5106	75	5	these	these	DET
cana-5106	75	6	findings	finding	NOUN
cana-5106	75	7	to	to	PART
cana-5106	75	8	enhance	enhance	VERB
cana-5106	75	9	our	our	PRON
cana-5106	75	10	captioning	captioning	NOUN
cana-5106	75	11	system	system	NOUN
cana-5106	75	12	.	.	PUNCT
cana-5106	76	1	however	however	ADV
cana-5106	76	2	,	,	PUNCT
cana-5106	76	3	the	the	DET
cana-5106	76	4	computational	computational	ADJ
cana-5106	76	5	complexity	complexity	NOUN
cana-5106	76	6	of	of	ADP
cana-5106	76	7	transformers	transformer	NOUN
cana-5106	76	8	poses	pose	VERB
cana-5106	76	9	challenges	challenge	NOUN
cana-5106	76	10	.	.	PUNCT
cana-5106	77	1	our	our	PRON
cana-5106	77	2	project	project	NOUN
cana-5106	77	3	will	will	AUX
cana-5106	77	4	explore	explore	VERB
cana-5106	77	5	optimization	optimization	NOUN
cana-5106	77	6	techniques	technique	NOUN
cana-5106	77	7	and	and	CCONJ
cana-5106	77	8	lightweight	lightweight	ADJ
cana-5106	77	9	designs	design	NOUN
cana-5106	77	10	to	to	PART
cana-5106	77	11	retain	retain	VERB
cana-5106	77	12	transformer	transformer	NOUN
cana-5106	77	13	advantages	advantage	NOUN
cana-5106	77	14	while	while	SCONJ
cana-5106	77	15	improving	improve	VERB
cana-5106	77	16	efficiency	efficiency	NOUN
cana-5106	77	17	,	,	PUNCT
cana-5106	77	18	ensuring	ensure	VERB
cana-5106	77	19	practical	practical	ADJ
cana-5106	77	20	real	real	ADJ
cana-5106	77	21	-	-	PUNCT
cana-5106	77	22	world	world	NOUN
cana-5106	77	23	applications.[10	applications.[10	NOUN
cana-5106	77	24	]	]	X
cana-5106	77	25	11	11	NUM
cana-5106	77	26	)	)	PUNCT
cana-5106	77	27	in	in	ADP
cana-5106	77	28	“	"	PUNCT
cana-5106	77	29	align	align	VERB
cana-5106	77	30	before	before	ADP
cana-5106	77	31	fuse	fuse	NOUN
cana-5106	77	32	:	:	PUNCT
cana-5106	77	33	vision	vision	NOUN
cana-5106	77	34	and	and	CCONJ
cana-5106	77	35	language	language	NOUN
cana-5106	77	36	representation	representation	NOUN
cana-5106	77	37	learning	learn	VERB
cana-5106	77	38	with	with	ADP
cana-5106	77	39	momentum	momentum	NOUN
cana-5106	77	40	distillation	distillation	NOUN
cana-5106	77	41	,	,	PUNCT
cana-5106	77	42	”	"	PUNCT
cana-5106	77	43	junnan	junnan	NOUN
cana-5106	77	44	li	li	PROPN
cana-5106	77	45	et	et	PROPN
cana-5106	77	46	al	al	PROPN
cana-5106	77	47	.	.	PROPN
cana-5106	78	1	propose	propose	VERB
cana-5106	78	2	a	a	DET
cana-5106	78	3	two	two	NUM
cana-5106	78	4	-	-	PUNCT
cana-5106	78	5	step	step	NOUN
cana-5106	78	6	approach	approach	NOUN
cana-5106	78	7	that	that	PRON
cana-5106	78	8	aligns	align	VERB
cana-5106	78	9	vision	vision	NOUN
cana-5106	78	10	and	and	CCONJ
cana-5106	78	11	language	language	NOUN
cana-5106	78	12	representations	representation	NOUN
cana-5106	78	13	before	before	ADP
cana-5106	78	14	fusion	fusion	NOUN
cana-5106	78	15	,	,	PUNCT
cana-5106	78	16	improving	improve	VERB
cana-5106	78	17	joint	joint	ADJ
cana-5106	78	18	model	model	NOUN
cana-5106	78	19	performance	performance	NOUN
cana-5106	78	20	.	.	PUNCT
cana-5106	79	1	the	the	DET
cana-5106	79	2	use	use	NOUN
cana-5106	79	3	of	of	ADP
cana-5106	79	4	momentum	momentum	NOUN
cana-5106	79	5	distillation	distillation	NOUN
cana-5106	79	6	ensures	ensure	VERB
cana-5106	79	7	robust	robust	ADJ
cana-5106	79	8	and	and	CCONJ
cana-5106	79	9	consistent	consistent	ADJ
cana-5106	79	10	representations	representation	NOUN
cana-5106	79	11	,	,	PUNCT
cana-5106	79	12	a	a	DET
cana-5106	79	13	strategy	strategy	NOUN
cana-5106	79	14	we	we	PRON
cana-5106	79	15	plan	plan	VERB
cana-5106	79	16	to	to	PART
cana-5106	79	17	incorporate	incorporate	VERB
cana-5106	79	18	for	for	ADP
cana-5106	79	19	better	well	ADJ
cana-5106	79	20	caption	caption	NOUN
cana-5106	79	21	coherence	coherence	NOUN
cana-5106	79	22	.	.	PUNCT
cana-5106	80	1	however	however	ADV
cana-5106	80	2	,	,	PUNCT
cana-5106	80	3	the	the	DET
cana-5106	80	4	separate	separate	ADJ
cana-5106	80	5	alignment	alignment	NOUN
cana-5106	80	6	and	and	CCONJ
cana-5106	80	7	fusion	fusion	NOUN
cana-5106	80	8	phases	phase	NOUN
cana-5106	80	9	may	may	AUX
cana-5106	80	10	increase	increase	VERB
cana-5106	80	11	training	training	NOUN
cana-5106	80	12	complexity	complexity	NOUN
cana-5106	80	13	.	.	PUNCT
cana-5106	81	1	our	our	PRON
cana-5106	81	2	project	project	NOUN
cana-5106	81	3	will	will	AUX
cana-5106	81	4	optimize	optimize	VERB
cana-5106	81	5	these	these	DET
cana-5106	81	6	stages	stage	NOUN
cana-5106	81	7	to	to	PART
cana-5106	81	8	ensure	ensure	VERB
cana-5106	81	9	efficient	efficient	ADJ
cana-5106	81	10	training	training	NOUN
cana-5106	81	11	while	while	SCONJ
cana-5106	81	12	maintaining	maintain	VERB
cana-5106	81	13	high	high	ADJ
cana-5106	81	14	-	-	PUNCT
cana-5106	81	15	quality	quality	NOUN
cana-5106	81	16	captioning	captioning	NOUN
cana-5106	81	17	,	,	PUNCT
cana-5106	81	18	balancing	balance	VERB
cana-5106	81	19	complexity	complexity	NOUN
cana-5106	81	20	and	and	CCONJ
cana-5106	81	21	performance.[11	performance.[11	PROPN
cana-5106	81	22	]	]	PUNCT
cana-5106	81	23	ii	ii	PROPN
cana-5106	81	24	.	.	PROPN
cana-5106	81	25	proposed	propose	VERB
cana-5106	81	26	work	work	NOUN
cana-5106	81	27	the	the	DET
cana-5106	81	28	proposed	propose	VERB
cana-5106	81	29	system	system	NOUN
cana-5106	81	30	integrates	integrate	VERB
cana-5106	81	31	vision	vision	NOUN
cana-5106	81	32	transformers	transformer	NOUN
cana-5106	81	33	(	(	PUNCT
cana-5106	81	34	vits	vit	NOUN
cana-5106	81	35	)	)	PUNCT
cana-5106	81	36	and	and	CCONJ
cana-5106	81	37	nlp	nlp	NOUN
cana-5106	81	38	transformers	transformer	NOUN
cana-5106	81	39	to	to	PART
cana-5106	81	40	create	create	VERB
cana-5106	81	41	a	a	DET
cana-5106	81	42	robust	robust	ADJ
cana-5106	81	43	and	and	CCONJ
cana-5106	81	44	intelligent	intelligent	ADJ
cana-5106	81	45	image	image	NOUN
cana-5106	81	46	captioning	captioning	NOUN
cana-5106	81	47	framework	framework	NOUN
cana-5106	81	48	.	.	PUNCT
cana-5106	82	1	the	the	DET
cana-5106	82	2	system	system	NOUN
cana-5106	82	3	operates	operate	VERB
cana-5106	82	4	in	in	ADP
cana-5106	82	5	two	two	NUM
cana-5106	82	6	key	key	ADJ
cana-5106	82	7	stages	stage	NOUN
cana-5106	82	8	.	.	PUNCT
cana-5106	83	1	in	in	ADP
cana-5106	83	2	the	the	DET
cana-5106	83	3	first	first	ADJ
cana-5106	83	4	stage	stage	NOUN
cana-5106	83	5	,	,	PUNCT
cana-5106	83	6	the	the	DET
cana-5106	83	7	vision	vision	NOUN
cana-5106	83	8	transformer	transformer	NOUN
cana-5106	83	9	processes	process	NOUN
cana-5106	83	10	input	input	NOUN
cana-5106	83	11	images	image	NOUN
cana-5106	83	12	by	by	ADP
cana-5106	83	13	dividing	divide	VERB
cana-5106	83	14	them	they	PRON
cana-5106	83	15	into	into	ADP
cana-5106	83	16	patches	patch	NOUN
cana-5106	83	17	and	and	CCONJ
cana-5106	83	18	extracting	extract	VERB
cana-5106	83	19	rich	rich	ADJ
cana-5106	83	20	visual	visual	ADJ
cana-5106	83	21	features	feature	NOUN
cana-5106	83	22	through	through	ADP
cana-5106	83	23	self	self	NOUN
cana-5106	83	24	-	-	PUNCT
cana-5106	83	25	attention	attention	NOUN
cana-5106	83	26	mechanisms	mechanism	NOUN
cana-5106	83	27	.	.	PUNCT
cana-5106	84	1	this	this	PRON
cana-5106	84	2	allows	allow	VERB
cana-5106	84	3	the	the	DET
cana-5106	84	4	model	model	NOUN
cana-5106	84	5	to	to	PART
cana-5106	84	6	effectively	effectively	ADV
cana-5106	84	7	capture	capture	VERB
cana-5106	84	8	complex	complex	ADJ
cana-5106	84	9	patterns	pattern	NOUN
cana-5106	84	10	,	,	PUNCT
cana-5106	84	11	relationships	relationship	NOUN
cana-5106	84	12	,	,	PUNCT
cana-5106	84	13	and	and	CCONJ
cana-5106	84	14	contextual	contextual	ADJ
cana-5106	84	15	details	detail	NOUN
cana-5106	84	16	within	within	ADP
cana-5106	84	17	the	the	DET
cana-5106	84	18	images	image	NOUN
cana-5106	84	19	.	.	PUNCT
cana-5106	85	1	in	in	ADP
cana-5106	85	2	the	the	DET
cana-5106	85	3	second	second	ADJ
cana-5106	85	4	stage	stage	NOUN
cana-5106	85	5	,	,	PUNCT
cana-5106	85	6	the	the	DET
cana-5106	85	7	extracted	extract	VERB
cana-5106	85	8	visual	visual	ADJ
cana-5106	85	9	features	feature	NOUN
cana-5106	85	10	are	be	AUX
cana-5106	85	11	passed	pass	VERB
cana-5106	85	12	to	to	ADP
cana-5106	85	13	an	an	DET
cana-5106	85	14	nlp	nlp	NOUN
cana-5106	85	15	transformer	transformer	NOUN
cana-5106	85	16	decoder	decoder	NOUN
cana-5106	85	17	,	,	PUNCT
cana-5106	85	18	which	which	PRON
cana-5106	85	19	generates	generate	VERB
cana-5106	85	20	coherent	coherent	ADJ
cana-5106	85	21	and	and	CCONJ
cana-5106	85	22	contextually	contextually	ADV
cana-5106	85	23	relevant	relevant	ADJ
cana-5106	85	24	captions	caption	NOUN
cana-5106	85	25	.	.	PUNCT
cana-5106	86	1	the	the	DET
cana-5106	86	2	decoder	decoder	NOUN
cana-5106	86	3	employs	employ	VERB
cana-5106	86	4	attention	attention	NOUN
cana-5106	86	5	mechanisms	mechanism	NOUN
cana-5106	86	6	to	to	PART
cana-5106	86	7	focus	focus	VERB
cana-5106	86	8	on	on	ADP
cana-5106	86	9	the	the	DET
cana-5106	86	10	most	most	ADV
cana-5106	86	11	important	important	ADJ
cana-5106	86	12	visual	visual	ADJ
cana-5106	86	13	cues	cue	NOUN
cana-5106	86	14	,	,	PUNCT
cana-5106	86	15	ensuring	ensure	VERB
cana-5106	86	16	the	the	DET
cana-5106	86	17	generated	generate	VERB
cana-5106	86	18	text	text	NOUN
cana-5106	86	19	accurately	accurately	ADV
cana-5106	86	20	reflects	reflect	VERB
cana-5106	86	21	the	the	DET
cana-5106	86	22	image	image	NOUN
cana-5106	86	23	's	's	PART
cana-5106	86	24	content	content	NOUN
cana-5106	86	25	and	and	CCONJ
cana-5106	86	26	dynamics	dynamic	NOUN
cana-5106	86	27	.	.	PUNCT
cana-5106	87	1	the	the	DET
cana-5106	87	2	system	system	NOUN
cana-5106	87	3	is	be	AUX
cana-5106	87	4	trained	train	VERB
cana-5106	87	5	using	use	VERB
cana-5106	87	6	a	a	DET
cana-5106	87	7	large	large	ADJ
cana-5106	87	8	dataset	dataset	NOUN
cana-5106	87	9	of	of	ADP
cana-5106	87	10	images	image	NOUN
cana-5106	87	11	paired	pair	VERB
cana-5106	87	12	with	with	ADP
cana-5106	87	13	captions	caption	NOUN
cana-5106	87	14	,	,	PUNCT
cana-5106	87	15	optimizing	optimize	VERB
cana-5106	87	16	the	the	DET
cana-5106	87	17	interaction	interaction	NOUN
cana-5106	87	18	between	between	ADP
cana-5106	87	19	the	the	DET
cana-5106	87	20	vision	vision	NOUN
cana-5106	87	21	transformer	transformer	NOUN
cana-5106	87	22	and	and	CCONJ
cana-5106	87	23	nlp	nlp	NOUN
cana-5106	87	24	transformer	transformer	NOUN
cana-5106	87	25	.	.	PUNCT
cana-5106	88	1	this	this	DET
cana-5106	88	2	synergy	synergy	NOUN
cana-5106	88	3	enhances	enhance	VERB
cana-5106	88	4	the	the	DET
cana-5106	88	5	quality	quality	NOUN
cana-5106	88	6	of	of	ADP
cana-5106	88	7	the	the	DET
cana-5106	88	8	generated	generate	VERB
cana-5106	88	9	descriptions	description	NOUN
cana-5106	88	10	,	,	PUNCT
cana-5106	88	11	enabling	enable	VERB
cana-5106	88	12	the	the	DET
cana-5106	88	13	system	system	NOUN
cana-5106	88	14	to	to	PART
cana-5106	88	15	produce	produce	VERB
cana-5106	88	16	detailed	detailed	ADJ
cana-5106	88	17	,	,	PUNCT
cana-5106	88	18	fluent	fluent	ADJ
cana-5106	88	19	captions	caption	NOUN
cana-5106	88	20	that	that	PRON
cana-5106	88	21	not	not	PART
cana-5106	88	22	only	only	ADV
cana-5106	88	23	identify	identify	VERB
cana-5106	88	24	objects	object	NOUN
cana-5106	88	25	but	but	CCONJ
cana-5106	88	26	also	also	ADV
cana-5106	88	27	describe	describe	VERB
cana-5106	88	28	their	their	PRON
cana-5106	88	29	interactions	interaction	NOUN
cana-5106	88	30	and	and	CCONJ
cana-5106	88	31	contextual	contextual	ADJ
cana-5106	88	32	relevance	relevance	NOUN
cana-5106	88	33	within	within	ADP
cana-5106	88	34	the	the	DET
cana-5106	88	35	scene	scene	NOUN
cana-5106	88	36	.	.	PUNCT
cana-5106	89	1	by	by	ADP
cana-5106	89	2	leveraging	leverage	VERB
cana-5106	89	3	the	the	DET
cana-5106	89	4	strengths	strength	NOUN
cana-5106	89	5	of	of	ADP
cana-5106	89	6	both	both	DET
cana-5106	89	7	architectures	architecture	NOUN
cana-5106	89	8	,	,	PUNCT
