Non-Salient Visual Content Grounding for Multimodal Relation Extraction

Non-Salient Visual Content Grounding for Multimodal Relation Extraction | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Non-Salient Visual Content Grounding for Multimodal Relation Extraction Zefan Zhang, Yanhui Li, Weiqi Zhang, Tian Bai This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7395065/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 20 You are reading this latest preprint version Abstract Multimodal Relation Extraction (MRE) seeks to identify relations between textual entities with visual context. However, current models struggle when visual information is non-salient or weakly relevant, primarily due to two factors: (1) independent visual feature extraction often prioritizes biased semantics for salient visual contents, hindering attention to non-salient details; (2) classifier-based methods struggle to filter out spurious multimodal correlations, particularly in scenarios involving non-salient visual cues. To address these issues, we propose a N on- S alient V isual C ontent G rounding ( NSVCG ) Network that leverages instruction-following Multimodal Large Language Models (MLLMs) with entity-guided prompts to extract relevant but non-salient visual features. Further, a conditional diffusion mechanism iteratively refines predictions by eliminating spurious multimodal correlations. As the test bed, we introduce a new dataset, VG-MNRE, extending MNRE test set with 1614 samples and manually annotated grounding labels. Experimental results show that NSVCG outperforms state-of-the-art baselines by 1.2% F1 on MNRE and 6.76% grounding accuracy on VG-MNRE, demonstrating improved robustness and relevant visual content grounding ability. The code will be released upon publication. Multimodal Relation Extraction Visual Content Grounding Multimodal Large Language Models Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 27 Mar, 2026 Reviews received at journal 16 Dec, 2025 Reviewers agreed at journal 15 Dec, 2025 Reviews received at journal 13 Dec, 2025 Reviews received at journal 12 Dec, 2025 Reviewers agreed at journal 11 Dec, 2025 Reviewers agreed at journal 11 Dec, 2025 Reviews received at journal 11 Dec, 2025 Reviewers agreed at journal 10 Dec, 2025 Reviewers agreed at journal 10 Dec, 2025 Reviewers agreed at journal 10 Dec, 2025 Reviewers agreed at journal 09 Dec, 2025 Reviewers agreed at journal 09 Dec, 2025 Reviewers agreed at journal 09 Dec, 2025 Reviewers agreed at journal 09 Dec, 2025 Reviewers agreed at journal 09 Dec, 2025 Reviewers invited by journal 09 Dec, 2025 Editor assigned by journal 02 Sep, 2025 Submission checks completed at journal 18 Aug, 2025 First submitted to journal 17 Aug, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7395065","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":559737766,"identity":"326b7176-2c7a-480b-be69-e8aeb63ea06e","order_by":0,"name":"Zefan Zhang","email":"","orcid":"","institution":"Jilin University","correspondingAuthor":false,"prefix":"","firstName":"Zefan","middleName":"","lastName":"Zhang","suffix":""},{"id":559737767,"identity":"669dd11c-7648-41a4-be31-c4d65fcc0b3d","order_by":1,"name":"Yanhui Li","email":"","orcid":"","institution":"Jilin University","correspondingAuthor":false,"prefix":"","firstName":"Yanhui","middleName":"","lastName":"Li","suffix":""},{"id":559737768,"identity":"b733b042-9d3c-4582-a609-b30eed77a72e","order_by":2,"name":"Weiqi Zhang","email":"","orcid":"","institution":"Jilin University","correspondingAuthor":false,"prefix":"","firstName":"Weiqi","middleName":"","lastName":"Zhang","suffix":""},{"id":559737769,"identity":"a211ef37-262d-4376-8986-17ea11e4f354","order_by":3,"name":"Tian Bai","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA00lEQVRIie3OMQuCQBTA8XcIthiuTvoVlAaRvoxOTkGTU8PJgS7SLPQp+gbGg0ZbHRzOpbmWcCm6c2jMG4PuvxwH78d7ADrdD+YDoRADuKSUX1ON5FSQFWHqRK4BSKihSsIFlpzv+pQxy4dbhmAf6HcSVYk47Hzd5IKQukVw+mbmsEYSEydiLAsE34lnyGUQ5IUpkeSpRDqxJSkwnghRIVE95HWyxyBn5vZUtanldDMktFN+Hx/oBSU78jFbu3Y9Qz4FFKARr6U4L/LUR3U6ne7fegPTQ0b4GZf6uAAAAABJRU5ErkJggg==","orcid":"","institution":"Jilin University","correspondingAuthor":true,"prefix":"","firstName":"Tian","middleName":"","lastName":"Bai","suffix":""}],"badges":[],"createdAt":"2025-08-18 02:38:05","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7395065/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7395065/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98210970,"identity":"c3d1bca9-10ec-4c31-b344-ad157c2261ee","added_by":"auto","created_at":"2025-12-15 09:36:28","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5374,"visible":true,"origin":"","legend":"","description":"","filename":"84908dd47d2b427d96b2c5f4d2e9095b.json","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/0e32d42b89795c1afe129d67.json"},{"id":98210973,"identity":"018162f4-1042-43e0-b37e-a30040570f22","added_by":"auto","created_at":"2025-12-15 09:36:28","extension":"xml","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":151083,"visible":true,"origin":"","legend":"","description":"","filename":"84908dd47d2b427d96b2c5f4d2e9095b1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/eb8689387975d4967779b64d.xml"},{"id":98432845,"identity":"2ba10393-bb23-4c57-bcf7-28ef0f582181","added_by":"auto","created_at":"2025-12-17 16:50:02","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1785640,"visible":true,"origin":"","legend":"","description":"","filename":"KAISVGMRE.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/ad67c64e4c37ac35fc837490.pdf"},{"id":98210974,"identity":"7636a762-3865-46ea-b2e2-45b95267d16f","added_by":"auto","created_at":"2025-12-15 09:36:28","extension":"pdf","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":14329,"visible":true,"origin":"","legend":"","description":"","filename":"beta.