TSPT ReID: Triple Streams with Pos-insertion Transformer for Person Re-identification

TSPT ReID: Triple Streams with Pos-insertion Transformer for 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 TSPT ReID: Triple Streams with Pos-insertion Transformer for Person Re-identification Xujun Li, Liming Rao, Tengze Zhang, Jia Chang, Zhicheng Duan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4271156/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 In the process of person re-identification (ReID), the disparity of various datasets caused by different cameras and viewing distances poses challenges. Additionally, the inherent limitation of convolutional neural networks (CNNs) in capturing long-range dependencies exacerbates these challenges. To address these issues, this paper proposes the Triple Streams with Pos-insertion Transformer ReID (TSPT ReID) based on the Transformer architecture. The key design of TSPT ReID, the Triple Streams Transformer Encoder (TST Encoder), establishes multi-head attention layers on three different scales to enhance the correlation between various scales. To enhance the recognizability and robustness of valid pedestrian edges, the Shift Pos-insertion Module (SPI Module) is introduced by adding shift-based position coding in the attention layers without introducing additional parameters. The experimental results show the effectiveness of the model on multiple datasets, achieving 82.9% mAP and 91.4% Rank-1 accuracy on the DukeMTMC, and 83.5% mAP and 85.9% Rank-1 accuracy on CUKH03. Person re-identification transformer multiple fine-grained position encoding 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-4271156","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":291965699,"identity":"b1711cfb-7188-4c9f-9faf-fca103fc0faf","order_by":0,"name":"Xujun Li","email":"","orcid":"","institution":"Xiangtan University","correspondingAuthor":false,"prefix":"","firstName":"Xujun","middleName":"","lastName":"Li","suffix":""},{"id":291965700,"identity":"9fd04bf9-5364-4681-8fe1-040298018c44","order_by":1,"name":"Liming Rao","email":"data:image/png;base64,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","orcid":"","institution":"Xiangtan University","correspondingAuthor":true,"prefix":"","firstName":"Liming","middleName":"","lastName":"Rao","suffix":""},{"id":291965701,"identity":"706dfe16-161b-448c-8527-f2cc2b189150","order_by":2,"name":"Tengze Zhang","email":"","orcid":"","institution":"Xiangtan University","correspondingAuthor":false,"prefix":"","firstName":"Tengze","middleName":"","lastName":"Zhang","suffix":""},{"id":291965702,"identity":"6ef0ab4d-3541-4909-a052-840acfe1b765","order_by":3,"name":"Jia Chang","email":"","orcid":"","institution":"Xiangtan University","correspondingAuthor":false,"prefix":"","firstName":"Jia","middleName":"","lastName":"Chang","suffix":""},{"id":291965703,"identity":"e6c7fa17-3d98-44af-9199-0f227df4ff3a","order_by":4,"name":"Zhicheng Duan","email":"","orcid":"","institution":"Xiangtan University","correspondingAuthor":false,"prefix":"","firstName":"Zhicheng","middleName":"","lastName":"Duan","suffix":""}],"badges":[],"createdAt":"2024-04-15 17:27:20","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4271156/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4271156/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":59526950,"identity":"e7cf3dd4-dd92-4892-b76b-6c005d0db89d","added_by":"auto","created_at":"2024-07-02 21:33:16","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":652038,"visible":true,"origin":"","legend":"","description":"","filename":"TSPTReIDTripleStreamswithPosinsertionTransformerforPersonReidentification.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4271156/v1_covered_1130ea12-2ae9-4d9c-90fb-7f84fbdef774.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"TSPT ReID: Triple Streams with Pos-insertion Transformer for 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":"Person re-identification, transformer, multiple fine-grained, position encoding","lastPublishedDoi":"10.21203/rs.3.rs-4271156/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4271156/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn the process of person re-identification (ReID), the disparity of various datasets caused by different cameras and viewing distances poses challenges. Additionally, the inherent limitation of convolutional neural networks (CNNs) in capturing long-range dependencies exacerbates these challenges. To address these issues, this paper proposes the Triple Streams with Pos-insertion Transformer ReID (TSPT ReID) based on the Transformer architecture. The key design of TSPT ReID, the Triple Streams Transformer Encoder (TST Encoder), establishes multi-head attention layers on three different scales to enhance the correlation between various scales. To enhance the recognizability and robustness of valid pedestrian edges, the Shift Pos-insertion Module (SPI Module) is introduced by adding shift-based position coding in the attention layers without introducing additional parameters. The experimental results show the effectiveness of the model on multiple datasets, achieving 82.9% mAP and 91.4% Rank-1 accuracy on the DukeMTMC, and 83.5% mAP and 85.9% Rank-1 accuracy on CUKH03.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e","manuscriptTitle":"TSPT ReID: Triple Streams with Pos-insertion Transformer for Person Re-identification","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-19 03:04:36","doi":"10.21203/rs.3.rs-4271156/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":"8abf79cf-e0bf-4ea9-a8eb-c80fa4542deb","owner":[],"postedDate":"April 19th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-07-02T21:25:04+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-19 03:04:36","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4271156","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4271156","identity":"rs-4271156","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}