Grounded Image Text Matching with Mismatched Relation Reasoning

This paper introduces Grounded Image Text Matching with Mismatched Relation (GITM-MR), a novel visual-linguistic joint task that evaluates the relation understanding capabilities of transformer-based pre-trained models. GITM-MR requires a model to first determine if an expression describes an image,...

Full description

Saved in:
Bibliographic Details
Published in:arXiv.org 2023-08
Main Authors: Wu, Yu, Wei, Yana, Wang, Haozhe, Liu, Yongfei, Yang, Sibei, He, Xuming
Format: Article
Language:English
Subjects:
Matching
Modular structures
Reasoning
Visual tasks
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
cited_by
cites
container_end_page
container_issue
container_start_page
container_title arXiv.org
container_volume
creator Wu, Yu
Wei, Yana
Wang, Haozhe
Liu, Yongfei
Yang, Sibei
He, Xuming
description This paper introduces Grounded Image Text Matching with Mismatched Relation (GITM-MR), a novel visual-linguistic joint task that evaluates the relation understanding capabilities of transformer-based pre-trained models. GITM-MR requires a model to first determine if an expression describes an image, then localize referred objects or ground the mismatched parts of the text. We provide a benchmark for evaluating pre-trained models on this task, with a focus on the challenging settings of limited data and out-of-distribution sentence lengths. Our evaluation demonstrates that pre-trained models lack data efficiency and length generalization ability. To address this, we propose the Relation-sensitive Correspondence Reasoning Network (RCRN), which incorporates relation-aware reasoning via bi-directional message propagation guided by language structure. RCRN can be interpreted as a modular program and delivers strong performance in both length generalization and data efficiency.
format article
fullrecord <record><control><sourceid>proquest</sourceid><recordid>TN_cdi_proquest_journals_2845345539</recordid><sourceformat>XML</sourceformat><sourcesystem>PC</sourcesystem><sourcerecordid>2845345539</sourcerecordid><originalsourceid>FETCH-proquest_journals_28453455393</originalsourceid><addsrcrecordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSwdy_KL81LSU1R8MxNTE9VCEmtKFHwTSxJzsjMS1cozyzJUPDNLM4FCQDVBKXmJJZk5ucBGYnF-XlAJTwMrGmJOcWpvFCam0HZzTXE2UO3oCi_sDS1uCQ-K7-0KA8oFW9kYWJqbAK01dKYOFUA2-c4jg</addsrcrecordid><sourcetype>Aggregation Database</sourcetype><iscdi>true</iscdi><recordtype>article</recordtype><pqid>2845345539</pqid></control><display><type>article</type><title>Grounded Image Text Matching with Mismatched Relation Reasoning</title><source>Publicly Available Content Database</source><creator>Wu, Yu ; Wei, Yana ; Wang, Haozhe ; Liu, Yongfei ; Yang, Sibei ; He, Xuming</creator><creatorcontrib>Wu, Yu ; Wei, Yana ; Wang, Haozhe ; Liu, Yongfei ; Yang, Sibei ; He, Xuming</creatorcontrib><description>This paper introduces Grounded Image Text Matching with Mismatched Relation (GITM-MR), a novel visual-linguistic joint task that evaluates the relation understanding capabilities of transformer-based pre-trained models. GITM-MR requires a model to first determine if an expression describes an image, then localize referred objects or ground the mismatched parts of the text. We provide a benchmark for evaluating pre-trained models on this task, with a focus on the challenging settings of limited data and out-of-distribution sentence lengths. Our evaluation demonstrates that pre-trained models lack data efficiency and length generalization ability. To address this, we propose the Relation-sensitive Correspondence Reasoning Network (RCRN), which incorporates relation-aware reasoning via bi-directional message propagation guided by language structure. RCRN can be interpreted as a modular program and delivers strong performance in both length generalization and data efficiency.