cana-5106	89	9	the	the	DET
cana-5106	89	10	proposed	propose	VERB
cana-5106	89	11	system	system	NOUN
cana-5106	89	12	aims	aim	VERB
cana-5106	89	13	to	to	PART
cana-5106	89	14	surpass	surpass	VERB
cana-5106	89	15	traditional	traditional	ADJ
cana-5106	89	16	image	image	NOUN
cana-5106	89	17	captioning	captioning	NOUN
cana-5106	89	18	methods	method	NOUN
cana-5106	89	19	,	,	PUNCT
cana-5106	89	20	offering	offer	VERB
cana-5106	89	21	a	a	DET
cana-5106	89	22	scalable	scalable	ADJ
cana-5106	89	23	and	and	CCONJ
cana-5106	89	24	efficient	efficient	ADJ
cana-5106	89	25	solution	solution	NOUN
cana-5106	89	26	.	.	PUNCT
cana-5106	90	1	potential	potential	ADJ
cana-5106	90	2	applications	application	NOUN
cana-5106	90	3	include	include	VERB
cana-5106	90	4	accessibility	accessibility	NOUN
cana-5106	90	5	tools	tool	NOUN
cana-5106	90	6	,	,	PUNCT
cana-5106	90	7	automated	automate	VERB
cana-5106	90	8	content	content	NOUN
cana-5106	90	9	generation	generation	NOUN
cana-5106	90	10	,	,	PUNCT
cana-5106	90	11	and	and	CCONJ
cana-5106	90	12	multimedia	multimedia	NOUN
cana-5106	90	13	search	search	NOUN
cana-5106	90	14	systems	system	NOUN
cana-5106	90	15	,	,	PUNCT
cana-5106	90	16	contributing	contribute	VERB
cana-5106	90	17	to	to	ADP
cana-5106	90	18	advancements	advancement	NOUN
cana-5106	90	19	in	in	ADP
cana-5106	90	20	multimodal	multimodal	NOUN
cana-5106	90	21	learning	learning	NOUN
cana-5106	90	22	.	.	PUNCT
cana-5106	91	1	communications	communication	NOUN
cana-5106	91	2	on	on	ADP
cana-5106	91	3	applied	apply	VERB
cana-5106	91	4	nonlinear	nonlinear	ADJ
cana-5106	91	5	analysis	analysis	NOUN
cana-5106	91	6	issn	issn	NOUN
cana-5106	91	7	:	:	PUNCT
cana-5106	91	8	1074	1074	NUM
cana-5106	91	9	-	-	PUNCT
cana-5106	91	10	133x	133x	NUM
cana-5106	91	11	vol	vol	VERB
cana-5106	91	12	32	32	NUM
cana-5106	91	13	no	no	NOUN
cana-5106	91	14	.	.	PUNCT
cana-5106	92	1	10s	10	NOUN
cana-5106	92	2	(	(	PUNCT
cana-5106	92	3	2025	2025	NUM
cana-5106	92	4	)	)	PUNCT
cana-5106	92	5	1125	1125	NUM
cana-5106	92	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	92	7	fig.1	fig.1	PROPN
cana-5106	92	8	.	.	PUNCT
cana-5106	92	9	system	system	NOUN
cana-5106	92	10	architecture	architecture	NOUN
cana-5106	92	11	iii	iii	X
cana-5106	92	12	.	.	PUNCT
cana-5106	93	1	methods	methods	PROPN
cana-5106	93	2	a.	a.	PROPN
cana-5106	93	3	dataset	dataset	PROPN
cana-5106	93	4	collection	collection	NOUN
cana-5106	93	5	:	:	PUNCT
cana-5106	93	6	datasets	dataset	NOUN
cana-5106	93	7	from	from	ADP
cana-5106	93	8	kaggle	kaggle	PROPN
cana-5106	93	9	,	,	PUNCT
cana-5106	93	10	like	like	ADP
cana-5106	93	11	ms	ms	PROPN
cana-5106	93	12	-	-	PUNCT
cana-5106	93	13	coco	coco	PROPN
cana-5106	93	14	or	or	CCONJ
cana-5106	93	15	other	other	ADJ
cana-5106	93	16	image	image	NOUN
cana-5106	93	17	-	-	PUNCT
cana-5106	93	18	captioning	caption	VERB
cana-5106	93	19	datasets	dataset	NOUN
cana-5106	93	20	,	,	PUNCT
cana-5106	93	21	are	be	AUX
cana-5106	93	22	utilized	utilize	VERB
cana-5106	93	23	to	to	PART
cana-5106	93	24	train	train	VERB
cana-5106	93	25	or	or	CCONJ
cana-5106	93	26	fine	fine	ADJ
cana-5106	93	27	-	-	PUNCT
cana-5106	93	28	tune	tune	NOUN
cana-5106	93	29	the	the	DET
cana-5106	93	30	smart	smart	ADJ
cana-5106	93	31	caption	caption	NOUN
cana-5106	93	32	’s	’s	PART
cana-5106	93	33	transformer	transformer	NOUN
cana-5106	93	34	model	model	NOUN
cana-5106	93	35	.	.	PUNCT
cana-5106	94	1	these	these	DET
cana-5106	94	2	datasets	dataset	NOUN
cana-5106	94	3	offer	offer	VERB
cana-5106	94	4	vast	vast	ADJ
cana-5106	94	5	amounts	amount	NOUN
cana-5106	94	6	of	of	ADP
cana-5106	94	7	labeled	label	VERB
cana-5106	94	8	images	image	NOUN
cana-5106	94	9	paired	pair	VERB
cana-5106	94	10	with	with	ADP
cana-5106	94	11	human	human	NOUN
cana-5106	94	12	-	-	PUNCT
cana-5106	94	13	generated	generate	VERB
cana-5106	94	14	captions	caption	NOUN
cana-5106	94	15	,	,	PUNCT
cana-5106	94	16	allowing	allow	VERB
cana-5106	94	17	the	the	DET
cana-5106	94	18	model	model	NOUN
cana-5106	94	19	to	to	PART
cana-5106	94	20	learn	learn	VERB
cana-5106	94	21	the	the	DET
cana-5106	94	22	relationships	relationship	NOUN
cana-5106	94	23	between	between	ADP
cana-5106	94	24	visual	visual	ADJ
cana-5106	94	25	content	content	NOUN
cana-5106	94	26	and	and	CCONJ
cana-5106	94	27	descriptive	descriptive	ADJ
cana-5106	94	28	language	language	NOUN
cana-5106	94	29	.	.	PUNCT
cana-5106	95	1	by	by	ADP
cana-5106	95	2	using	use	VERB
cana-5106	95	3	high	high	ADJ
cana-5106	95	4	-	-	PUNCT
cana-5106	95	5	quality	quality	NOUN
cana-5106	95	6	,	,	PUNCT
cana-5106	95	7	diverse	diverse	ADJ
cana-5106	95	8	datasets	dataset	NOUN
cana-5106	95	9	from	from	ADP
cana-5106	95	10	kaggle	kaggle	PROPN
cana-5106	95	11	,	,	PUNCT
cana-5106	95	12	the	the	DET
cana-5106	95	13	system	system	NOUN
cana-5106	95	14	is	be	AUX
cana-5106	95	15	able	able	ADJ
cana-5106	95	16	to	to	PART
cana-5106	95	17	generate	generate	VERB
cana-5106	95	18	accurate	accurate	ADJ
cana-5106	95	19	and	and	CCONJ
cana-5106	95	20	contextually	contextually	ADV
cana-5106	95	21	relevant	relevant	ADJ
cana-5106	95	22	captions	caption	NOUN
cana-5106	95	23	across	across	ADP
cana-5106	95	24	various	various	ADJ
cana-5106	95	25	image	image	NOUN
cana-5106	95	26	types	type	NOUN
cana-5106	95	27	,	,	PUNCT
cana-5106	95	28	enhancing	enhance	VERB
cana-5106	95	29	its	its	PRON
cana-5106	95	30	generalization	generalization	NOUN
cana-5106	95	31	and	and	CCONJ
cana-5106	95	32	performance	performance	NOUN
cana-5106	95	33	in	in	ADP
cana-5106	95	34	real	real	ADJ
cana-5106	95	35	-	-	PUNCT
cana-5106	95	36	world	world	NOUN
cana-5106	95	37	applications	application	NOUN
cana-5106	95	38	.	.	PUNCT
cana-5106	96	1	b.	b.	PROPN
cana-5106	96	2	methodology	methodology	PROPN
cana-5106	96	3	:	:	PUNCT
cana-5106	96	4	the	the	DET
cana-5106	96	5	methodology	methodology	NOUN
cana-5106	96	6	for	for	ADP
cana-5106	96	7	developing	develop	VERB
cana-5106	96	8	smart	smart	ADJ
cana-5106	96	9	caption	caption	NOUN
cana-5106	96	10	:	:	PUNCT
cana-5106	96	11	intelligent	intelligent	ADJ
cana-5106	96	12	image	image	NOUN
cana-5106	96	13	description	description	NOUN
cana-5106	96	14	using	use	VERB
cana-5106	96	15	transformer	transformer	NOUN
cana-5106	96	16	decoder	decoder	NOUN
cana-5106	96	17	follows	follow	VERB
cana-5106	96	18	a	a	DET
cana-5106	96	19	structured	structured	ADJ
cana-5106	96	20	approach	approach	NOUN
cana-5106	96	21	that	that	PRON
cana-5106	96	22	integrates	integrate	VERB
cana-5106	96	23	vision	vision	NOUN
cana-5106	96	24	transformers	transformer	NOUN
cana-5106	96	25	(	(	PUNCT
cana-5106	96	26	vits	vit	NOUN
cana-5106	96	27	)	)	PUNCT
cana-5106	96	28	and	and	CCONJ
cana-5106	96	29	nlp	nlp	NOUN
cana-5106	96	30	transformers	transformer	NOUN
cana-5106	96	31	to	to	PART
cana-5106	96	32	build	build	VERB
cana-5106	96	33	an	an	DET
cana-5106	96	34	advanced	advanced	ADJ
cana-5106	96	35	image	image	NOUN
cana-5106	96	36	captioning	captioning	NOUN
cana-5106	96	37	system	system	NOUN
cana-5106	96	38	.	.	PUNCT
cana-5106	97	1	the	the	DET
cana-5106	97	2	process	process	NOUN
cana-5106	97	3	begins	begin	VERB
cana-5106	97	4	with	with	ADP
cana-5106	97	5	data	datum	NOUN
cana-5106	97	6	collection	collection	NOUN
cana-5106	97	7	and	and	CCONJ
cana-5106	97	8	preprocessing	preprocessing	NOUN
cana-5106	97	9	,	,	PUNCT
cana-5106	97	10	where	where	SCONJ
cana-5106	97	11	a	a	DET
cana-5106	97	12	diverse	diverse	ADJ
cana-5106	97	13	dataset	dataset	NOUN
cana-5106	97	14	of	of	ADP
cana-5106	97	15	images	image	NOUN
cana-5106	97	16	paired	pair	VERB
cana-5106	97	17	with	with	ADP
cana-5106	97	18	corresponding	correspond	VERB
cana-5106	97	19	captions	caption	NOUN
cana-5106	97	20	is	be	AUX
cana-5106	97	21	curated	curate	VERB
cana-5106	97	22	.	.	PUNCT
cana-5106	98	1	the	the	DET
cana-5106	98	2	images	image	NOUN
cana-5106	98	3	are	be	AUX
cana-5106	98	4	divided	divide	VERB
cana-5106	98	5	into	into	ADP
cana-5106	98	6	patches	patch	NOUN
cana-5106	98	7	to	to	PART
cana-5106	98	8	align	align	VERB
cana-5106	98	9	with	with	ADP
cana-5106	98	10	the	the	DET
cana-5106	98	11	vision	vision	NOUN
cana-5106	98	12	transformer	transformer	NOUN
cana-5106	98	13	architecture	architecture	NOUN
cana-5106	98	14	,	,	PUNCT
cana-5106	98	15	enabling	enable	VERB
cana-5106	98	16	the	the	DET
cana-5106	98	17	extraction	extraction	NOUN
cana-5106	98	18	of	of	ADP
cana-5106	98	19	rich	rich	ADJ
cana-5106	98	20	visual	visual	ADJ
cana-5106	98	21	features	feature	NOUN
cana-5106	98	22	.	.	PUNCT
cana-5106	99	1	the	the	DET
cana-5106	99	2	vit	vit	NOUN
cana-5106	99	3	is	be	AUX
cana-5106	99	4	trained	train	VERB
cana-5106	99	5	to	to	PART
cana-5106	99	6	learn	learn	VERB
cana-5106	99	7	effective	effective	ADJ
cana-5106	99	8	representations	representation	NOUN
cana-5106	99	9	of	of	ADP
cana-5106	99	10	image	image	NOUN
cana-5106	99	11	content	content	NOUN
cana-5106	99	12	,	,	PUNCT
cana-5106	99	13	capturing	capture	VERB
cana-5106	99	14	complex	complex	ADJ
cana-5106	99	15	patterns	pattern	NOUN
cana-5106	99	16	,	,	PUNCT
cana-5106	99	17	object	object	VERB
cana-5106	99	18	relationships	relationship	NOUN
cana-5106	99	19	,	,	PUNCT
cana-5106	99	20	and	and	CCONJ
cana-5106	99	21	contextual	contextual	ADJ
cana-5106	99	22	details	detail	NOUN
cana-5106	99	23	within	within	ADP
cana-5106	99	24	the	the	DET
cana-5106	99	25	images.the	images.the	PRON
cana-5106	99	26	extracted	extract	VERB
cana-5106	99	27	visual	visual	ADJ
cana-5106	99	28	features	feature	NOUN
cana-5106	99	29	are	be	AUX
cana-5106	99	30	fed	feed	VERB
cana-5106	99	31	into	into	ADP
cana-5106	99	32	an	an	DET
cana-5106	99	33	nlp	nlp	NOUN
cana-5106	99	34	transformer	transformer	NOUN
cana-5106	99	35	decoder	decoder	NOUN
cana-5106	99	36	,	,	PUNCT
cana-5106	99	37	which	which	PRON
cana-5106	99	38	generates	generate	VERB
cana-5106	99	39	descriptive	descriptive	ADJ
cana-5106	99	40	captions	caption	NOUN
cana-5106	99	41	.	.	PUNCT
cana-5106	100	1	the	the	DET
cana-5106	100	2	decoder	decoder	NOUN
cana-5106	100	3	employs	employ	VERB
cana-5106	100	4	attention	attention	NOUN
cana-5106	100	5	mechanisms	mechanism	NOUN
cana-5106	100	6	to	to	PART
cana-5106	100	7	focus	focus	VERB
cana-5106	100	8	on	on	ADP
cana-5106	100	9	the	the	DET
cana-5106	100	10	most	most	ADV
cana-5106	100	11	relevant	relevant	ADJ
cana-5106	100	12	visual	visual	ADJ
cana-5106	100	13	elements	element	NOUN
cana-5106	100	14	,	,	PUNCT
cana-5106	100	15	ensuring	ensure	VERB
cana-5106	100	16	the	the	DET
cana-5106	100	17	generated	generate	VERB
cana-5106	100	18	text	text	NOUN
cana-5106	100	19	accurately	accurately	ADV
cana-5106	100	20	reflects	reflect	VERB
cana-5106	100	21	the	the	DET
cana-5106	100	22	image	image	NOUN
cana-5106	100	23	's	's	PART
cana-5106	100	24	content	content	NOUN
cana-5106	100	25	and	and	CCONJ
cana-5106	100	26	context	context	NOUN
cana-5106	100	27	.	.	PUNCT
cana-5106	101	1	the	the	DET
cana-5106	101	2	training	training	NOUN
cana-5106	101	3	process	process	NOUN
cana-5106	101	4	involves	involve	VERB
cana-5106	101	5	fine	fine	ADV
cana-5106	101	6	-	-	PUNCT
cana-5106	101	7	tuning	tune	VERB
cana-5106	101	8	both	both	CCONJ
cana-5106	101	9	the	the	DET
cana-5106	101	10	vit	vit	NOUN
cana-5106	101	11	and	and	CCONJ
cana-5106	101	12	nlp	nlp	NOUN
cana-5106	101	13	transformer	transformer	NOUN
cana-5106	101	14	using	use	VERB
cana-5106	101	15	loss	loss	NOUN
cana-5106	101	16	functions	function	NOUN
cana-5106	101	17	that	that	PRON
cana-5106	101	18	optimize	optimize	VERB
cana-5106	101	19	the	the	DET
cana-5106	101	20	alignment	alignment	NOUN
cana-5106	101	21	between	between	ADP
cana-5106	101	22	visual	visual	ADJ
cana-5106	101	23	inputs	input	NOUN
cana-5106	101	24	and	and	CCONJ
