Cross-Modal Fine-grained Feature Alignment for Text-Based Person Re-identification

Cross-Modal Fine-grained Feature Alignment for Text-Based Person Re-identification | 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 Cross-Modal Fine-grained Feature Alignment for Text-Based Person Re-identification Qingfeng Lin, Chengfang Zhang, Xucheng Zhou, Ziliang Feng This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6769139/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Text-based Person Re-Identification aims to identify the corresponding target person based on natural language descriptions. Previous works focuse mainly on coarse-grained features while neglecting the critical role of fine-grained features in identification. To address this limitation, we propose a Cross-Modal Fine-grained Feature Alignment (CFFA) method to mine and align fine-grained information in text-based person re-identification. Firstly, we introduce a Fine-grained Local Feature Refinement (FLFR) approach to extract essential local fine-grained features. This approach also discards redundant information, introducing noise in the model. To better uncover latent fine-grained cues within texts, we adopt an implicit cross-modal relation reasoning method. It implicitly infers complex relationships between textual descriptions and image content, enhancing the model's understanding of fine-grained features. Then, we propose a Multi-association Learning (MAL) method to strengthen the model’s ability to distinguish between positive and negative samples. It improves classification performance, reinforcing interactions across different semantic levels. Comprehensive evaluations on three public datasets validate the effectiveness of our CFFA method, demonstrating superior performance compared to state-of-the-art approaches on key metrics. Our source code is available at https://github.com/lqfhello/CFFA . Text-based person re-identification Fine-grained information refinement Masked model Cross-modal feature align Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted 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-6769139","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":469646225,"identity":"f5c665af-234f-4551-bfd9-12eccf673352","order_by":0,"name":"Qingfeng Lin","email":"","orcid":"","institution":"Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Qingfeng","middleName":"","lastName":"Lin","suffix":""},{"id":469646226,"identity":"73c80a7e-4809-400d-96b8-73a33852cadf","order_by":1,"name":"Chengfang Zhang","email":"data:image/png;base64,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","orcid":"","institution":"Sichuan police college","correspondingAuthor":true,"prefix":"","firstName":"Chengfang","middleName":"","lastName":"Zhang","suffix":""},{"id":469646227,"identity":"03fedb0c-5e65-4208-a2e2-115921b0ad09","order_by":2,"name":"Xucheng Zhou","email":"","orcid":"","institution":"Sichuan Changhong NeoNet Technologies Co.. Ltd.","correspondingAuthor":false,"prefix":"","firstName":"Xucheng","middleName":"","lastName":"Zhou","suffix":""},{"id":469646228,"identity":"bc3b9040-22b4-41b6-b181-03ef18abab64","order_by":3,"name":"Ziliang Feng","email":"","orcid":"","institution":"Sichuan University","correspondingAuthor":false,"prefix":"","firstName":"Ziliang","middleName":"","lastName":"Feng","suffix":""}],"badges":[],"createdAt":"2025-05-28 14:38:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6769139/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6769139/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":86447541,"identity":"6d8834f2-f94e-434c-903b-2ee9f5f5044f","added_by":"auto","created_at":"2025-07-10 18:08:25","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":938905,"visible":true,"origin":"","legend":"","description":"","filename":"CrossModalFinegrainedFeatureAlignmentforTextBasedPersonReidentification.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6769139/v1_covered_1de949da-2dc7-4f57-b19d-565ef83d6f42.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Cross-Modal Fine-grained Feature Alignment for Text-Based Person Re-identification","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Text-based person re-identification, Fine-grained information refinement, Masked model, Cross-modal feature align","lastPublishedDoi":"10.21203/rs.3.rs-6769139/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6769139/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eText-based Person Re-Identification aims to identify the corresponding target person based on natural language descriptions. Previous works focuse mainly on coarse-grained features while neglecting the critical role of fine-grained features in identification. To address this limitation, we propose a Cross-Modal Fine-grained Feature Alignment (CFFA) method to mine and align fine-grained information in text-based person re-identification. Firstly, we introduce a Fine-grained Local Feature Refinement (FLFR) approach to extract essential local fine-grained features. This approach also discards redundant information, introducing noise in the model. To better uncover latent fine-grained cues within texts, we adopt an implicit cross-modal relation reasoning method. It implicitly infers complex relationships between textual descriptions and image content, enhancing the model's understanding of fine-grained features. Then, we propose a Multi-association Learning (MAL) method to strengthen the model\u0026rsquo;s ability to distinguish between positive and negative samples. It improves classification performance, reinforcing interactions across different semantic levels. Comprehensive evaluations on three public datasets validate the effectiveness of our CFFA method, demonstrating superior performance compared to state-of-the-art approaches on key metrics. Our source code is available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/lqfhello/CFFA\u003c/span\u003e\u003cspan address=\"https://github.com/lqfhello/CFFA\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e","manuscriptTitle":"Cross-Modal Fine-grained Feature Alignment for Text-Based Person Re-identification","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-06-12 06:57:26","doi":"10.21203/rs.3.rs-6769139/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"5b6902fe-74a2-4769-a373-51a8a61e8974","owner":[],"postedDate":"June 12th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-07-10T18:08:12+00:00","versionOfRecord":[],"versionCreatedAt":"2025-06-12 06:57:26","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6769139","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6769139","identity":"rs-6769139","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}