</description><identifier>EISSN: 2331-8422</identifier><language>eng</language><publisher>Ithaca: Cornell University Library, arXiv.org</publisher><subject>Matching ; Modular structures ; Reasoning ; Visual tasks</subject><ispartof>arXiv.org, 2023-08</ispartof><rights>2023. This work is published under http://creativecommons.org/licenses/by/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.</rights><oa>free_for_read</oa><woscitedreferencessubscribed>false</woscitedreferencessubscribed></display><links><openurl>$$Topenurl_article</openurl><openurlfulltext>$$Topenurlfull_article</openurlfulltext><thumbnail>$$Tsyndetics_thumb_exl</thumbnail><linktohtml>$$Uhttps://www.proquest.com/docview/2845345539?pq-origsite=primo$$EHTML$$P50$$Gproquest$$Hfree_for_read</linktohtml><link.rule.ids>780,784,25753,37012,44590</link.rule.ids></links><search><creatorcontrib>Wu, Yu</creatorcontrib><creatorcontrib>Wei, Yana</creatorcontrib><creatorcontrib>Wang, Haozhe</creatorcontrib><creatorcontrib>Liu, Yongfei</creatorcontrib><creatorcontrib>Yang, Sibei</creatorcontrib><creatorcontrib>He, Xuming</creatorcontrib><title>Grounded Image Text Matching with Mismatched Relation Reasoning</title><title>arXiv.org</title><description>This paper introduces Grounded Image Text Matching with Mismatched Relation (GITM-MR), a novel visual-linguistic joint task that evaluates the relation understanding capabilities of transformer-based pre-trained models. GITM-MR requires a model to first determine if an expression describes an image, then localize referred objects or ground the mismatched parts of the text. We provide a benchmark for evaluating pre-trained models on this task, with a focus on the challenging settings of limited data and out-of-distribution sentence lengths. Our evaluation demonstrates that pre-trained models lack data efficiency and length generalization ability. To address this, we propose the Relation-sensitive Correspondence Reasoning Network (RCRN), which incorporates relation-aware reasoning via bi-directional message propagation guided by language structure. RCRN can be interpreted as a modular program and delivers strong performance in both length generalization and data efficiency.</description><subject>Matching</subject><subject>Modular structures</subject><subject>Reasoning</subject><subject>Visual tasks</subject><issn>2331-8422</issn><fulltext>true</fulltext><rsrctype>article</rsrctype><creationdate>2023</creationdate><recordtype>article</recordtype><sourceid>PIMPY</sourceid><recordid>eNpjYuA0MjY21LUwMTLiYOAtLs4yMDAwMjM3MjU15mSwdy_KL81LSU1R8MxNTE9VCEmtKFHwTSxJzsjMS1cozyzJUPDNLM4FCQDVBKXmJJZk5ucBGYnF-XlAJTwMrGmJOcWpvFCam0HZzTXE2UO3oCi_sDS1uCQ-K7-0KA8oFW9kYWJqbAK01dKYOFUA2-c4jg</recordid><startdate>20230804</startdate><enddate>20230804</enddate><creator>Wu, Yu</creator><creator>Wei, Yana</creator><creator>Wang, Haozhe</creator><creator>Liu, Yongfei</creator><creator>Yang, Sibei</creator><creator>He, Xuming</creator><general>Cornell University Library, arXiv.org</general><scope>8FE</scope><scope>8FG</scope><scope>ABJCF</scope><scope>ABUWG</scope><scope>AFKRA</scope><scope>AZQEC</scope><scope>BENPR</scope><scope>BGLVJ</scope><scope>CCPQU</scope><scope>DWQXO</scope><scope>HCIFZ</scope><scope>L6V</scope><scope>M7S</scope><scope>PIMPY</scope><scope>PQEST</scope><scope>PQQKQ</scope><scope>PQUKI</scope><scope>PTHSS</scope></search><sort><creationdate>20230804</creationdate><title>Grounded Image Text Matching with Mismatched Relation Reasoning</title><author>Wu, Yu ; Wei, Yana ; Wang, Haozhe ; Liu, Yongfei ; Yang, Sibei ; He, Xuming</author></sort><facets><frbrtype>5</frbrtype><frbrgroupid>cdi_FETCH-proquest_journals_28453455393</frbrgroupid><rsrctype>articles</rsrctype><prefilter>articles</prefilter><language>eng</language><creationdate>2023</creationdate><topic>Matching</topic><topic>Modular structures</topic><topic>Reasoning</topic><topic>Visual