cana-5106	101	25	textual	textual	ADJ
cana-5106	101	26	outputs	output	NOUN
cana-5106	101	27	.	.	PUNCT
cana-5106	102	1	to	to	PART
cana-5106	102	2	evaluate	evaluate	VERB
cana-5106	102	3	the	the	DET
cana-5106	102	4	system	system	NOUN
cana-5106	102	5	's	's	PART
cana-5106	102	6	performance	performance	NOUN
cana-5106	102	7	,	,	PUNCT
cana-5106	102	8	metrics	metric	NOUN
cana-5106	102	9	such	such	ADJ
cana-5106	102	10	as	as	ADP
cana-5106	102	11	bleu	bleu	NOUN
cana-5106	102	12	,	,	PUNCT
cana-5106	102	13	meteor	meteor	NOUN
cana-5106	102	14	,	,	PUNCT
cana-5106	102	15	and	and	CCONJ
cana-5106	102	16	cider	cider	NOUN
cana-5106	102	17	are	be	AUX
cana-5106	102	18	used	use	VERB
cana-5106	102	19	to	to	PART
cana-5106	102	20	compare	compare	VERB
cana-5106	102	21	the	the	DET
cana-5106	102	22	generated	generate	VERB
cana-5106	102	23	captions	caption	NOUN
cana-5106	102	24	against	against	ADP
cana-5106	102	25	ground	ground	NOUN
cana-5106	102	26	truth	truth	NOUN
cana-5106	102	27	descriptions.through	descriptions.through	NOUN
cana-5106	102	28	iterative	iterative	NOUN
cana-5106	102	29	training	training	NOUN
cana-5106	102	30	and	and	CCONJ
cana-5106	102	31	evaluation	evaluation	NOUN
cana-5106	102	32	,	,	PUNCT
cana-5106	102	33	the	the	DET
cana-5106	102	34	methodology	methodology	NOUN
cana-5106	102	35	aims	aim	VERB
cana-5106	102	36	to	to	PART
cana-5106	102	37	improve	improve	VERB
cana-5106	102	38	the	the	DET
cana-5106	102	39	fluency	fluency	NOUN
cana-5106	102	40	,	,	PUNCT
cana-5106	102	41	accuracy	accuracy	NOUN
cana-5106	102	42	,	,	PUNCT
cana-5106	102	43	and	and	CCONJ
cana-5106	102	44	relevance	relevance	NOUN
cana-5106	102	45	of	of	ADP
cana-5106	102	46	the	the	DET
cana-5106	102	47	captions	caption	NOUN
cana-5106	102	48	.	.	PUNCT
cana-5106	103	1	this	this	DET
cana-5106	103	2	systematic	systematic	ADJ
cana-5106	103	3	approach	approach	NOUN
cana-5106	103	4	establishes	establish	VERB
cana-5106	103	5	a	a	DET
cana-5106	103	6	robust	robust	ADJ
cana-5106	103	7	framework	framework	NOUN
cana-5106	103	8	for	for	ADP
cana-5106	103	9	automated	automate	VERB
cana-5106	103	10	image	image	NOUN
cana-5106	103	11	description	description	NOUN
cana-5106	103	12	,	,	PUNCT
cana-5106	103	13	leveraging	leverage	VERB
cana-5106	103	14	the	the	DET
cana-5106	103	15	strengths	strength	NOUN
cana-5106	103	16	of	of	ADP
cana-5106	103	17	transformer	transformer	NOUN
cana-5106	103	18	-	-	PUNCT
cana-5106	103	19	based	base	VERB
cana-5106	103	20	architectures	architecture	NOUN
cana-5106	103	21	to	to	PART
cana-5106	103	22	deliver	deliver	VERB
cana-5106	103	23	highquality	highquality	NOUN
cana-5106	103	24	results	result	NOUN
cana-5106	103	25	.	.	PUNCT
cana-5106	104	1	communications	communication	NOUN
cana-5106	104	2	on	on	ADP
cana-5106	104	3	applied	apply	VERB
cana-5106	104	4	nonlinear	nonlinear	ADJ
cana-5106	104	5	analysis	analysis	NOUN
cana-5106	104	6	issn	issn	NOUN
cana-5106	104	7	:	:	PUNCT
cana-5106	104	8	1074	1074	NUM
cana-5106	104	9	-	-	PUNCT
cana-5106	104	10	133x	133x	NUM
cana-5106	104	11	vol	vol	VERB
cana-5106	104	12	32	32	NUM
cana-5106	104	13	no	no	NOUN
cana-5106	104	14	.	.	PUNCT
cana-5106	105	1	10s	10	NOUN
cana-5106	105	2	(	(	PUNCT
cana-5106	105	3	2025	2025	NUM
cana-5106	105	4	)	)	PUNCT
cana-5106	105	5	1126	1126	NUM
cana-5106	105	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	105	7	iv	iv	X
cana-5106	105	8	.	.	PUNCT
cana-5106	105	9	algorithm	algorithm	PROPN
cana-5106	105	10	a.	a.	PROPN
cana-5106	105	11	vision	vision	PROPN
cana-5106	105	12	transformers	transformer	NOUN
cana-5106	105	13	(	(	PUNCT
cana-5106	105	14	vit):the	vit):the	PROPN
cana-5106	105	15	vision	vision	PROPN
cana-5106	105	16	transformer	transformer	NOUN
cana-5106	105	17	(	(	PUNCT
cana-5106	105	18	vit	vit	NOUN
cana-5106	105	19	)	)	PUNCT
cana-5106	105	20	algorithm	algorithm	NOUN
cana-5106	105	21	is	be	AUX
cana-5106	105	22	a	a	DET
cana-5106	105	23	key	key	ADJ
cana-5106	105	24	element	element	NOUN
cana-5106	105	25	in	in	ADP
cana-5106	105	26	this	this	DET
cana-5106	105	27	paper	paper	NOUN
cana-5106	105	28	,	,	PUNCT
cana-5106	105	29	transforming	transform	VERB
cana-5106	105	30	the	the	DET
cana-5106	105	31	way	way	NOUN
cana-5106	105	32	visual	visual	ADJ
cana-5106	105	33	features	feature	NOUN
cana-5106	105	34	are	be	AUX
cana-5106	105	35	extracted	extract	VERB
cana-5106	105	36	from	from	ADP
cana-5106	105	37	images	image	NOUN
cana-5106	105	38	.	.	PUNCT
cana-5106	106	1	unlike	unlike	ADP
cana-5106	106	2	conventional	conventional	ADJ
cana-5106	106	3	convolutional	convolutional	ADJ
cana-5106	106	4	neural	neural	ADJ
cana-5106	106	5	networks	network	NOUN
cana-5106	106	6	(	(	PUNCT
cana-5106	106	7	cnns	cnns	PROPN
cana-5106	106	8	)	)	PUNCT
cana-5106	106	9	that	that	PRON
cana-5106	106	10	process	process	NOUN
cana-5106	106	11	images	image	NOUN
cana-5106	106	12	through	through	ADP
cana-5106	106	13	hierarchical	hierarchical	ADJ
cana-5106	106	14	layers	layer	NOUN
cana-5106	106	15	of	of	ADP
cana-5106	106	16	convolutions	convolution	NOUN
cana-5106	106	17	,	,	PUNCT
cana-5106	106	18	the	the	DET
cana-5106	106	19	vit	vit	NOUN
cana-5106	106	20	treats	treat	VERB
cana-5106	106	21	an	an	DET
cana-5106	106	22	image	image	NOUN
cana-5106	106	23	as	as	ADP
cana-5106	106	24	a	a	DET
cana-5106	106	25	sequence	sequence	NOUN
cana-5106	106	26	of	of	ADP
cana-5106	106	27	nonoverlapping	nonoverlapping	ADJ
cana-5106	106	28	patches	patch	NOUN
cana-5106	106	29	,	,	PUNCT
cana-5106	106	30	akin	akin	ADJ
cana-5106	106	31	to	to	ADP
cana-5106	106	32	how	how	SCONJ
cana-5106	106	33	words	word	NOUN
cana-5106	106	34	are	be	AUX
cana-5106	106	35	treated	treat	VERB
cana-5106	106	36	in	in	ADP
cana-5106	106	37	natural	natural	ADJ
cana-5106	106	38	language	language	NOUN
cana-5106	106	39	processing	processing	NOUN
cana-5106	106	40	.	.	PUNCT
cana-5106	107	1	this	this	DET
cana-5106	107	2	novel	novel	ADJ
cana-5106	107	3	approach	approach	NOUN
cana-5106	107	4	enables	enable	VERB
cana-5106	107	5	the	the	DET
cana-5106	107	6	vit	vit	NOUN
cana-5106	107	7	to	to	PART
cana-5106	107	8	capture	capture	VERB
cana-5106	107	9	both	both	DET
cana-5106	107	10	local	local	ADJ
cana-5106	107	11	and	and	CCONJ
cana-5106	107	12	global	global	ADJ
cana-5106	107	13	contextual	contextual	ADJ
cana-5106	107	14	information	information	NOUN
cana-5106	107	15	effectively	effectively	ADV
cana-5106	107	16	.	.	PUNCT
cana-5106	108	1	by	by	ADP
cana-5106	108	2	leveraging	leverage	VERB
cana-5106	108	3	self	self	NOUN
cana-5106	108	4	-	-	PUNCT
cana-5106	108	5	attention	attention	NOUN
cana-5106	108	6	mechanisms	mechanism	NOUN
cana-5106	108	7	,	,	PUNCT
cana-5106	108	8	the	the	DET
cana-5106	108	9	model	model	NOUN
cana-5106	108	10	can	can	AUX
cana-5106	108	11	assess	assess	VERB
cana-5106	108	12	the	the	DET
cana-5106	108	13	significance	significance	NOUN
cana-5106	108	14	of	of	ADP
cana-5106	108	15	different	different	ADJ
cana-5106	108	16	patches	patch	NOUN
cana-5106	108	17	in	in	ADP
cana-5106	108	18	relation	relation	NOUN
cana-5106	108	19	to	to	ADP
cana-5106	108	20	one	one	NUM
cana-5106	108	21	another	another	DET
cana-5106	108	22	,	,	PUNCT
cana-5106	108	23	allowing	allow	VERB
cana-5106	108	24	it	it	PRON
cana-5106	108	25	to	to	PART
cana-5106	108	26	focus	focus	VERB
cana-5106	108	27	on	on	ADP
cana-5106	108	28	crucial	crucial	ADJ
cana-5106	108	29	visual	visual	ADJ
cana-5106	108	30	elements	element	NOUN
cana-5106	108	31	and	and	CCONJ
cana-5106	108	32	their	their	PRON
cana-5106	108	33	relationships	relationship	NOUN
cana-5106	108	34	.	.	PUNCT
cana-5106	109	1	this	this	PRON
cana-5106	109	2	results	result	VERB
cana-5106	109	3	in	in	ADP
cana-5106	109	4	a	a	DET
cana-5106	109	5	deeper	deep	ADJ
cana-5106	109	6	understanding	understanding	NOUN
cana-5106	109	7	of	of	ADP
cana-5106	109	8	the	the	DET
cana-5106	109	9	image	image	NOUN
cana-5106	109	10	's	's	PART
cana-5106	109	11	content	content	NOUN
cana-5106	109	12	.	.	PUNCT
cana-5106	110	1	for	for	ADP
cana-5106	110	2	the	the	DET
cana-5106	110	3	smart	smart	ADJ
cana-5106	110	4	caption	caption	NOUN
cana-5106	110	5	system	system	NOUN
cana-5106	110	6	,	,	PUNCT
cana-5106	110	7	the	the	DET
cana-5106	110	8	vit	vit	NOUN
cana-5106	110	9	is	be	AUX
cana-5106	110	10	trained	train	VERB
cana-5106	110	11	on	on	ADP
cana-5106	110	12	an	an	DET
cana-5106	110	13	extensive	extensive	ADJ
cana-5106	110	14	image	image	NOUN
cana-5106	110	15	dataset	dataset	VERB
cana-5106	110	16	to	to	PART
cana-5106	110	17	learn	learn	VERB
cana-5106	110	18	representations	representation	NOUN
cana-5106	110	19	sensitive	sensitive	ADJ
cana-5106	110	20	to	to	ADP
cana-5106	110	21	variations	variation	NOUN
cana-5106	110	22	in	in	ADP
cana-5106	110	23	objects	object	NOUN
cana-5106	110	24	,	,	PUNCT
cana-5106	110	25	backgrounds	background	NOUN
cana-5106	110	26	,	,	PUNCT
cana-5106	110	27	and	and	CCONJ
cana-5106	110	28	complex	complex	ADJ
cana-5106	110	29	scene	scene	NOUN
cana-5106	110	30	interactions	interaction	NOUN
cana-5106	110	31	.	.	PUNCT
cana-5106	111	1	during	during	ADP
cana-5106	111	2	training	training	NOUN
cana-5106	111	3	,	,	PUNCT
cana-5106	111	4	the	the	DET
cana-5106	111	5	model	model	NOUN
cana-5106	111	6	is	be	AUX
cana-5106	111	7	optimized	optimize	VERB
cana-5106	111	8	to	to	PART
cana-5106	111	9	minimize	minimize	VERB
cana-5106	111	10	the	the	DET
cana-5106	111	11	difference	difference	NOUN
cana-5106	111	12	between	between	ADP
cana-5106	111	13	predicted	predict	VERB
cana-5106	111	14	and	and	CCONJ
cana-5106	111	15	actual	actual	ADJ
cana-5106	111	16	image	image	NOUN
cana-5106	111	17	features	feature	NOUN
cana-5106	111	18	.	.	PUNCT
cana-5106	112	1	once	once	ADV
cana-5106	112	2	trained	train	VERB
cana-5106	112	3	,	,	PUNCT
cana-5106	112	4	the	the	DET
cana-5106	112	5	vision	vision	NOUN
cana-5106	112	6	transformer	transformer	NOUN
cana-5106	112	7	produces	produce	VERB
cana-5106	112	8	high	high	ADJ
cana-5106	112	9	-	-	PUNCT
cana-5106	112	10	dimensional	dimensional	ADJ
cana-5106	112	11	feature	feature	NOUN
cana-5106	112	12	vectors	vector	NOUN
cana-5106	112	13	that	that	PRON
cana-5106	112	14	encapsulate	encapsulate	VERB
cana-5106	112	15	key	key	ADJ
cana-5106	112	16	visual	visual	ADJ
cana-5106	112	17	information	information	NOUN
cana-5106	112	18	,	,	PUNCT
cana-5106	112	19	which	which	PRON
cana-5106	112	20	are	be	AUX
cana-5106	112	21	then	then	ADV
cana-5106	112	22	passed	pass	VERB
cana-5106	112	23	to	to	ADP
cana-5106	112	24	the	the	DET
cana-5106	112	25	nlp	nlp	NOUN
cana-5106	112	26	transformer	transformer	NOUN
cana-5106	112	27	decoder	decoder	NOUN
cana-5106	112	28	.	.	PUNCT
cana-5106	113	1	the	the	DET
cana-5106	113	2	collaboration	collaboration	NOUN
cana-5106	113	3	between	between	ADP
cana-5106	113	4	the	the	DET
cana-5106	113	5	vit	vit	NOUN
cana-5106	113	6	and	and	CCONJ
cana-5106	113	7	the	the	DET
cana-5106	113	8	nlp	nlp	NOUN
cana-5106	113	9	transformer	transformer	NOUN
cana-5106	113	10	enhances	enhance	VERB
cana-5106	113	11	the	the	DET
cana-5106	113	12	generation	generation	NOUN
cana-5106	113	13	of	of	ADP
cana-5106	113	14	accurate	accurate	ADJ
cana-5106	113	15	,	,	PUNCT
cana-5106	113	16	context	context	NOUN
cana-5106	113	17	-	-	PUNCT
cana-5106	113	18	aware	aware	ADJ
cana-5106	113	19	captions	caption	NOUN
cana-5106	113	20	,	,	PUNCT
cana-5106	113	21	describing	describe	VERB
cana-5106	113	22	not	not	PART
cana-5106	113	23	only	only	ADV
cana-5106	113	24	the	the	DET
cana-5106	113	25	objects	object	NOUN
cana-5106	113	26	in	in	ADP
cana-5106	113	27	the	the	DET
cana-5106	113	28	image	image	NOUN
cana-5106	113	29	but	but	CCONJ
cana-5106	113	30	also	also	ADV
cana-5106	113	31	their	their	PRON