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/8ddd93aad3ed4815e9c39e0a.pdf"},{"id":98210971,"identity":"37e4df0e-bf90-48b1-a07a-643bdcdd40bb","added_by":"auto","created_at":"2025-12-15 09:36:28","extension":"eps","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2890,"visible":true,"origin":"","legend":"","description":"","filename":"empty.eps","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/81ba1761d18e7cd518040feb.eps"},{"id":98210980,"identity":"751c729b-b4e1-4213-9e15-ff6eb7064c5a","added_by":"auto","created_at":"2025-12-15 09:36:29","extension":"eps","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":91593,"visible":true,"origin":"","legend":"","description":"","filename":"fig.eps","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/2d9b9ac9c4684b1e0c4b2005.eps"},{"id":98210982,"identity":"8045d0de-d031-481d-972e-7512ef8c69c3","added_by":"auto","created_at":"2025-12-15 09:36:29","extension":"bst","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":147148,"visible":true,"origin":"","legend":"","description":"","filename":"snapacite.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/25e86a3d947aca4c7f8225b2.bst"},{"id":98431605,"identity":"4c1c1240-9ef8-44b0-9728-b573d167178b","added_by":"auto","created_at":"2025-12-17 16:48:01","extension":"bst","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":30423,"visible":true,"origin":"","legend":"","description":"","filename":"snaps.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/6334070c8847749bd6e3f304.bst"},{"id":98432182,"identity":"91ec2076-a739-4a91-86f4-b524f8c405c2","added_by":"auto","created_at":"2025-12-17 16:49:10","extension":"pdf","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":390370,"visible":true,"origin":"","legend":"","description":"","filename":"snarticle.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/d59f863a5811960ff88ddab9.pdf"},{"id":98431440,"identity":"8e9c026d-1f4e-435e-a0af-513f938dff86","added_by":"auto","created_at":"2025-12-17 16:47:41","extension":"bst","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":35733,"visible":true,"origin":"","legend":"","description":"","filename":"snbasic.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/11aacc478c04733094691c66.bst"},{"id":98210978,"identity":"13a6d86b-afcb-46f4-ac8c-61e56ab2afa0","added_by":"auto","created_at":"2025-12-15 09:36:28","extension":"bst","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":40398,"visible":true,"origin":"","legend":"","description":"","filename":"snchicago.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/ee6cb19ba2dab57307f752d4.bst"},{"id":98433339,"identity":"e9e836a8-9d2e-4bc8-8954-5ff8cd8139ca","added_by":"auto","created_at":"2025-12-17 16:50:39","extension":"cls","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55331,"visible":true,"origin":"","legend":"","description":"","filename":"snjnl.cls","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/c116d9d10b91becbc6cfc07f.cls"},{"id":98210976,"identity":"b9621b54-99cc-4b7e-a629-ce6877767903","added_by":"auto","created_at":"2025-12-15 09:36:28","extension":"bst","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":64140,"visible":true,"origin":"","legend":"","description":"","filename":"snmathphysay.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/be5c85279ed5f0394affff4a.bst"},{"id":98433577,"identity":"d284ced9-e4fc-483f-bbc4-02ab4e242c26","added_by":"auto","created_at":"2025-12-17 16:50:55","extension":"bst","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":64141,"visible":true,"origin":"","legend":"","description":"","filename":"snmathphysnum.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/30769ee0cb0b7851d78c48b9.bst"},{"id":98432126,"identity":"1210471b-af5b-4f1a-a09d-0e3392a996c6","added_by":"auto","created_at":"2025-12-17 16:49:03","extension":"bst","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38349,"visible":true,"origin":"","legend":"","description":"","filename":"snnature.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/60eccb955b6f7772e12d51d3.bst"},{"id":98432112,"identity":"e584fd90-b2a0-4a1d-9975-aa5e2b836ee0","added_by":"auto","created_at":"2025-12-17 16:49:00","extension":"bst","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":41304,"visible":true,"origin":"","legend":"","description":"","filename":"snvancouver.bst","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/1645e191f9321e7e57917498.bst"},{"id":98210987,"identity":"e1ba3c57-f1ff-44c1-b3c1-823d3f5ba38c","added_by":"auto","created_at":"2025-12-15 09:36:29","extension":"pdf","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":415247,"visible":true,"origin":"","legend":"","description":"","filename":"usermanual.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/d10383d56295996b57ab4124.pdf"},{"id":98210985,"identity":"9d934641-a711-49ed-a9cc-6912cd018992","added_by":"auto","created_at":"2025-12-15 09:36:29","extension":"xml","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":161157,"visible":true,"origin":"","legend":"","description":"","filename":"84908dd47d2b427d96b2c5f4d2e9095b1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/17753b41c8d2f1bec28c7d12.xml"},{"id":98433072,"identity":"a6836fe5-0502-4d1a-ab47-9d195d75c518","added_by":"auto","created_at":"2025-12-17 16:50:13","extension":"html","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":170841,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1/c5588a1ffb08acd5a8829e7e.html"},{"id":98444974,"identity":"5b41e995-1552-4462-89ec-f8840593b571","added_by":"auto","created_at":"2025-12-17 17:18:30","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1135281,"visible":true,"origin":"","legend":"","description":"","filename":"KAISVGMRE.