tasks</topic><toplevel>online_resources</toplevel><creatorcontrib>Wu, Yu</creatorcontrib><creatorcontrib>Wei, Yana</creatorcontrib><creatorcontrib>Wang, Haozhe</creatorcontrib><creatorcontrib>Liu, Yongfei</creatorcontrib><creatorcontrib>Yang, Sibei</creatorcontrib><creatorcontrib>He, Xuming</creatorcontrib><collection>ProQuest SciTech Collection</collection><collection>ProQuest Technology Collection</collection><collection>Materials Science &amp; Engineering Collection</collection><collection>ProQuest Central (Alumni)</collection><collection>ProQuest Central</collection><collection>ProQuest Central Essentials</collection><collection>ProQuest Central</collection><collection>Technology Collection</collection><collection>ProQuest One Community College</collection><collection>ProQuest Central</collection><collection>SciTech Premium Collection</collection><collection>ProQuest Engineering Collection</collection><collection>Engineering Database</collection><collection>Publicly Available Content Database</collection><collection>ProQuest One Academic Eastern Edition (DO NOT USE)</collection><collection>ProQuest One Academic</collection><collection>ProQuest One Academic UKI Edition</collection><collection>Engineering collection</collection></facets><delivery><delcategory>Remote Search Resource</delcategory><fulltext>fulltext</fulltext></delivery><addata><au>Wu, Yu</au><au>Wei, Yana</au><au>Wang, Haozhe</au><au>Liu, Yongfei</au><au>Yang, Sibei</au><au>He, Xuming</au><format>book</format><genre>document</genre><ristype>GEN</ristype><atitle>Grounded Image Text Matching with Mismatched Relation Reasoning</atitle><jtitle>arXiv.org</jtitle><date>2023-08-04</date><risdate>2023</risdate><eissn>2331-8422</eissn><abstract>This paper introduces Grounded Image Text Matching with Mismatched Relation (GITM-MR), a novel visual-linguistic joint task that evaluates the relation understanding capabilities of transformer-based pre-trained models. GITM-MR requires a model to first determine if an expression describes an image, then localize referred objects or ground the mismatched parts of the text. We provide a benchmark for evaluating pre-trained models on this task, with a focus on the challenging settings of limited data and out-of-distribution sentence lengths. Our evaluation demonstrates that pre-trained models lack data efficiency and length generalization ability. To address this, we propose the Relation-sensitive Correspondence Reasoning Network (RCRN), which incorporates relation-aware reasoning via bi-directional message propagation guided by language structure. RCRN can be interpreted as a modular program and delivers strong performance in both length generalization and data efficiency.</abstract><cop>Ithaca</cop><pub>Cornell University Library, arXiv.org</pub><oa>free_for_read</oa></addata></record>
fulltext fulltext
identifier EISSN: 2331-8422
ispartof arXiv.org, 2023-08
issn 2331-8422
language eng
recordid cdi_proquest_journals_2845345539
source Publicly Available Content Database
subjects Matching
Modular structures
Reasoning
Visual tasks
title Grounded Image Text Matching with Mismatched Relation Reasoning
url http://sfxeu10.hosted.exlibrisgroup.com/loughborough?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF-8&ctx_tim=2025-01-02T12%3A56%3A11IST&url_ver=Z39.88-2004&url_ctx_fmt=infofi/fmt:kev:mtx:ctx&rfr_id=info:sid/primo.exlibrisgroup.com:primo3-Article-proquest&rft_val_fmt=info:ofi/fmt:kev:mtx:book&rft.genre=document&rft.atitle=Grounded%20Image%20Text%20Matching%20with%20Mismatched%20Relation%20Reasoning&rft.jtitle=arXiv.org&rft.au=Wu,%20Yu&rft.date=2023-08-04&rft.eissn=2331-8422&rft_id=info:doi/&rft_dat=%3Cproquest%3E2845345539%3C/proquest%3E%3Cgrp_id%3Ecdi_FETCH-proquest_journals_28453455393%3C/grp_id%3E%3Coa%3E%3C/oa%3E%3Curl%3E%3C/url%3E&rft_id=info:oai/&rft_pqid=2845345539&rft_id=info:pmid/&rfr_iscdi=true