cana-5106	113	32	interactions	interaction	NOUN
cana-5106	113	33	and	and	CCONJ
cana-5106	113	34	relationships	relationship	NOUN
cana-5106	113	35	,	,	PUNCT
cana-5106	113	36	significantly	significantly	ADV
cana-5106	113	37	enriching	enrich	VERB
cana-5106	113	38	the	the	DET
cana-5106	113	39	quality	quality	NOUN
cana-5106	113	40	of	of	ADP
cana-5106	113	41	the	the	DET
cana-5106	113	42	generated	generate	VERB
cana-5106	113	43	descriptions	description	NOUN
cana-5106	113	44	.	.	PUNCT
cana-5106	114	1	b.	b.	PROPN
cana-5106	114	2	nlp	nlp	PROPN
cana-5106	114	3	transformer	transformer	PROPN
cana-5106	114	4	:	:	PUNCT
cana-5106	114	5	the	the	DET
cana-5106	114	6	nlp	nlp	PROPN
cana-5106	114	7	transformer	transformer	NOUN
cana-5106	114	8	algorithm	algorithm	NOUN
cana-5106	114	9	plays	play	VERB
cana-5106	114	10	a	a	DET
cana-5106	114	11	vital	vital	ADJ
cana-5106	114	12	role	role	NOUN
cana-5106	114	13	in	in	ADP
cana-5106	114	14	this	this	DET
cana-5106	114	15	paper	paper	NOUN
cana-5106	114	16	,	,	PUNCT
cana-5106	114	17	offering	offer	VERB
cana-5106	114	18	an	an	DET
cana-5106	114	19	advanced	advanced	ADJ
cana-5106	114	20	mechanism	mechanism	NOUN
cana-5106	114	21	for	for	ADP
cana-5106	114	22	creating	create	VERB
cana-5106	114	23	coherent	coherent	ADJ
cana-5106	114	24	,	,	PUNCT
cana-5106	114	25	contextually	contextually	ADV
cana-5106	114	26	relevant	relevant	ADJ
cana-5106	114	27	captions	caption	NOUN
cana-5106	114	28	from	from	ADP
cana-5106	114	29	the	the	DET
cana-5106	114	30	visual	visual	ADJ
cana-5106	114	31	features	feature	NOUN
cana-5106	114	32	provided	provide	VERB
cana-5106	114	33	by	by	ADP
cana-5106	114	34	the	the	DET
cana-5106	114	35	vision	vision	NOUN
cana-5106	114	36	transformer	transformer	NOUN
cana-5106	114	37	(	(	PUNCT
cana-5106	114	38	vit	vit	NOUN
cana-5106	114	39	)	)	PUNCT
cana-5106	114	40	.	.	PUNCT
cana-5106	115	1	this	this	DET
cana-5106	115	2	model	model	NOUN
cana-5106	115	3	uses	use	VERB
cana-5106	115	4	an	an	DET
cana-5106	115	5	attention	attention	NOUN
cana-5106	115	6	mechanism	mechanism	NOUN
cana-5106	115	7	to	to	PART
cana-5106	115	8	prioritize	prioritize	VERB
cana-5106	115	9	the	the	DET
cana-5106	115	10	importance	importance	NOUN
cana-5106	115	11	of	of	ADP
cana-5106	115	12	different	different	ADJ
cana-5106	115	13	words	word	NOUN
cana-5106	115	14	in	in	ADP
cana-5106	115	15	the	the	DET
cana-5106	115	16	input	input	NOUN
cana-5106	115	17	sequence	sequence	NOUN
cana-5106	115	18	while	while	SCONJ
cana-5106	115	19	generating	generate	VERB
cana-5106	115	20	output	output	NOUN
cana-5106	115	21	text	text	NOUN
cana-5106	115	22	.	.	PUNCT
cana-5106	116	1	by	by	ADP
cana-5106	116	2	focusing	focus	VERB
cana-5106	116	3	on	on	ADP
cana-5106	116	4	key	key	ADJ
cana-5106	116	5	visual	visual	ADJ
cana-5106	116	6	cues	cue	NOUN
cana-5106	116	7	and	and	CCONJ
cana-5106	116	8	their	their	PRON
cana-5106	116	9	relationships	relationship	NOUN
cana-5106	116	10	within	within	ADP
cana-5106	116	11	the	the	DET
cana-5106	116	12	feature	feature	NOUN
cana-5106	116	13	vectors	vector	NOUN
cana-5106	116	14	,	,	PUNCT
cana-5106	116	15	the	the	DET
cana-5106	116	16	nlp	nlp	NOUN
cana-5106	116	17	transformer	transformer	NOUN
cana-5106	116	18	constructs	construct	VERB
cana-5106	116	19	descriptive	descriptive	ADJ
cana-5106	116	20	sentences	sentence	NOUN
cana-5106	116	21	that	that	PRON
cana-5106	116	22	accurately	accurately	ADV
cana-5106	116	23	reflect	reflect	VERB
cana-5106	116	24	the	the	DET
cana-5106	116	25	image	image	NOUN
cana-5106	116	26	's	's	PART
cana-5106	116	27	content	content	NOUN
cana-5106	116	28	.	.	PUNCT
cana-5106	117	1	this	this	DET
cana-5106	117	2	approach	approach	NOUN
cana-5106	117	3	outperforms	outperform	VERB
cana-5106	117	4	traditional	traditional	ADJ
cana-5106	117	5	sequence	sequence	NOUN
cana-5106	117	6	models	model	NOUN
cana-5106	117	7	,	,	PUNCT
cana-5106	117	8	such	such	ADJ
cana-5106	117	9	as	as	ADP
cana-5106	117	10	rnns	rnn	NOUN
cana-5106	117	11	,	,	PUNCT
cana-5106	117	12	by	by	ADP
cana-5106	117	13	enabling	enable	VERB
cana-5106	117	14	parallel	parallel	ADJ
cana-5106	117	15	processing	processing	NOUN
cana-5106	117	16	and	and	CCONJ
cana-5106	117	17	effectively	effectively	ADV
cana-5106	117	18	managing	manage	VERB
cana-5106	117	19	long	long	ADJ
cana-5106	117	20	-	-	PUNCT
cana-5106	117	21	range	range	NOUN
cana-5106	117	22	dependencies	dependency	NOUN
cana-5106	117	23	in	in	ADP
cana-5106	117	24	language	language	NOUN
cana-5106	117	25	.	.	PUNCT
cana-5106	118	1	the	the	DET
cana-5106	118	2	architecture	architecture	NOUN
cana-5106	118	3	of	of	ADP
cana-5106	118	4	the	the	DET
cana-5106	118	5	nlp	nlp	NOUN
cana-5106	118	6	transformer	transformer	NOUN
cana-5106	118	7	consists	consist	VERB
cana-5106	118	8	of	of	ADP
cana-5106	118	9	multiple	multiple	ADJ
cana-5106	118	10	encoder	encoder	NOUN
cana-5106	118	11	and	and	CCONJ
cana-5106	118	12	decoder	decoder	NOUN
cana-5106	118	13	layers	layer	NOUN
cana-5106	118	14	,	,	PUNCT
cana-5106	118	15	with	with	ADP
cana-5106	118	16	the	the	DET
cana-5106	118	17	decoder	decoder	NOUN
cana-5106	118	18	specifically	specifically	ADV
cana-5106	118	19	designed	design	VERB
cana-5106	118	20	for	for	ADP
cana-5106	118	21	generating	generate	VERB
cana-5106	118	22	output	output	NOUN
cana-5106	118	23	text	text	NOUN
cana-5106	118	24	.	.	PUNCT
cana-5106	119	1	in	in	ADP
cana-5106	119	2	the	the	DET
cana-5106	119	3	context	context	NOUN
cana-5106	119	4	of	of	ADP
cana-5106	119	5	image	image	NOUN
cana-5106	119	6	captioning	captioning	NOUN
cana-5106	119	7	,	,	PUNCT
cana-5106	119	8	the	the	DET
cana-5106	119	9	process	process	NOUN
cana-5106	119	10	begins	begin	VERB
cana-5106	119	11	with	with	ADP
cana-5106	119	12	inputting	inputte	VERB
cana-5106	119	13	the	the	DET
cana-5106	119	14	vit	vit	NOUN
cana-5106	119	15	's	's	PART
cana-5106	119	16	output	output	NOUN
cana-5106	119	17	features	feature	VERB
cana-5106	119	18	into	into	ADP
cana-5106	119	19	the	the	DET
cana-5106	119	20	decoder	decoder	NOUN
cana-5106	119	21	.	.	PUNCT
cana-5106	120	1	the	the	DET
cana-5106	120	2	model	model	NOUN
cana-5106	120	3	generates	generate	VERB
cana-5106	120	4	captions	caption	NOUN
cana-5106	120	5	word	word	NOUN
cana-5106	120	6	by	by	ADP
cana-5106	120	7	word	word	NOUN
cana-5106	120	8	,	,	PUNCT
cana-5106	120	9	incorporating	incorporate	VERB
cana-5106	120	10	information	information	NOUN
cana-5106	120	11	from	from	ADP
cana-5106	120	12	previously	previously	ADV
cana-5106	120	13	generated	generate	VERB
cana-5106	120	14	words	word	NOUN
cana-5106	120	15	through	through	ADP
cana-5106	120	16	self	self	NOUN
cana-5106	120	17	-	-	PUNCT
cana-5106	120	18	attention	attention	NOUN
cana-5106	120	19	.	.	PUNCT
cana-5106	121	1	this	this	DET
cana-5106	121	2	iterative	iterative	NOUN
cana-5106	121	3	process	process	NOUN
cana-5106	121	4	ensures	ensure	VERB
cana-5106	121	5	that	that	SCONJ
cana-5106	121	6	the	the	DET
cana-5106	121	7	generated	generate	VERB
cana-5106	121	8	text	text	NOUN
cana-5106	121	9	is	be	AUX
cana-5106	121	10	relevant	relevant	ADJ
cana-5106	121	11	to	to	ADP
cana-5106	121	12	the	the	DET
cana-5106	121	13	visual	visual	ADJ
cana-5106	121	14	content	content	NOUN
cana-5106	121	15	while	while	SCONJ
cana-5106	121	16	maintaining	maintain	VERB
cana-5106	121	17	fluency	fluency	NOUN
cana-5106	121	18	and	and	CCONJ
cana-5106	121	19	coherence	coherence	NOUN
cana-5106	121	20	.	.	PUNCT
cana-5106	122	1	training	train	VERB
cana-5106	122	2	the	the	DET
cana-5106	122	3	nlp	nlp	NOUN
cana-5106	122	4	transformer	transformer	NOUN
cana-5106	122	5	involves	involve	VERB
cana-5106	122	6	optimizing	optimize	VERB
cana-5106	122	7	loss	loss	NOUN
cana-5106	122	8	functions	function	NOUN
cana-5106	122	9	that	that	PRON
cana-5106	122	10	evaluate	evaluate	VERB
cana-5106	122	11	the	the	DET
cana-5106	122	12	accuracy	accuracy	NOUN
cana-5106	122	13	of	of	ADP
cana-5106	122	14	the	the	DET
cana-5106	122	15	generated	generate	VERB
cana-5106	122	16	captions	caption	NOUN
cana-5106	122	17	against	against	ADP
cana-5106	122	18	ground	ground	NOUN
cana-5106	122	19	truth	truth	NOUN
cana-5106	122	20	descriptions	description	NOUN
cana-5106	122	21	,	,	PUNCT
cana-5106	122	22	enabling	enable	VERB
cana-5106	122	23	the	the	DET
cana-5106	122	24	smart	smart	ADJ
cana-5106	122	25	caption	caption	NOUN
cana-5106	122	26	system	system	NOUN
cana-5106	122	27	to	to	PART
cana-5106	122	28	produce	produce	VERB
cana-5106	122	29	high	high	ADJ
cana-5106	122	30	-	-	PUNCT
cana-5106	122	31	quality	quality	NOUN
cana-5106	122	32	,	,	PUNCT
cana-5106	122	33	intelligent	intelligent	ADJ
cana-5106	122	34	image	image	NOUN
cana-5106	122	35	descriptions	description	NOUN
cana-5106	122	36	that	that	PRON
cana-5106	122	37	enhance	enhance	VERB
cana-5106	122	38	user	user	NOUN
cana-5106	122	39	understanding	understanding	NOUN
cana-5106	122	40	and	and	CCONJ
cana-5106	122	41	accessibility	accessibility	NOUN
cana-5106	122	42	across	across	ADP
cana-5106	122	43	various	various	ADJ
cana-5106	122	44	applications	application	NOUN
cana-5106	122	45	.	.	PUNCT
cana-5106	123	1	communications	communication	NOUN
cana-5106	123	2	on	on	ADP
cana-5106	123	3	applied	apply	VERB
cana-5106	123	4	nonlinear	nonlinear	ADJ
cana-5106	123	5	analysis	analysis	NOUN
cana-5106	123	6	issn	issn	NOUN
cana-5106	123	7	:	:	PUNCT
cana-5106	123	8	1074	1074	NUM
cana-5106	123	9	-	-	PUNCT
cana-5106	123	10	133x	133x	NUM
cana-5106	123	11	vol	vol	VERB
cana-5106	123	12	32	32	NUM
cana-5106	123	13	no	no	NOUN
cana-5106	123	14	.	.	PUNCT
cana-5106	124	1	10s	10	NOUN
cana-5106	124	2	(	(	PUNCT
cana-5106	124	3	2025	2025	NUM
cana-5106	124	4	)	)	PUNCT
cana-5106	124	5	1127	1127	NUM
cana-5106	124	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	124	7	v.	v.	ADP
cana-5106	124	8	results	result	NOUN
cana-5106	124	9	and	and	CCONJ
cana-5106	124	10	discussion	discussion	VERB
cana-5106	124	11	the	the	DET
cana-5106	124	12	results	result	NOUN
cana-5106	124	13	of	of	ADP
cana-5106	124	14	the	the	DET
cana-5106	124	15	smart	smart	ADJ
cana-5106	124	16	caption	caption	NOUN
cana-5106	124	17	:	:	PUNCT
cana-5106	124	18	intelligent	intelligent	ADJ
cana-5106	124	19	image	image	NOUN
cana-5106	124	20	description	description	NOUN
cana-5106	124	21	using	use	VERB
cana-5106	124	22	transformer	transformer	NOUN
cana-5106	124	23	decoder	decoder	NOUN
cana-5106	124	24	system	system	NOUN
cana-5106	124	25	demonstrate	demonstrate	VERB
cana-5106	124	26	its	its	PRON
cana-5106	124	27	effectiveness	effectiveness	NOUN
cana-5106	124	28	in	in	ADP
cana-5106	124	29	generating	generate	VERB
cana-5106	124	30	accurate	accurate	ADJ
cana-5106	124	31	,	,	PUNCT
cana-5106	124	32	contextually	contextually	ADV
cana-5106	124	33	rich	rich	ADJ
cana-5106	124	34	,	,	PUNCT
cana-5106	124	35	and	and	CCONJ
cana-5106	124	36	fluent	fluent	ADJ
cana-5106	124	37	captions	caption	NOUN
cana-5106	124	38	for	for	ADP
cana-5106	124	39	a	a	DET
cana-5106	124	40	wide	wide	ADJ
cana-5106	124	41	range	range	NOUN
cana-5106	124	42	of	of	ADP
cana-5106	124	43	images	image	NOUN
cana-5106	124	44	.	.	PUNCT
cana-5106	125	1	leveraging	leverage	VERB
cana-5106	125	2	the	the	DET
cana-5106	125	3	vision	vision	NOUN
cana-5106	125	4	transformer	transformer	NOUN
cana-5106	125	5	(	(	PUNCT
cana-5106	125	6	vit	vit	NOUN
cana-5106	125	7	)	)	PUNCT
cana-5106	125	8	for	for	ADP
cana-5106	125	9	robust	robust	ADJ
cana-5106	125	10	visual	visual	ADJ
cana-5106	125	11	feature	feature	NOUN
cana-5106	125	12	extraction	extraction	NOUN
cana-5106	125	13	and	and	CCONJ
cana-5106	125	14	the	the	DET
cana-5106	125	15	nlp	nlp	NOUN
cana-5106	125	16	transformer	transformer	NOUN
cana-5106	125	17	for	for	ADP