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7395065/v1_covered_faa8031e-d3b7-47b7-8666-e9f02d23350c.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Non-Salient Visual Content Grounding for Multimodal Relation Extraction","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"knowledge-and-information-systems","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"kais","sideBox":"Learn more about [Knowledge and Information Systems](http://link.springer.com/journal/10115)","snPcode":"10115","submissionUrl":"https://submission.nature.com/new-submission/10115/3","title":"Knowledge and Information Systems","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Multimodal Relation Extraction, Visual Content Grounding, Multimodal Large Language Models","lastPublishedDoi":"10.21203/rs.3.rs-7395065/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7395065/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eMultimodal Relation Extraction (MRE) seeks to identify relations between textual entities with visual context. However, current models struggle when visual information is non-salient or weakly relevant, primarily due to two factors: (1) independent visual feature extraction often prioritizes biased semantics for salient visual contents, hindering attention to non-salient details; (2) classifier-based methods struggle to filter out spurious multimodal correlations, particularly in scenarios involving non-salient visual cues. To address these issues, we propose a \u003cb\u003eN\u003c/b\u003eon-\u003cb\u003eS\u003c/b\u003ealient \u003cb\u003eV\u003c/b\u003eisual \u003cb\u003eC\u003c/b\u003eontent \u003cb\u003eG\u003c/b\u003erounding (\u003cb\u003eNSVCG\u003c/b\u003e) Network that leverages instruction-following Multimodal Large Language Models (MLLMs) with entity-guided prompts to extract relevant but non-salient visual features. Further, a conditional diffusion mechanism iteratively refines predictions by eliminating spurious multimodal correlations. As the test bed, we introduce a new dataset, VG-MNRE, extending MNRE test set with 1614 samples and manually annotated grounding labels. Experimental results show that NSVCG outperforms state-of-the-art baselines by 1.2% F1 on MNRE and 6.76% grounding accuracy on VG-MNRE, demonstrating improved robustness and relevant visual content grounding ability. The code will be released upon publication.\u003c/p\u003e","manuscriptTitle":"Non-Salient Visual Content Grounding for Multimodal Relation Extraction","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-15 09:36:22","doi":"10.21203/rs.3.rs-7395065/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-03-27T10:50:03+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-17T01:43:57+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"303967756651839345044770672862857662191","date":"2025-12-15T12:34:44+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-13T08:45:24+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-12T08:15:25+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"191285620665717728378119761604221891992","date":"2025-12-11T13:58:01+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"104727522369653625158397744975906455417","date":"2025-12-11T13:50:37+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2025-12-11T05:46:29+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"206112043518507199180611588188641285907","date":"2025-12-10T14:51:08+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"339630017576680118699762161918511819191","date":"2025-12-10T12:57:08+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"177950069566949934133276638515067090151","date":"2025-12-10T06:10:11+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"264418069353198587973003889883899064978","date":"2025-12-10T02:07:54+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"74074959052836036352907262411562944884","date":"2025-12-09T13:53:49+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"251901713571344179979681111311274890613","date":"2025-12-09T13:49:36+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"266338330870355312445430893823597509265","date":"2025-12-09T13:46:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"224308591972158163235711002479491970654","date":"2025-12-09T13:44:42+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2025-12-09T13:38:04+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-09-02T13:07:05+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-08-18T15:08:47+00:00","index":"","fulltext":""},{"type":"submitted","content":"Knowledge and Information Systems","date":"2025-08-18T02:22:05+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"knowledge-and-information-systems","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"kais","sideBox":"Learn more about [Knowledge and Information Systems](http://link.springer.com/journal/10115)","snPcode":"10115","submissionUrl":"https://submission.nature.com/new-submission/10115/3","title":"Knowledge and Information Systems","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"88dca9f6-faac-465f-a695-e54d17382c0b","owner":[],"postedDate":"December 15th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-08T14:25:00+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-15 09:36:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7395065","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7395065","identity":"rs-7395065","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}