cana-5106	125	18	coherent	coherent	ADJ
cana-5106	125	19	text	text	NOUN
cana-5106	125	20	generation	generation	NOUN
cana-5106	125	21	,	,	PUNCT
cana-5106	125	22	the	the	DET
cana-5106	125	23	system	system	NOUN
cana-5106	125	24	achieves	achieve	VERB
cana-5106	125	25	state	state	NOUN
cana-5106	125	26	-	-	PUNCT
cana-5106	125	27	of	of	ADP
cana-5106	125	28	-	-	PUNCT
cana-5106	125	29	the	the	DET
cana-5106	125	30	-	-	PUNCT
cana-5106	125	31	art	art	NOUN
cana-5106	125	32	performance	performance	NOUN
cana-5106	125	33	on	on	ADP
cana-5106	125	34	benchmark	benchmark	ADJ
cana-5106	125	35	datasets	dataset	NOUN
cana-5106	125	36	,	,	PUNCT
cana-5106	125	37	as	as	SCONJ
cana-5106	125	38	evidenced	evidence	VERB
cana-5106	125	39	by	by	ADP
cana-5106	125	40	high	high	ADJ
cana-5106	125	41	scores	score	NOUN
cana-5106	125	42	on	on	ADP
cana-5106	125	43	evaluation	evaluation	NOUN
cana-5106	125	44	metrics	metric	NOUN
cana-5106	125	45	such	such	ADJ
cana-5106	125	46	as	as	ADP
cana-5106	125	47	bleu	bleu	NOUN
cana-5106	125	48	,	,	PUNCT
cana-5106	125	49	meteor	meteor	NOUN
cana-5106	125	50	,	,	PUNCT
cana-5106	125	51	and	and	CCONJ
cana-5106	125	52	cider	cider	NOUN
cana-5106	125	53	.	.	PUNCT
cana-5106	126	1	the	the	DET
cana-5106	126	2	vit	vit	NOUN
cana-5106	126	3	's	's	PART
cana-5106	126	4	ability	ability	NOUN
cana-5106	126	5	to	to	PART
cana-5106	126	6	capture	capture	VERB
cana-5106	126	7	intricate	intricate	ADJ
cana-5106	126	8	object	object	NOUN
cana-5106	126	9	relationships	relationship	NOUN
cana-5106	126	10	and	and	CCONJ
cana-5106	126	11	contextual	contextual	ADJ
cana-5106	126	12	details	detail	NOUN
cana-5106	126	13	,	,	PUNCT
cana-5106	126	14	combined	combine	VERB
cana-5106	126	15	with	with	ADP
cana-5106	126	16	the	the	DET
cana-5106	126	17	nlp	nlp	NOUN
cana-5106	126	18	transformer	transformer	NOUN
cana-5106	126	19	's	's	PART
cana-5106	126	20	capacity	capacity	NOUN
cana-5106	126	21	to	to	PART
cana-5106	126	22	produce	produce	VERB
cana-5106	126	23	linguistically	linguistically	ADV
cana-5106	126	24	accurate	accurate	ADJ
cana-5106	126	25	descriptions	description	NOUN
cana-5106	126	26	,	,	PUNCT
cana-5106	126	27	ensures	ensure	VERB
cana-5106	126	28	that	that	SCONJ
cana-5106	126	29	the	the	DET
cana-5106	126	30	generated	generate	VERB
cana-5106	126	31	captions	caption	NOUN
cana-5106	126	32	are	be	AUX
cana-5106	126	33	not	not	PART
cana-5106	126	34	only	only	ADV
cana-5106	126	35	precise	precise	ADJ
cana-5106	126	36	but	but	CCONJ
cana-5106	126	37	also	also	ADV
cana-5106	126	38	contextually	contextually	ADV
cana-5106	126	39	meaningful	meaningful	ADJ
cana-5106	126	40	.	.	PUNCT
cana-5106	127	1	qualitative	qualitative	ADJ
cana-5106	127	2	analysis	analysis	NOUN
cana-5106	127	3	reveals	reveal	VERB
cana-5106	127	4	that	that	SCONJ
cana-5106	127	5	the	the	DET
cana-5106	127	6	system	system	NOUN
cana-5106	127	7	excels	excel	VERB
cana-5106	127	8	in	in	ADP
cana-5106	127	9	describing	describe	VERB
cana-5106	127	10	complex	complex	ADJ
cana-5106	127	11	scenes	scene	NOUN
cana-5106	127	12	,	,	PUNCT
cana-5106	127	13	including	include	VERB
cana-5106	127	14	interactions	interaction	NOUN
cana-5106	127	15	between	between	ADP
cana-5106	127	16	objects	object	NOUN
cana-5106	127	17	and	and	CCONJ
cana-5106	127	18	nuanced	nuanced	ADJ
cana-5106	127	19	visual	visual	ADJ
cana-5106	127	20	elements	element	NOUN
cana-5106	127	21	,	,	PUNCT
cana-5106	127	22	which	which	PRON
cana-5106	127	23	traditional	traditional	ADJ
cana-5106	127	24	methods	method	NOUN
cana-5106	127	25	often	often	ADV
cana-5106	127	26	struggle	struggle	VERB
cana-5106	127	27	to	to	PART
cana-5106	127	28	capture	capture	VERB
cana-5106	127	29	.	.	PUNCT
cana-5106	128	1	however	however	ADV
cana-5106	128	2	,	,	PUNCT
cana-5106	128	3	challenges	challenge	NOUN
cana-5106	128	4	remain	remain	VERB
cana-5106	128	5	in	in	ADP
cana-5106	128	6	handling	handle	VERB
cana-5106	128	7	highly	highly	ADV
cana-5106	128	8	abstract	abstract	ADJ
cana-5106	128	9	or	or	CCONJ
cana-5106	128	10	ambiguous	ambiguous	ADJ
cana-5106	128	11	images	image	NOUN
cana-5106	128	12	,	,	PUNCT
cana-5106	128	13	where	where	SCONJ
cana-5106	128	14	the	the	DET
cana-5106	128	15	system	system	NOUN
cana-5106	128	16	occasionally	occasionally	ADV
cana-5106	128	17	produces	produce	VERB
cana-5106	128	18	less	less	ADV
cana-5106	128	19	accurate	accurate	ADJ
cana-5106	128	20	descriptions	description	NOUN
cana-5106	128	21	.	.	PUNCT
cana-5106	129	1	overall	overall	ADV
cana-5106	129	2	,	,	PUNCT
cana-5106	129	3	the	the	DET
cana-5106	129	4	results	result	NOUN
cana-5106	129	5	highlight	highlight	VERB
cana-5106	129	6	the	the	DET
cana-5106	129	7	system	system	NOUN
cana-5106	129	8	's	's	PART
cana-5106	129	9	potential	potential	NOUN
cana-5106	129	10	for	for	ADP
cana-5106	129	11	applications	application	NOUN
cana-5106	129	12	in	in	ADP
cana-5106	129	13	accessibility	accessibility	NOUN
cana-5106	129	14	tools	tool	NOUN
cana-5106	129	15	,	,	PUNCT
cana-5106	129	16	automated	automate	VERB
cana-5106	129	17	content	content	NOUN
cana-5106	129	18	generation	generation	NOUN
cana-5106	129	19	,	,	PUNCT
cana-5106	129	20	and	and	CCONJ
cana-5106	129	21	multimedia	multimedia	NOUN
cana-5106	129	22	search	search	NOUN
cana-5106	129	23	systems	system	NOUN
cana-5106	129	24	,	,	PUNCT
cana-5106	129	25	while	while	SCONJ
cana-5106	129	26	also	also	ADV
cana-5106	129	27	identifying	identify	VERB
cana-5106	129	28	areas	area	NOUN
cana-5106	129	29	for	for	ADP
cana-5106	129	30	future	future	ADJ
cana-5106	129	31	improvement	improvement	NOUN
cana-5106	129	32	,	,	PUNCT
cana-5106	129	33	such	such	ADJ
cana-5106	129	34	as	as	ADP
cana-5106	129	35	enhancing	enhance	VERB
cana-5106	129	36	robustness	robustness	NOUN
cana-5106	129	37	for	for	ADP
cana-5106	129	38	diverse	diverse	ADJ
cana-5106	129	39	and	and	CCONJ
cana-5106	129	40	ambiguous	ambiguous	ADJ
cana-5106	129	41	visual	visual	ADJ
cana-5106	129	42	inputs	input	NOUN
cana-5106	129	43	.	.	PUNCT
cana-5106	130	1	fig	fig	NOUN
cana-5106	130	2	:	:	PUNCT
cana-5106	130	3	welcome	welcome	ADJ
cana-5106	130	4	page	page	NOUN
cana-5106	130	5	this	this	DET
cana-5106	130	6	image	image	NOUN
cana-5106	130	7	is	be	AUX
cana-5106	130	8	a	a	DET
cana-5106	130	9	shows	show	NOUN
cana-5106	130	10	"	"	PUNCT
cana-5106	130	11	smart	smart	ADJ
cana-5106	130	12	caption	caption	NOUN
cana-5106	130	13	,	,	PUNCT
cana-5106	130	14	"	"	PUNCT
cana-5106	130	15	an	an	DET
cana-5106	130	16	ai	ai	NOUN
cana-5106	130	17	-	-	PUNCT
cana-5106	130	18	powered	power	VERB
cana-5106	130	19	tool	tool	NOUN
cana-5106	130	20	that	that	PRON
cana-5106	130	21	generates	generate	VERB
cana-5106	130	22	image	image	NOUN
cana-5106	130	23	descriptions	description	NOUN
cana-5106	130	24	using	use	VERB
cana-5106	130	25	a	a	DET
cana-5106	130	26	transformer	transformer	NOUN
cana-5106	130	27	decoder	decoder	NOUN
cana-5106	130	28	.	.	PUNCT
cana-5106	131	1	the	the	DET
cana-5106	131	2	banner	banner	NOUN
cana-5106	131	3	features	feature	VERB
cana-5106	131	4	a	a	DET
cana-5106	131	5	clean	clean	ADJ
cana-5106	131	6	,	,	PUNCT
cana-5106	131	7	minimalist	minimalist	ADJ
cana-5106	131	8	design	design	NOUN
cana-5106	131	9	with	with	ADP
cana-5106	131	10	bold	bold	ADJ
cana-5106	131	11	typography	typography	NOUN
cana-5106	131	12	emphasizing	emphasize	VERB
cana-5106	131	13	the	the	DET
cana-5106	131	14	tool	tool	NOUN
cana-5106	131	15	's	's	PART
cana-5106	131	16	ability	ability	NOUN
cana-5106	131	17	to	to	PART
cana-5106	131	18	"	"	PUNCT
cana-5106	131	19	transform	transform	VERB
cana-5106	131	20	images	image	NOUN
cana-5106	131	21	into	into	ADP
cana-5106	131	22	captions	caption	NOUN
cana-5106	131	23	.	.	PUNCT
cana-5106	131	24	"	"	PUNCT
cana-5106	132	1	it	it	PRON
cana-5106	132	2	highlights	highlight	VERB
cana-5106	132	3	the	the	DET
cana-5106	132	4	use	use	NOUN
cana-5106	132	5	of	of	ADP
cana-5106	132	6	artificial	artificial	ADJ
cana-5106	132	7	intelligence	intelligence	NOUN
cana-5106	132	8	(	(	PUNCT
cana-5106	132	9	"	"	PUNCT
cana-5106	132	10	powered	power	VERB
cana-5106	132	11	by	by	ADP
cana-5106	132	12	ai	ai	VERB
cana-5106	132	13	"	"	PUNCT
cana-5106	132	14	)	)	PUNCT
cana-5106	132	15	and	and	CCONJ
cana-5106	132	16	briefly	briefly	ADV
cana-5106	132	17	explains	explain	VERB
cana-5106	132	18	the	the	DET
cana-5106	132	19	functionality	functionality	NOUN
cana-5106	132	20	:	:	PUNCT
cana-5106	132	21	generating	generate	VERB
cana-5106	132	22	descriptive	descriptive	ADJ
cana-5106	132	23	captions	caption	NOUN
cana-5106	132	24	instantly	instantly	ADV
cana-5106	132	25	using	use	VERB
cana-5106	132	26	advanced	advanced	ADJ
cana-5106	132	27	deep	deep	ADJ
cana-5106	132	28	learning	learning	NOUN
cana-5106	132	29	technology	technology	NOUN
cana-5106	132	30	.	.	PUNCT
cana-5106	133	1	a	a	DET
cana-5106	133	2	clear	clear	ADJ
cana-5106	133	3	call	call	NOUN
cana-5106	133	4	to	to	ADP
cana-5106	133	5	action	action	NOUN
cana-5106	133	6	(	(	PUNCT
cana-5106	133	7	"	"	PUNCT
cana-5106	133	8	get	get	AUX
cana-5106	133	9	started	start	VERB
cana-5106	133	10	"	"	PUNCT
cana-5106	133	11	)	)	PUNCT
cana-5106	133	12	encourages	encourage	VERB
cana-5106	133	13	user	user	NOUN
cana-5106	133	14	engagement	engagement	NOUN
cana-5106	133	15	,	,	PUNCT
cana-5106	133	16	while	while	SCONJ
cana-5106	133	17	icons	icon	NOUN
cana-5106	133	18	at	at	ADP
cana-5106	133	19	the	the	DET
cana-5106	133	20	bottom	bottom	ADJ
cana-5106	133	21	showcase	showcase	NOUN
cana-5106	133	22	key	key	NOUN
cana-5106	133	23	features	feature	NOUN
cana-5106	133	24	:	:	PUNCT
cana-5106	133	25	deep	deep	ADJ
cana-5106	133	26	learning	learning	NOUN
cana-5106	133	27	,	,	PUNCT
cana-5106	133	28	instant	instant	ADJ
cana-5106	133	29	results	result	NOUN
cana-5106	133	30	,	,	PUNCT
cana-5106	133	31	and	and	CCONJ
cana-5106	133	32	privacy	privacy	NOUN
cana-5106	133	33	focus	focus	NOUN
cana-5106	133	34	.	.	PUNCT
cana-5106	134	1	overall	overall	ADV
cana-5106	134	2	,	,	PUNCT
cana-5106	134	3	the	the	DET
cana-5106	134	4	banner	banner	NOUN
cana-5106	134	5	effectively	effectively	ADV
cana-5106	134	6	communicates	communicate	VERB
cana-5106	134	7	the	the	DET
cana-5106	134	8	core	core	NOUN
cana-5106	134	9	value	value	NOUN
cana-5106	134	10	proposition	proposition	NOUN
cana-5106	134	11	of	of	ADP
cana-5106	134	12	smart	smart	ADJ
cana-5106	134	13	caption	caption	NOUN
cana-5106	134	14	as	as	ADP
cana-5106	134	15	a	a	DET
cana-5106	134	16	fast	fast	ADJ
cana-5106	134	17	,	,	PUNCT
cana-5106	134	18	intelligent	intelligent	ADJ
cana-5106	134	19	,	,	PUNCT
cana-5106	134	20	and	and	CCONJ
cana-5106	134	21	privacyconscious	privacyconscious	ADJ
cana-5106	134	22	solution	solution	NOUN
cana-5106	134	23	for	for	ADP
cana-5106	134	24	automated	automate	VERB
cana-5106	134	25	image	image	NOUN
cana-5106	134	26	captioning	captioning	NOUN
cana-5106	134	27	.	.	PUNCT
cana-5106	135	1	communications	communication	NOUN
cana-5106	135	2	on	on	ADP
cana-5106	135	3	applied	apply	VERB
cana-5106	135	4	nonlinear	nonlinear	ADJ
cana-5106	135	5	analysis	analysis	NOUN
cana-5106	135	6	issn	issn	NOUN
cana-5106	135	7	:	:	PUNCT
cana-5106	135	8	1074	1074	NUM
cana-5106	135	9	-	-	PUNCT
cana-5106	135	10	133x	133x	NUM
cana-5106	135	11	vol	vol	VERB
cana-5106	135	12	32	32	NUM
cana-5106	135	13	no	no	NOUN
cana-5106	135	14	.	.	PUNCT
cana-5106	136	1	10s	10	NOUN
cana-5106	136	2	(	(	PUNCT
cana-5106	136	3	2025	2025	NUM
cana-5106	136	4	)	)	PUNCT
cana-5106	136	5	1128	1128	NUM
cana-5106	136	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	136	7	fig	fig	NOUN
cana-5106	136	8	:	:	PUNCT
cana-5106	136	9	login	login	NOUN
cana-5106	136	10	page	page	NOUN
cana-5106	136	11	fig	fig	NOUN
cana-5106	136	12	:	:	PUNCT
cana-5106	136	13	upload	upload	VERB
cana-5106	136	14	image	image	NOUN
cana-5106	136	15	this	this	DET
cana-5106	136	16	image	image	NOUN
cana-5106	136	17	shows	show	VERB
cana-5106	136	18	the	the	DET
cana-5106	136	19	user	user	NOUN
cana-5106	136	20	interface	interface	NOUN
cana-5106	136	21	for	for	ADP
cana-5106	136	22	uploading	upload	VERB
cana-5106	136	23	an	an	DET
cana-5106	136	24	image	image	NOUN
cana-5106	136	25	to	to	ADP
cana-5106	136	26	"	"	PUNCT
cana-5106	136	27	smart	smart	ADJ
cana-5106	136	28	caption	caption	NOUN
cana-5106	136	29	,	,	PUNCT
cana-5106	136	30	"	"	PUNCT
cana-5106	136	31	an	an	DET
cana-5106	136	32	aipowered	aipowere	VERB
cana-5106	136	33	tool	tool	NOUN
cana-5106	136	34	for	for	ADP
cana-5106	136	35	generating	generate	VERB
cana-5106	136	36	image	image	NOUN
cana-5106	136	37	descriptions	description	NOUN
cana-5106	136	38	.	.	PUNCT
cana-5106	137	1	the	the	PRON
cana-5106	137	2	prominent	prominent	ADJ
cana-5106	137	3	"	"	PUNCT
cana-5106	137	4	upload	upload	VERB
cana-5106	137	5	your	your	PRON
cana-5106	137	6	image	image	NOUN
cana-5106	137	7	"	"	PUNCT
cana-5106	137	8	text	text	NOUN
cana-5106	137	9	and	and	CCONJ
cana-5106	137	10	a	a	DET
cana-5106	137	11	symbolic	symbolic	ADJ
cana-5106	137	12	upload	upload	NOUN
cana-5106	137	13	icon	icon	NOUN
cana-5106	137	14	guide	guide	VERB
cana-5106	137	15	the	the	DET
cana-5106	137	16	user	user	NOUN
cana-5106	137	17	to	to	ADP
cana-5106	137	18	the	the	DET
cana-5106	137	19	central	central	ADJ
cana-5106	137	20	function	function	NOUN
cana-5106	137	21	.	.	PUNCT
cana-5106	138	1	below	below	ADV
cana-5106	138	2	,	,	PUNCT
cana-5106	138	3	a	a	DET
cana-5106	138	4	"	"	PUNCT
cana-5106	138	5	choose	choose	NOUN
cana-5106	138	6	file	file	NOUN
cana-5106	138	7	"	"	PUNCT
cana-5106	138	8	button	button	NOUN
cana-5106	138	9	allows	allow	VERB
cana-5106	138	10	users	user	NOUN
cana-5106	138	11	to	to	PART
cana-5106	138	12	select	select	VERB
cana-5106	138	13	an	an	DET
cana-5106	138	14	image	image	NOUN
cana-5106	138	15	from	from	ADP
cana-5106	138	16	their	their	PRON
cana-5106	138	17	device	device	NOUN
cana-5106	138	18	,	,	PUNCT
cana-5106	138	19	with	with	ADP
cana-5106	138	20	a	a	DET
cana-5106	138	21	note	note	NOUN
cana-5106	138	22	specifying	specify	VERB
cana-5106	138	23	supported	support	VERB
cana-5106	138	24	file	file	NOUN
cana-5106	138	25	formats	format	NOUN
cana-5106	138	26	(	(	PUNCT
cana-5106	138	27	jpg	jpg	NOUN
cana-5106	138	28	,	,	PUNCT
cana-5106	138	29	png	png	PROPN
cana-5106	138	30	,	,	PUNCT
cana-5106	138	31	jpeg	jpeg	NOUN
cana-5106	138	32	)	)	PUNCT
cana-5106	138	33	.	.	PUNCT
cana-5106	139	1	a	a	DET
cana-5106	139	2	secondary	secondary	ADJ
cana-5106	139	3	call	call	NOUN
cana-5106	139	4	to	to	ADP
cana-5106	139	5	action	action	NOUN
cana-5106	139	6	,	,	PUNCT
cana-5106	139	7	"	"	PUNCT
cana-5106	139	8	generate	generate	VERB
cana-5106	139	9	caption	caption	NOUN
cana-5106	139	10	,	,	PUNCT
cana-5106	139	11	"	"	PUNCT
cana-5106	139	12	suggests	suggest	VERB
cana-5106	139	13	the	the	DET
cana-5106	139	14	next	next	ADJ
cana-5106	139	15	step	step	NOUN
cana-5106	139	16	in	in	ADP
cana-5106	139	17	the	the	DET
cana-5106	139	18	process	process	NOUN
cana-5106	139	19	.	.	PUNCT
cana-5106	140	1	the	the	DET
cana-5106	140	2	clean	clean	ADJ
cana-5106	140	3	,	,	PUNCT
cana-5106	140	4	white	white	ADJ
cana-5106	140	5	background	background	NOUN
cana-5106	140	6	and	and	CCONJ
cana-5106	140	7	blue	blue	ADJ
cana-5106	140	8	accents	accent	NOUN
cana-5106	140	9	maintain	maintain	VERB
cana-5106	140	10	the	the	DET
cana-5106	140	11	minimalist	minimalist	ADJ
cana-5106	140	12	design	design	NOUN
cana-5106	140	13	seen	see	VERB
cana-5106	140	14	in	in	ADP
cana-5106	140	15	other	other	ADJ
cana-5106	140	16	smart	smart	ADJ
cana-5106	140	17	caption	caption	NOUN
cana-5106	140	18	promotional	promotional	ADJ
cana-5106	140	19	materials	material	NOUN
cana-5106	140	20	.	.	PUNCT
cana-5106	141	1	overall	overall	ADV
cana-5106	141	2	,	,	PUNCT
cana-5106	141	3	the	the	DET
cana-5106	141	4	image	image	NOUN
cana-5106	141	5	showcases	showcase	VERB
cana-5106	141	6	a	a	DET
cana-5106	141	7	simple	simple	ADJ
cana-5106	141	8	and	and	CCONJ
cana-5106	141	9	user	user	NOUN
cana-5106	141	10	-	-	PUNCT
cana-5106	141	11	friendly	friendly	ADJ
cana-5106	141	12	upload	upload	NOUN
cana-5106	141	13	interface	interface	NOUN
cana-5106	141	14	designed	design	VERB
cana-5106	141	15	to	to	PART
cana-5106	141	16	initiate	initiate	VERB
cana-5106	141	17	the	the	DET
cana-5106	141	18	intelligent	intelligent	ADJ
cana-5106	141	19	image	image	NOUN
cana-5106	141	20	description	description	NOUN
cana-5106	141	21	process	process	NOUN
cana-5106	141	22	offered	offer	VERB
cana-5106	141	23	by	by	ADP
cana-5106	141	24	smart	smart	ADJ
cana-5106	141	25	caption	caption	NOUN
cana-5106	141	26	.	.	PUNCT
cana-5106	142	1	fig	fig	NOUN
cana-5106	142	2	:	:	PUNCT
cana-5106	142	3	caption	caption	NOUN
cana-5106	142	4	generated	generate	VERB
cana-5106	142	5	communications	communication	NOUN
cana-5106	142	6	on	on	ADP
cana-5106	142	7	applied	apply	VERB
cana-5106	142	8	nonlinear	nonlinear	ADJ
cana-5106	142	9	analysis	analysis	NOUN
cana-5106	142	10	issn	issn	NOUN
cana-5106	142	11	:	:	PUNCT
cana-5106	142	12	1074	1074	NUM
cana-5106	142	13	-	-	PUNCT
cana-5106	142	14	133x	133x	NUM
cana-5106	142	15	vol	vol	VERB
cana-5106	142	16	32	32	NUM
cana-5106	142	17	no	no	NOUN
cana-5106	142	18	.	.	PUNCT
cana-5106	143	1	10s	10	NOUN
cana-5106	143	2	(	(	PUNCT
cana-5106	143	3	2025	2025	NUM
cana-5106	143	4	)	)	PUNCT
cana-5106	143	5	1129	1129	NUM
cana-5106	143	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	144	1	this	this	DET
cana-5106	144	2	image	image	NOUN
cana-5106	144	3	shows	show	VERB
cana-5106	144	4	the	the	DET
cana-5106	144	5	smart	smart	ADJ
cana-5106	144	6	caption	caption	NOUN
cana-5106	144	7	interface	interface	NOUN
cana-5106	144	8	after	after	SCONJ
cana-5106	144	9	an	an	DET
cana-5106	144	10	image	image	NOUN
cana-5106	144	11	has	have	AUX
cana-5106	144	12	been	be	AUX
cana-5106	144	13	uploaded	upload	VERB
cana-5106	144	14	and	and	CCONJ
cana-5106	144	15	processed	process	VERB
cana-5106	144	16	.	.	PUNCT
cana-5106	145	1	the	the	DET
cana-5106	145	2	user	user	NOUN
cana-5106	145	3	is	be	AUX
cana-5106	145	4	presented	present	VERB
cana-5106	145	5	with	with	ADP
cana-5106	145	6	the	the	DET
cana-5106	145	7	uploaded	uploaded	ADJ
cana-5106	145	8	image	image	NOUN
cana-5106	145	9	,	,	PUNCT
cana-5106	145	10	in	in	ADP
cana-5106	145	11	this	this	DET
cana-5106	145	12	case	case	NOUN
cana-5106	145	13	,	,	PUNCT
cana-5106	145	14	two	two	NUM
cana-5106	145	15	dogs	dog	NOUN
cana-5106	145	16	running	run	VERB
cana-5106	145	17	on	on	ADP
cana-5106	145	18	a	a	DET
cana-5106	145	19	beach	beach	NOUN
cana-5106	145	20	,	,	PUNCT
cana-5106	145	21	and	and	CCONJ
cana-5106	145	22	below	below	ADP
cana-5106	145	23	it	it	PRON
cana-5106	145	24	,	,	PUNCT
cana-5106	145	25	the	the	DET
cana-5106	145	26	ai	ai	NOUN
cana-5106	145	27	-	-	PUNCT
cana-5106	145	28	generated	generate	VERB
cana-5106	145	29	caption	caption	NOUN
cana-5106	145	30	:	:	PUNCT
cana-5106	145	31	"	"	PUNCT
cana-5106	145	32	two	two	NUM
cana-5106	145	33	dogs	dog	NOUN
cana-5106	145	34	running	run	VERB
cana-5106	145	35	around	around	ADV
cana-5106	145	36	.	.	PUNCT
cana-5106	145	37	"	"	PUNCT
cana-5106	146	1	the	the	DET
cana-5106	146	2	interface	interface	NOUN
cana-5106	146	3	is	be	AUX
cana-5106	146	4	clean	clean	ADJ
cana-5106	146	5	and	and	CCONJ
cana-5106	146	6	simple	simple	ADJ
cana-5106	146	7	,	,	PUNCT
cana-5106	146	8	with	with	ADP
cana-5106	146	9	a	a	DET
cana-5106	146	10	"	"	PUNCT
cana-5106	146	11	back	back	NOUN
cana-5106	146	12	to	to	PART
cana-5106	146	13	upload	upload	VERB
cana-5106	146	14	"	"	PUNCT
cana-5106	146	15	button	button	NOUN
cana-5106	146	16	to	to	PART
cana-5106	146	17	facilitate	facilitate	VERB
cana-5106	146	18	further	further	ADJ
cana-5106	146	19	use	use	NOUN
cana-5106	146	20	.	.	PUNCT
cana-5106	147	1	this	this	DET
cana-5106	147	2	image	image	NOUN
cana-5106	147	3	exemplifies	exemplify	VERB
cana-5106	147	4	the	the	DET
cana-5106	147	5	core	core	NOUN
cana-5106	147	6	functionality	functionality	NOUN
cana-5106	147	7	of	of	ADP
cana-5106	147	8	smart	smart	ADJ
cana-5106	147	9	caption	caption	NOUN
cana-5106	147	10	,	,	PUNCT
cana-5106	147	11	demonstrating	demonstrate	VERB
cana-5106	147	12	its	its	PRON
cana-5106	147	13	ability	ability	NOUN
cana-5106	147	14	to	to	PART
cana-5106	147	15	analyze	analyze	VERB
cana-5106	147	16	an	an	DET
cana-5106	147	17	image	image	NOUN
cana-5106	147	18	and	and	CCONJ
cana-5106	147	19	provide	provide	VERB
cana-5106	147	20	a	a	DET
cana-5106	147	21	concise	concise	ADJ
cana-5106	147	22	,	,	PUNCT
cana-5106	147	23	descriptive	descriptive	ADJ
cana-5106	147	24	caption	caption	NOUN
cana-5106	147	25	.	.	PUNCT
cana-5106	148	1	the	the	DET
cana-5106	148	2	straightforward	straightforward	ADJ
cana-5106	148	3	design	design	NOUN
cana-5106	148	4	and	and	CCONJ
cana-5106	148	5	clear	clear	ADJ
cana-5106	148	6	presentation	presentation	NOUN
cana-5106	148	7	of	of	ADP
cana-5106	148	8	the	the	DET
cana-5106	148	9	results	result	NOUN
cana-5106	148	10	highlight	highlight	VERB
cana-5106	148	11	the	the	DET
cana-5106	148	12	tool	tool	NOUN
cana-5106	148	13	's	's	PART
cana-5106	148	14	ease	ease	NOUN
cana-5106	148	15	of	of	ADP
cana-5106	148	16	use	use	NOUN
cana-5106	148	17	and	and	CCONJ
cana-5106	148	18	efficiency	efficiency	NOUN
cana-5106	148	19	in	in	ADP
cana-5106	148	20	generating	generate	VERB
cana-5106	148	21	image	image	NOUN
cana-5106	148	22	descriptions	description	NOUN
cana-5106	148	23	.	.	PUNCT
cana-5106	149	1	vi	vi	X
cana-5106	149	2	.	.	NOUN
cana-5106	149	3	conclusion	conclusion	NOUN
cana-5106	149	4	in	in	ADP
cana-5106	149	5	conclusion	conclusion	NOUN
cana-5106	149	6	,	,	PUNCT
cana-5106	149	7	this	this	DET
cana-5106	149	8	research	research	NOUN
cana-5106	149	9	underscores	underscore	VERB
cana-5106	149	10	the	the	DET
cana-5106	149	11	transformative	transformative	ADJ
cana-5106	149	12	impact	impact	NOUN
cana-5106	149	13	of	of	ADP
cana-5106	149	14	integrating	integrate	VERB
cana-5106	149	15	vision	vision	NOUN
cana-5106	149	16	transformers	transformer	NOUN
cana-5106	149	17	(	(	PUNCT
cana-5106	149	18	vits	vit	NOUN
cana-5106	149	19	)	)	PUNCT
cana-5106	149	20	and	and	CCONJ
cana-5106	149	21	nlp	nlp	NOUN
cana-5106	149	22	transformers	transformer	NOUN
cana-5106	149	23	for	for	ADP
cana-5106	149	24	automated	automate	VERB
cana-5106	149	25	image	image	NOUN
cana-5106	149	26	captioning	captioning	NOUN
cana-5106	149	27	.	.	PUNCT
cana-5106	150	1	by	by	ADP
cana-5106	150	2	harnessing	harness	VERB
cana-5106	150	3	the	the	DET
cana-5106	150	4	complementary	complementary	ADJ
cana-5106	150	5	strengths	strength	NOUN
cana-5106	150	6	of	of	ADP
cana-5106	150	7	these	these	DET
cana-5106	150	8	advanced	advanced	ADJ
cana-5106	150	9	architectures	architecture	NOUN
cana-5106	150	10	,	,	PUNCT
cana-5106	150	11	the	the	DET
cana-5106	150	12	proposed	propose	VERB
cana-5106	150	13	system	system	NOUN
cana-5106	150	14	achieves	achieve	VERB
cana-5106	150	15	significant	significant	ADJ
cana-5106	150	16	advancements	advancement	NOUN
cana-5106	150	17	in	in	ADP
cana-5106	150	18	generating	generate	VERB
cana-5106	150	19	accurate	accurate	ADJ
cana-5106	150	20	,	,	PUNCT
cana-5106	150	21	context	context	NOUN
cana-5106	150	22	-	-	PUNCT
cana-5106	150	23	aware	aware	ADJ
cana-5106	150	24	,	,	PUNCT
cana-5106	150	25	and	and	CCONJ
cana-5106	150	26	fluent	fluent	ADJ
cana-5106	150	27	captions	caption	NOUN
cana-5106	150	28	,	,	PUNCT
cana-5106	150	29	greatly	greatly	ADV
cana-5106	150	30	enhancing	enhance	VERB
cana-5106	150	31	the	the	DET
cana-5106	150	32	interpretation	interpretation	NOUN
cana-5106	150	33	of	of	ADP
cana-5106	150	34	visual	visual	ADJ
cana-5106	150	35	content	content	NOUN
cana-5106	150	36	.	.	PUNCT
cana-5106	151	1	the	the	DET
cana-5106	151	2	results	result	NOUN
cana-5106	151	3	demonstrate	demonstrate	VERB
cana-5106	151	4	that	that	SCONJ
cana-5106	151	5	this	this	DET
cana-5106	151	6	integration	integration	NOUN
cana-5106	151	7	not	not	PART
cana-5106	151	8	only	only	ADV
cana-5106	151	9	elevates	elevate	VERB
cana-5106	151	10	the	the	DET
cana-5106	151	11	quality	quality	NOUN
cana-5106	151	12	of	of	ADP
cana-5106	151	13	image	image	NOUN
cana-5106	151	14	descriptions	description	NOUN
cana-5106	151	15	but	but	CCONJ
cana-5106	151	16	also	also	ADV
cana-5106	151	17	overcomes	overcome	VERB
cana-5106	151	18	key	key	ADJ
cana-5106	151	19	limitations	limitation	NOUN
cana-5106	151	20	of	of	ADP
cana-5106	151	21	traditional	traditional	ADJ
cana-5106	151	22	captioning	caption	VERB
cana-5106	151	23	methods	method	NOUN
cana-5106	151	24	.	.	PUNCT
cana-5106	152	1	looking	look	VERB
cana-5106	152	2	ahead	ahead	ADV
cana-5106	152	3	,	,	PUNCT
cana-5106	152	4	the	the	DET
cana-5106	152	5	system	system	NOUN
cana-5106	152	6	holds	hold	VERB
cana-5106	152	7	immense	immense	ADJ
cana-5106	152	8	potential	potential	NOUN
cana-5106	152	9	for	for	ADP
cana-5106	152	10	diverse	diverse	ADJ
cana-5106	152	11	applications	application	NOUN
cana-5106	152	12	,	,	PUNCT
cana-5106	152	13	including	include	VERB
cana-5106	152	14	accessibility	accessibility	NOUN
cana-5106	152	15	tools	tool	NOUN
cana-5106	152	16	,	,	PUNCT
cana-5106	152	17	e	e	NOUN
cana-5106	152	18	-	-	NOUN
cana-5106	152	19	commerce	commerce	NOUN
cana-5106	152	20	,	,	PUNCT
cana-5106	152	21	and	and	CCONJ
cana-5106	152	22	automated	automate	VERB
cana-5106	152	23	content	content	NOUN
cana-5106	152	24	generation	generation	NOUN
cana-5106	152	25	.	.	PUNCT
cana-5106	153	1	this	this	DET
cana-5106	153	2	work	work	NOUN
cana-5106	153	3	highlights	highlight	VERB
cana-5106	153	4	the	the	DET
cana-5106	153	5	critical	critical	ADJ
cana-5106	153	6	need	need	NOUN
cana-5106	153	7	for	for	ADP
cana-5106	153	8	further	further	ADJ
cana-5106	153	9	exploration	exploration	NOUN
cana-5106	153	10	in	in	ADP
cana-5106	153	11	multimodal	multimodal	ADJ
cana-5106	153	12	learning	learning	NOUN
cana-5106	153	13	,	,	PUNCT
cana-5106	153	14	with	with	ADP
cana-5106	153	15	the	the	DET
cana-5106	153	16	goal	goal	NOUN
cana-5106	153	17	of	of	ADP
cana-5106	153	18	developing	develop	VERB
cana-5106	153	19	systems	system	NOUN
cana-5106	153	20	that	that	PRON
cana-5106	153	21	can	can	AUX
cana-5106	153	22	better	well	ADV
cana-5106	153	23	understand	understand	VERB
cana-5106	153	24	and	and	CCONJ
cana-5106	153	25	articulate	articulate	VERB
cana-5106	153	26	the	the	DET
cana-5106	153	27	complex	complex	ADJ
cana-5106	153	28	relationships	relationship	NOUN
cana-5106	153	29	between	between	ADP
cana-5106	153	30	visual	visual	ADJ
cana-5106	153	31	and	and	CCONJ
cana-5106	153	32	textual	textual	ADJ
cana-5106	153	33	information	information	NOUN
cana-5106	153	34	.	.	PUNCT
cana-5106	154	1	such	such	ADJ
cana-5106	154	2	advancements	advancement	NOUN
cana-5106	154	3	promise	promise	VERB
cana-5106	154	4	to	to	PART
cana-5106	154	5	contribute	contribute	VERB
cana-5106	154	6	to	to	ADP
cana-5106	154	7	a	a	DET
cana-5106	154	8	more	more	ADV
cana-5106	154	9	inclusive	inclusive	ADJ
cana-5106	154	10	,	,	PUNCT
cana-5106	154	11	intelligent	intelligent	ADJ
cana-5106	154	12	,	,	PUNCT
cana-5106	154	13	and	and	CCONJ
cana-5106	154	14	interconnected	interconnect	VERB
cana-5106	154	15	digital	digital	ADJ
cana-5106	154	16	future	future	NOUN
cana-5106	154	17	.	.	PUNCT
cana-5106	155	1	references	reference	NOUN
cana-5106	155	2	[	[	X
cana-5106	155	3	1	1	NUM
cana-5106	155	4	]	]	PUNCT
cana-5106	155	5	r.	r.	NOUN
cana-5106	155	6	mulyawan	mulyawan	PROPN
cana-5106	155	7	,	,	PUNCT
cana-5106	155	8	a.	a.	NOUN
cana-5106	155	9	sunyoto	sunyoto	PROPN
cana-5106	155	10	,	,	PUNCT
cana-5106	155	11	and	and	CCONJ
cana-5106	155	12	a.	a.	PROPN
cana-5106	155	13	h.	h.	PROPN
cana-5106	155	14	muhammad	muhammad	PROPN
cana-5106	155	15	proposed	propose	VERB
cana-5106	155	16	an	an	DET
cana-5106	155	17	approach	approach	NOUN
cana-5106	155	18	for	for	ADP
cana-5106	155	19	automatic	automatic	ADJ
cana-5106	155	20	indonesian	indonesian	ADJ
cana-5106	155	21	image	image	NOUN
cana-5106	155	22	captioning	captioning	NOUN
cana-5106	155	23	using	use	VERB
cana-5106	155	24	a	a	DET
cana-5106	155	25	combination	combination	NOUN
cana-5106	155	26	of	of	ADP
cana-5106	155	27	cnn	cnn	PROPN
cana-5106	155	28	and	and	CCONJ
cana-5106	155	29	transformer	transformer	NOUN
cana-5106	155	30	models	model	NOUN
cana-5106	155	31	.	.	PUNCT
cana-5106	156	1	their	their	PRON
cana-5106	156	2	work	work	NOUN
cana-5106	156	3	was	be	AUX
cana-5106	156	4	presented	present	VERB
cana-5106	156	5	at	at	ADP
cana-5106	156	6	the	the	DET
cana-5106	156	7	*	*	NOUN
cana-5106	156	8	2022	2022	NUM
cana-5106	156	9	5th	5th	ADJ
cana-5106	156	10	international	international	ADJ
cana-5106	156	11	conference	conference	NOUN
cana-5106	156	12	on	on	ADP
cana-5106	156	13	information	information	NOUN
cana-5106	156	14	and	and	CCONJ
cana-5106	156	15	communications	communication	NOUN
cana-5106	156	16	technology	technology	NOUN
cana-5106	156	17	(	(	PUNCT
cana-5106	156	18	icoiact	icoiact	NOUN
cana-5106	156	19	)	)	PUNCT
cana-5106	156	20	*	*	PUNCT
cana-5106	156	21	,	,	PUNCT
cana-5106	156	22	with	with	ADP
cana-5106	156	23	the	the	DET
cana-5106	156	24	findings	finding	NOUN
cana-5106	156	25	published	publish	VERB
cana-5106	156	26	on	on	ADP
cana-5106	156	27	pages	page	NOUN
cana-5106	156	28	355	355	NUM
cana-5106	156	29	–	–	PUNCT
cana-5106	156	30	360	360	NUM
cana-5106	156	31	.	.	PUNCT
cana-5106	157	1	the	the	DET
cana-5106	157	2	doi	doi	NOUN
cana-5106	157	3	for	for	ADP
cana-5106	157	4	the	the	DET
cana-5106	157	5	paper	paper	NOUN
cana-5106	157	6	is	be	AUX
cana-5106	157	7	10.1109	10.1109	NUM
cana-5106	157	8	/	/	SYM
cana-5106	157	9	icoiact55506.2022.9971855	icoiact55506.2022.9971855	PROPN
cana-5106	157	10	.	.	PUNCT
cana-5106	158	1	[	[	X
cana-5106	158	2	2	2	NUM
cana-5106	158	3	]	]	X
cana-5106	158	4	patil	patil	PROPN
cana-5106	158	5	,	,	PUNCT
cana-5106	158	6	j.	j.	PROPN
cana-5106	158	7	y.	y.	PROPN
cana-5106	158	8	,	,	PUNCT
cana-5106	158	9	&	&	CCONJ
cana-5106	158	10	wankhade	wankhade	PROPN
cana-5106	158	11	,	,	PUNCT
cana-5106	158	12	n.	n.	PROPN
cana-5106	158	13	r.	r.	PROPN
cana-5106	158	14	(	(	PUNCT
cana-5106	158	15	2017	2017	NUM
cana-5106	158	16	)	)	PUNCT
cana-5106	158	17	.	.	PUNCT
cana-5106	159	1	web	web	NOUN
cana-5106	159	2	image	image	NOUN
cana-5106	159	3	re	re	VERB
cana-5106	159	4	-	-	VERB
cana-5106	159	5	ranking	ranking	ADJ
cana-5106	159	6	using	use	VERB
cana-5106	159	7	semantic	semantic	ADJ
cana-5106	159	8	signature	signature	NOUN
cana-5106	159	9	.	.	PUNCT
cana-5106	160	1	international	international	ADJ
cana-5106	160	2	journal	journal	PROPN
cana-5106	160	3	,	,	PUNCT
cana-5106	160	4	5(6	5(6	NUM
cana-5106	160	5	)	)	PUNCT
cana-5106	160	6	.	.	PUNCT
cana-5106	161	1	[	[	X
cana-5106	161	2	3	3	X
cana-5106	161	3	]	]	PUNCT
cana-5106	161	4	x.	x.	NOUN
cana-5106	161	5	dai	dai	PROPN
cana-5106	161	6	,	,	PUNCT
cana-5106	161	7	y.	y.	PROPN
cana-5106	161	8	chen	chen	PROPN
cana-5106	161	9	,	,	PUNCT
cana-5106	161	10	b.	b.	PROPN
cana-5106	161	11	xiao	xiao	PROPN
cana-5106	161	12	,	,	PUNCT
cana-5106	161	13	d.	d.	PROPN
cana-5106	161	14	chen	chen	PROPN
cana-5106	161	15	,	,	PUNCT
cana-5106	161	16	m.	m.	PROPN
cana-5106	161	17	liu	liu	PROPN
cana-5106	161	18	,	,	PUNCT
cana-5106	161	19	l.	l.	PROPN
cana-5106	161	20	yuan	yuan	PROPN
cana-5106	161	21	,	,	PUNCT
cana-5106	161	22	and	and	CCONJ
cana-5106	161	23	l.	l.	PROPN
cana-5106	161	24	zhang	zhang	PROPN
cana-5106	161	25	introduced	introduce	VERB
cana-5106	161	26	a	a	DET
cana-5106	161	27	method	method	NOUN
cana-5106	161	28	called	call	VERB
cana-5106	161	29	"	"	PUNCT
cana-5106	161	30	dynamic	dynamic	ADJ
cana-5106	161	31	head	head	NOUN
cana-5106	161	32	,	,	PUNCT
cana-5106	161	33	"	"	PUNCT
cana-5106	161	34	which	which	PRON
cana-5106	161	35	unifies	unify	VERB
cana-5106	161	36	object	object	NOUN
cana-5106	161	37	detection	detection	NOUN
cana-5106	161	38	heads	head	NOUN
cana-5106	161	39	using	use	VERB
cana-5106	161	40	attention	attention	NOUN
cana-5106	161	41	mechanisms	mechanism	NOUN
cana-5106	161	42	.	.	PUNCT
cana-5106	162	1	this	this	DET
cana-5106	162	2	research	research	NOUN
cana-5106	162	3	was	be	AUX
cana-5106	162	4	published	publish	VERB
cana-5106	162	5	in	in	ADP
cana-5106	162	6	the	the	DET
cana-5106	162	7	*	*	ADJ
cana-5106	162	8	proceedings	proceeding	NOUN
cana-5106	162	9	of	of	ADP
cana-5106	162	10	the	the	DET
cana-5106	162	11	ieee	ieee	NOUN
cana-5106	162	12	/	/	SYM
cana-5106	162	13	cvf	cvf	NOUN
cana-5106	162	14	conference	conference	NOUN
cana-5106	162	15	on	on	ADP
cana-5106	162	16	computer	computer	NOUN
cana-5106	162	17	vision	vision	NOUN
cana-5106	162	18	and	and	CCONJ
cana-5106	162	19	pattern	pattern	NOUN
cana-5106	162	20	recognition	recognition	NOUN
cana-5106	162	21	(	(	PUNCT
cana-5106	162	22	cvpr	cvpr	NOUN
cana-5106	162	23	)	)	PUNCT
cana-5106	162	24	*	*	PUNCT
cana-5106	162	25	in	in	ADP
cana-5106	162	26	2021	2021	NUM
cana-5106	162	27	,	,	PUNCT
cana-5106	162	28	spanning	span	VERB
cana-5106	162	29	pages	page	NOUN
cana-5106	162	30	7373–7382	7373–7382	NUM
cana-5106	162	31	.	.	PUNCT
cana-5106	163	1	[	[	X
cana-5106	163	2	4	4	NUM
cana-5106	163	3	]	]	X
cana-5106	163	4	y.	y.	PROPN
cana-5106	163	5	fang	fang	PROPN
cana-5106	163	6	,	,	PUNCT
cana-5106	163	7	b.	b.	PROPN
cana-5106	163	8	liao	liao	PROPN
cana-5106	163	9	,	,	PUNCT
cana-5106	163	10	x.	x.	PROPN
cana-5106	163	11	wang	wang	PROPN
cana-5106	163	12	,	,	PUNCT
cana-5106	163	13	j.	j.	PROPN
cana-5106	163	14	fang	fang	PROPN
cana-5106	163	15	,	,	PUNCT
cana-5106	163	16	j.	j.	PROPN
cana-5106	163	17	qi	qi	PROPN
cana-5106	163	18	,	,	PUNCT
cana-5106	163	19	r.	r.	PROPN
cana-5106	163	20	wu	wu	PROPN
cana-5106	163	21	,	,	PUNCT
cana-5106	163	22	j.	j.	PROPN
cana-5106	163	23	niu	niu	PROPN
cana-5106	163	24	,	,	PUNCT
cana-5106	163	25	and	and	CCONJ
cana-5106	163	26	w.	w.	PROPN
cana-5106	163	27	liu	liu	PROPN
cana-5106	163	28	,	,	PUNCT
cana-5106	163	29	“	"	PUNCT
cana-5106	163	30	you	you	PRON
cana-5106	163	31	only	only	ADV
cana-5106	163	32	look	look	VERB
cana-5106	163	33	at	at	ADP
cana-5106	163	34	one	one	NUM
cana-5106	163	35	sequence	sequence	NOUN
cana-5106	163	36	:	:	PUNCT
cana-5106	163	37	rethinking	rethink	VERB
cana-5106	163	38	transformer	transformer	NOUN
cana-5106	163	39	in	in	ADP
cana-5106	163	40	vision	vision	NOUN
cana-5106	163	41	through	through	ADP
cana-5106	163	42	object	object	NOUN
cana-5106	163	43	detection	detection	NOUN
cana-5106	163	44	,	,	PUNCT
cana-5106	163	45	”	"	PUNCT
cana-5106	163	46	arxiv	arxiv	PROPN
cana-5106	163	47	preprint	preprint	NOUN
cana-5106	163	48	arxiv:2106.00666	arxiv:2106.00666	NUM
cana-5106	163	49	,	,	PUNCT
cana-5106	163	50	2021	2021	NUM
cana-5106	163	51	.	.	PUNCT
cana-5106	164	1	communications	communication	NOUN
cana-5106	164	2	on	on	ADP
cana-5106	164	3	applied	apply	VERB
cana-5106	164	4	nonlinear	nonlinear	ADJ
cana-5106	164	5	analysis	analysis	NOUN
cana-5106	164	6	issn	issn	NOUN
cana-5106	164	7	:	:	PUNCT
cana-5106	164	8	1074	1074	NUM
cana-5106	164	9	-	-	PUNCT
cana-5106	164	10	133x	133x	NUM
cana-5106	164	11	vol	vol	VERB
cana-5106	164	12	32	32	NUM
cana-5106	164	13	no	no	NOUN
cana-5106	164	14	.	.	PUNCT
cana-5106	165	1	10s	10	NOUN
cana-5106	165	2	(	(	PUNCT
cana-5106	165	3	2025	2025	NUM
cana-5106	165	4	)	)	PUNCT
cana-5106	165	5	1130	1130	NUM
cana-5106	165	6	https://internationalpubls.com	https://internationalpubls.com	X
cana-5106	166	1	[	[	X
cana-5106	166	2	5	5	NUM
cana-5106	166	3	]	]	PUNCT
cana-5106	166	4	c.	c.	PROPN
cana-5106	166	5	feng	feng	PROPN
cana-5106	166	6	,	,	PUNCT
cana-5106	166	7	y.	y.	PROPN
cana-5106	166	8	zhong	zhong	PROPN
cana-5106	166	9	,	,	PUNCT
cana-5106	166	10	y.	y.	PROPN
cana-5106	166	11	gao	gao	PROPN
cana-5106	166	12	,	,	PUNCT
cana-5106	166	13	m.	m.	PROPN
cana-5106	166	14	r.	r.	PROPN
cana-5106	166	15	scott	scott	PROPN
cana-5106	166	16	,	,	PUNCT
cana-5106	166	17	and	and	CCONJ
cana-5106	166	18	w.	w.	PROPN
cana-5106	166	19	huang	huang	PROPN
cana-5106	166	20	,	,	PUNCT
cana-5106	166	21	“	"	PUNCT
cana-5106	166	22	tood	tood	PROPN
cana-5106	166	23	:	:	PUNCT
cana-5106	166	24	task	task	NOUN
cana-5106	166	25	-	-	PUNCT
cana-5106	166	26	aligned	align	VERB
cana-5106	166	27	onestage	onestage	NOUN
cana-5106	166	28	object	object	NOUN
cana-5106	166	29	detection	detection	NOUN
cana-5106	166	30	,	,	PUNCT
cana-5106	166	31	”	"	PUNCT
cana-5106	166	32	in	in	ADP
cana-5106	166	33	proceedings	proceeding	NOUN
cana-5106	166	34	of	of	ADP
cana-5106	166	35	the	the	DET
cana-5106	166	36	ieee	ieee	NOUN
cana-5106	166	37	/	/	SYM
cana-5106	166	38	cvf	cvf	NOUN
cana-5106	166	39	international	international	ADJ
cana-5106	166	40	conference	conference	NOUN
cana-5106	166	41	on	on	ADP
cana-5106	166	42	computer	computer	NOUN
cana-5106	166	43	vision	vision	NOUN
cana-5106	166	44	(	(	PUNCT
cana-5106	166	45	iccv	iccv	PROPN
cana-5106	166	46	)	)	PUNCT
cana-5106	166	47	,	,	PUNCT
cana-5106	166	48	2021	2021	NUM
cana-5106	166	49	,	,	PUNCT
cana-5106	166	50	pp	pp	ADJ
cana-5106	166	51	.	.	PUNCT
cana-5106	167	1	3510–3519	3510–3519	NUM
cana-5106	167	2	.	.	PUNCT
cana-5106	168	1	[	[	X
cana-5106	168	2	6	6	NUM
cana-5106	168	3	]	]	PUNCT
cana-5106	168	4	p.	p.	PROPN
cana-5106	168	5	gao	gao	PROPN
cana-5106	168	6	,	,	PUNCT
cana-5106	168	7	m.	m.	NOUN
cana-5106	168	8	zheng	zheng	PROPN
cana-5106	168	9	,	,	PUNCT
cana-5106	168	10	x.	x.	PROPN
cana-5106	168	11	wang	wang	PROPN
cana-5106	168	12	,	,	PUNCT
cana-5106	168	13	j.	j.	PROPN
cana-5106	168	14	dai	dai	PROPN
cana-5106	168	15	,	,	PUNCT
cana-5106	168	16	and	and	CCONJ
cana-5106	168	17	h.	h.	PROPN
cana-5106	168	18	li	li	PROPN
cana-5106	168	19	,	,	PUNCT
cana-5106	168	20	“	"	PUNCT
cana-5106	168	21	accelerating	accelerate	VERB
cana-5106	168	22	detr	detr	NOUN
cana-5106	168	23	convergence	convergence	NOUN
cana-5106	168	24	through	through	ADP
cana-5106	168	25	spatially	spatially	ADV
cana-5106	168	26	modulated	modulate	VERB
cana-5106	168	27	co	co	NOUN
cana-5106	168	28	-	-	NOUN
cana-5106	168	29	attention	attention	NOUN
cana-5106	168	30	,	,	PUNCT
cana-5106	168	31	”	"	PUNCT
cana-5106	168	32	arxiv	arxiv	PROPN
cana-5106	168	33	preprint	preprint	NOUN
cana-5106	168	34	arxiv:2108.02404	arxiv:2108.02404	NOUN
cana-5106	168	35	,	,	PUNCT
cana-5106	168	36	2021	2021	NUM
cana-5106	168	37	.	.	PUNCT
cana-5106	169	1	[	[	X
cana-5106	169	2	7	7	X
cana-5106	169	3	]	]	PUNCT
cana-5106	169	4	z.	z.	PROPN
cana-5106	169	5	zhang	zhang	PROPN
cana-5106	169	6	,	,	PUNCT
cana-5106	169	7	q.	q.	PROPN
cana-5106	169	8	wu	wu	PROPN
cana-5106	169	9	,	,	PUNCT
cana-5106	169	10	y.	y.	PROPN
cana-5106	169	11	wang	wang	PROPN
cana-5106	169	12	,	,	PUNCT
cana-5106	169	13	and	and	CCONJ
cana-5106	169	14	f.	f.	PROPN
cana-5106	169	15	chen	chen	PROPN
cana-5106	169	16	,	,	PUNCT
cana-5106	169	17	“	"	PUNCT
cana-5106	169	18	exploring	explore	VERB
cana-5106	169	19	region	region	NOUN
cana-5106	169	20	connections	connection	NOUN
cana-5106	169	21	implicitly	implicitly	ADV
cana-5106	169	22	:	:	PUNCT
cana-5106	169	23	image	image	NOUN
cana-5106	169	24	captioning	captioning	NOUN
cana-5106	169	25	with	with	ADP
cana-5106	169	26	visual	visual	ADJ
cana-5106	169	27	relationship	relationship	NOUN
cana-5106	169	28	attention	attention	NOUN
cana-5106	169	29	,	,	PUNCT
cana-5106	169	30	”	"	PUNCT
cana-5106	169	31	image	image	NOUN
cana-5106	169	32	and	and	CCONJ
cana-5106	169	33	vision	vision	NOUN
cana-5106	169	34	computing	computing	NOUN
cana-5106	169	35	,	,	PUNCT
cana-5106	169	36	vol	vol	NOUN
cana-5106	169	37	.	.	PROPN
cana-5106	169	38	109	109	NUM
cana-5106	169	39	,	,	PUNCT
cana-5106	169	40	p.	p.	NOUN
cana-5106	169	41	104146	104146	NUM
cana-5106	169	42	,	,	PUNCT
cana-5106	169	43	2021	2021	NUM
cana-5106	169	44	.	.	PUNCT
cana-5106	170	1	[	[	X
cana-5106	170	2	8	8	NUM
cana-5106	170	3	]	]	PUNCT
cana-5106	170	4	z.	z.	PROPN
cana-5106	170	5	gao	gao	PROPN
cana-5106	170	6	,	,	PUNCT
cana-5106	170	7	l.	l.	PROPN
cana-5106	170	8	wang	wang	PROPN
cana-5106	170	9	,	,	PUNCT
cana-5106	170	10	and	and	CCONJ
cana-5106	170	11	g.	g.	PROPN
cana-5106	170	12	wu	wu	PROPN
cana-5106	170	13	,	,	PUNCT
cana-5106	170	14	“	"	PUNCT
cana-5106	170	15	collaborative	collaborative	ADJ
cana-5106	170	16	supervision	supervision	NOUN
cana-5106	170	17	for	for	ADP
cana-5106	170	18	dense	dense	ADJ
cana-5106	170	19	object	object	NOUN
cana-5106	170	20	detection	detection	NOUN
cana-5106	170	21	,	,	PUNCT
cana-5106	170	22	”	"	PUNCT
cana-5106	170	23	in	in	ADP
cana-5106	170	24	proceedings	proceeding	NOUN
cana-5106	170	25	of	of	ADP
cana-5106	170	26	the	the	DET
cana-5106	170	27	ieee	ieee	NOUN
cana-5106	170	28	/	/	SYM
cana-5106	170	29	cvf	cvf	NOUN
cana-5106	170	30	international	international	ADJ
cana-5106	170	31	conference	conference	NOUN
cana-5106	170	32	on	on	ADP
cana-5106	170	33	computer	computer	NOUN
cana-5106	170	34	vision	vision	NOUN
cana-5106	170	35	(	(	PUNCT
cana-5106	170	36	iccv	iccv	PROPN
cana-5106	170	37	)	)	PUNCT
cana-5106	170	38	,	,	PUNCT
cana-5106	170	39	2021	2021	NUM
cana-5106	170	40	,	,	PUNCT
cana-5106	170	41	pp	pp	ADJ
cana-5106	170	42	.	.	PUNCT
cana-5106	170	43	3641–3650	3641–3650	NUM
cana-5106	170	44	.	.	PUNCT
cana-5106	171	1	[	[	X
cana-5106	171	2	9	9	NUM
cana-5106	171	3	]	]	PUNCT
cana-5106	171	4	j.	j.	PROPN
cana-5106	171	5	cho	cho	PROPN
cana-5106	171	6	,	,	PUNCT
cana-5106	171	7	j.	j.	PROPN
cana-5106	171	8	lei	lei	PROPN
cana-5106	171	9	,	,	PUNCT
cana-5106	171	10	h.	h.	PROPN
cana-5106	171	11	tan	tan	PROPN
cana-5106	171	12	,	,	PUNCT
cana-5106	171	13	and	and	CCONJ
cana-5106	171	14	m.	m.	NOUN
cana-5106	171	15	bansal	bansal	NOUN
cana-5106	171	16	explored	explore	VERB
cana-5106	171	17	the	the	DET
cana-5106	171	18	unification	unification	NOUN
cana-5106	171	19	of	of	ADP
cana-5106	171	20	vision	vision	NOUN
cana-5106	171	21	-	-	PUNCT
cana-5106	171	22	and	and	CCONJ
cana-5106	171	23	-	-	PUNCT
cana-5106	171	24	language	language	NOUN
cana-5106	171	25	tasks	task	NOUN
cana-5106	171	26	through	through	ADP
cana-5106	171	27	text	text	NOUN
cana-5106	171	28	generation	generation	NOUN
cana-5106	171	29	in	in	ADP
cana-5106	171	30	their	their	PRON
cana-5106	171	31	work	work	NOUN
cana-5106	171	32	titled	title	VERB
cana-5106	171	33	“	"	PUNCT
cana-5106	171	34	unifying	unify	VERB
cana-5106	171	35	vision	vision	NOUN
cana-5106	171	36	-	-	PUNCT
cana-5106	171	37	and	and	CCONJ
cana-5106	171	38	-	-	PUNCT
cana-5106	171	39	language	language	NOUN
cana-5106	171	40	tasks	task	NOUN
cana-5106	171	41	via	via	ADP
cana-5106	171	42	text	text	NOUN
cana-5106	171	43	generation	generation	NOUN
cana-5106	171	44	.	.	PUNCT
cana-5106	171	45	”	"	PUNCT
cana-5106	172	1	this	this	DET
cana-5106	172	2	study	study	NOUN
cana-5106	172	3	was	be	AUX
cana-5106	172	4	published	publish	VERB
cana-5106	172	5	as	as	ADP
cana-5106	172	6	an	an	DET
cana-5106	172	7	*	*	PROPN
cana-5106	172	8	arxiv	arxiv	PROPN
cana-5106	172	9	preprint	preprint	NOUN
cana-5106	172	10	*	*	PUNCT
cana-5106	172	11	(	(	PUNCT
cana-5106	172	12	arxiv:2102.02779	arxiv:2102.02779	NUM
cana-5106	172	13	)	)	PUNCT
cana-5106	172	14	in	in	ADP
cana-5106	172	15	2021	2021	NUM
cana-5106	172	16	.	.	PUNCT
cana-5106	173	1	[	[	X
cana-5106	173	2	10	10	NUM
cana-5106	173	3	]	]	X
cana-5106	173	4	s.	s.	PROPN
cana-5106	173	5	khan	khan	PROPN
cana-5106	173	6	,	,	PUNCT
cana-5106	173	7	m.	m.	NOUN
cana-5106	173	8	naseer	naseer	PROPN
cana-5106	173	9	,	,	PUNCT
cana-5106	173	10	m.	m.	NOUN
cana-5106	173	11	hayat	hayat	PROPN
cana-5106	173	12	,	,	PUNCT
cana-5106	173	13	s.	s.	PROPN
cana-5106	173	14	w.	w.	PROPN
cana-5106	173	15	zamir	zamir	PROPN
cana-5106	173	16	,	,	PUNCT
cana-5106	173	17	f.	f.	PROPN
cana-5106	173	18	s.	s.	PROPN
cana-5106	173	19	khan	khan	PROPN
cana-5106	173	20	,	,	PUNCT
cana-5106	173	21	and	and	CCONJ
cana-5106	173	22	m.	m.	NOUN
cana-5106	173	23	shah	shah	PROPN
cana-5106	173	24	conducted	conduct	VERB
cana-5106	173	25	a	a	DET
cana-5106	173	26	comprehensive	comprehensive	ADJ
cana-5106	173	27	survey	survey	NOUN
cana-5106	173	28	on	on	ADP
cana-5106	173	29	the	the	DET
cana-5106	173	30	application	application	NOUN
cana-5106	173	31	of	of	ADP
cana-5106	173	32	transformers	transformer	NOUN
cana-5106	173	33	in	in	ADP
cana-5106	173	34	vision	vision	NOUN
cana-5106	173	35	-	-	PUNCT
cana-5106	173	36	related	relate	VERB
cana-5106	173	37	tasks	task	NOUN
cana-5106	173	38	,	,	PUNCT
cana-5106	173	39	titled	title	VERB
cana-5106	173	40	“	"	PUNCT
cana-5106	173	41	transformers	transformer	NOUN
cana-5106	173	42	in	in	ADP
cana-5106	173	43	vision	vision	NOUN
cana-5106	173	44	:	:	PUNCT
cana-5106	173	45	a	a	DET
cana-5106	173	46	survey	survey	NOUN
cana-5106	173	47	.	.	PUNCT
cana-5106	173	48	”	"	PUNCT
cana-5106	174	1	their	their	PRON
cana-5106	174	2	findings	finding	NOUN
cana-5106	174	3	were	be	AUX
cana-5106	174	4	published	publish	VERB
cana-5106	174	5	as	as	ADP
cana-5106	174	6	an	an	DET
cana-5106	174	7	*	*	PROPN
cana-5106	174	8	arxiv	arxiv	PROPN
cana-5106	174	9	preprint	preprint	NOUN
cana-5106	174	10	*	*	PUNCT
cana-5106	174	11	(	(	PUNCT
cana-5106	174	12	arxiv:2101.01169	arxiv:2101.01169	NUM
cana-5106	174	13	)	)	PUNCT
cana-5106	174	14	in	in	ADP
cana-5106	174	15	2021	2021	NUM
cana-5106	174	16	.	.	PUNCT
cana-5106	175	1	[	[	X
cana-5106	175	2	11	11	NUM
cana-5106	175	3	]	]	PUNCT
cana-5106	175	4	j.	j.	PROPN
cana-5106	175	5	li	li	PROPN
cana-5106	175	6	,	,	PUNCT
cana-5106	175	7	r.	r.	PROPN
cana-5106	175	8	selvaraju	selvaraju	PROPN
cana-5106	175	9	,	,	PUNCT
cana-5106	175	10	a.	a.	PROPN
cana-5106	175	11	gotmare	gotmare	PROPN
cana-5106	175	12	,	,	PUNCT
cana-5106	175	13	s.	s.	PROPN
cana-5106	175	14	joty	joty	PROPN
cana-5106	175	15	,	,	PUNCT
cana-5106	175	16	c.	c.	PROPN
cana-5106	175	17	xiong	xiong	PROPN
cana-5106	175	18	,	,	PUNCT
cana-5106	175	19	and	and	CCONJ
cana-5106	175	20	s.	s.	PROPN
cana-5106	175	21	c.	c.	PROPN
cana-5106	175	22	h.	h.	PROPN
cana-5106	175	23	hoi	hoi	PROPN
cana-5106	175	24	,	,	PUNCT
cana-5106	175	25	“	"	PUNCT
cana-5106	175	26	align	align	VERB
cana-5106	175	27	before	before	ADP
cana-5106	175	28	fuse	fuse	NOUN
cana-5106	175	29	:	:	PUNCT
cana-5106	175	30	vision	vision	NOUN
cana-5106	175	31	and	and	CCONJ
cana-5106	175	32	language	language	NOUN
cana-5106	175	33	representation	representation	NOUN
cana-5106	175	34	learning	learn	VERB
cana-5106	175	35	with	with	ADP
cana-5106	175	36	momentum	momentum	NOUN
cana-5106	175	37	distillation	distillation	NOUN
cana-5106	175	38	,	,	PUNCT
cana-5106	175	39	”	"	PUNCT
cana-5106	175	40	advances	advance	NOUN
cana-5106	175	41	in	in	ADP
cana-5106	175	42	neural	neural	ADJ
cana-5106	175	43	information	information	NOUN
cana-5106	175	44	processing	processing	NOUN
cana-5106	175	45	systems	system	NOUN
cana-5106	175	46	(	(	PUNCT
cana-5106	175	47	neurips	neurip	NOUN
cana-5106	175	48	)	)	PUNCT
cana-5106	175	49	,	,	PUNCT
cana-5106	175	50	vol	vol	NOUN
cana-5106	175	51	.	.	PROPN
cana-5106	175	52	34	34	NUM
cana-5106	175	53	,	,	PUNCT
cana-5106	175	54	2021	2021	NUM
cana-5106	175	55	.	.	PUNCT