Feature Fusion Units for Fine-grained Image Categorization

Feature Fusion Units for Fine-grained Image Categorization | 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 Feature Fusion Units for Fine-grained Image Categorization Hua Zhao, Zujun Liu, Bin Yang, Tianyu Lu, Ying Xing This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7717226/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 Fine-grained image categorization aims to categorize subclasses by processing detailed features, which is still a critical problem to be solved in computer version due to the small differences between subclasses. The traditional methods are usually to find features by manual annotation, using specific sliding Windows, using different thresholds and other methods. These methods are not only costly, but also ineffective. In computer version, by calculating attention scores between parts of the picture multiple times and weighting them, the transformer greatly improves the accuracy of categorization. In this paper, we propose a feature weight units. Specifically, transformer is used as the backbone to capture image feature(these features are called patches in transformer), and then all patches are weighted by our feature weight unit. The computal result of feature fusion unit represents the importance of the patch should to be forced on. To verify the effectiveness of our method, we conducted experiments on the CUB-200-2011 and stanford-dog datasets. Fine-grained image categorization transformer computer vision feature weight unit 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-7717226","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":543352474,"identity":"9a1791be-bb56-48a7-9756-1997d8b5bbfa","order_by":0,"name":"Hua Zhao","email":"","orcid":"","institution":"Smart Steps Digital Technology Co., Ltd.","correspondingAuthor":false,"prefix":"","firstName":"Hua","middleName":"","lastName":"Zhao","suffix":""},{"id":543352475,"identity":"773fd4ac-7ee1-4212-b8e8-01cac2a1c642","order_by":1,"name":"Zujun Liu","email":"","orcid":"","institution":"Smart Steps Digital Technology Co., Ltd.","correspondingAuthor":false,"prefix":"","firstName":"Zujun","middleName":"","lastName":"Liu","suffix":""},{"id":543352476,"identity":"0a4d1784-bbb2-493e-b679-ef4788cbc678","order_by":2,"name":"Bin Yang","email":"","orcid":"","institution":"China Unicom Research Institute","correspondingAuthor":false,"prefix":"","firstName":"Bin","middleName":"","lastName":"Yang","suffix":""},{"id":543352477,"identity":"70377936-3346-4153-bc17-8685acd6ef42","order_by":3,"name":"Tianyu Lu","email":"","orcid":"","institution":"Beijing University of Posts and Telecommunications","correspondingAuthor":false,"prefix":"","firstName":"Tianyu","middleName":"","lastName":"Lu","suffix":""},{"id":543352478,"identity":"e85ca6a7-e059-4895-b87c-0d8243f3030d","order_by":4,"name":"Ying Xing","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA50lEQVRIie3RLQvCQBjA8WcMlk5Xb4Lf4YGDMyjzqzgGJoPNxcGqmBX9EAPBfHCgZWAVXBCEFQ2zGQSdE5vcFg33D/cSfhzHA6DT/WPme0EAm37uRlibOPNiF7XINzzUJbhryAsZp+76OOXnHHrtWJjZSUWcqDnsEsz8TZp0UMCQxcLqoIrYJuGMoPT5YcSpAOnFglhURawvYfOSPKtJ8Qo7F8RFWhJRTZyIcGOFckDT7QQT9NlCWlxJcJ+w/PqQfXsZbU5B4LZnuyhTkvd3WgTAC4tDOdPPcNWZtztAH0qi0+l0uh+9ADYmR/Ilvs1HAAAAAElFTkSuQmCC","orcid":"","institution":"Beijing University of Posts and Telecommunications","correspondingAuthor":true,"prefix":"","firstName":"Ying","middleName":"","lastName":"Xing","suffix":""}],"badges":[],"createdAt":"2025-09-26 03:38:08","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7717226/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7717226/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":95836808,"identity":"e2fc7671-2e0c-479f-a5c7-4b9abbaaaac2","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"json","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":6390,"visible":true,"origin":"","legend":"","description":"","filename":"5f5c9fbd1b5d49369cbaaf01fc5ad29a.json","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/0bd05ae4b6ced97f82f7fe0a.json"},{"id":95836809,"identity":"dd484204-ef3f-41ad-ae74-0bd1c2f95992","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"xml","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":58243,"visible":true,"origin":"","legend":"","description":"","filename":"5f5c9fbd1b5d49369cbaaf01fc5ad29a1enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/64ecedd62ec9f459d3d3ded5.xml"},{"id":95836819,"identity":"c980d5e6-b7db-4cbe-b42b-1785d9aa9eb3","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"pdf","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2845625,"visible":true,"origin":"","legend":"","description":"","filename":"FeatureFusionUnitsforFinegrainedImageCategorization.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/574be614135044d72e92edcb.pdf"},{"id":95836811,"identity":"3ec97fe4-1889-41e2-abbb-109b83381c8f","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"pdf","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":156905,"visible":true,"origin":"","legend":"","description":"","filename":"Pointbypointresponse.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/896c250effea70f830fba0c5.pdf"},{"id":96239783,"identity":"196d06e2-f1cd-48f8-bd02-aacfed2c4c09","added_by":"auto","created_at":"2025-11-19 07:07:39","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":136431,"visible":true,"origin":"","legend":"","description":"","filename":"all.png","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/3977571bb7025d646c4b0ed8.png"},{"id":95836816,"identity":"73249b86-2478-48e7-8439-d7f5fc1268c7","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1097364,"visible":true,"origin":"","legend":"","description":"","filename":"fenxi.png","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/b63260ec092d858b8b59e1a0.png"},{"id":96240750,"identity":"d15773a7-23f4-4777-b847-15b20d00c78c","added_by":"auto","created_at":"2025-11-19 07:09:29","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1597946,"visible":true,"origin":"","legend":"","description":"","filename":"show.png","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/842ba1a734c542b2026d8cee.png"},{"id":95836815,"identity":"0034d6ca-ab97-4090-a5fd-8292c8af5878","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"cls","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":55857,"visible":true,"origin":"","legend":"","description":"","filename":"snjnl.cls","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/161641df1edd839da93a56a0.cls"},{"id":95836810,"identity":"7ecbf8ae-e430-46fd-bc33-6d5d4a34d4b7","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"bst","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":64166,"visible":true,"origin":"","legend":"","description":"","filename":"snmathphysnum.bst","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/9993c36672b98abc123356f8.bst"},{"id":96240189,"identity":"7ba45191-ed8d-41c5-9b4f-9be7f6c59499","added_by":"auto","created_at":"2025-11-19 07:08:33","extension":"xml","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":69640,"visible":true,"origin":"","legend":"","description":"","filename":"5f5c9fbd1b5d49369cbaaf01fc5ad29a1structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/badcba2429fe55be6ec5fcb4.xml"},{"id":95836813,"identity":"dc8612dd-d856-437c-8a29-12c937f739bc","added_by":"auto","created_at":"2025-11-13 13:28:32","extension":"html","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":68485,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1/4701684947b5de5c0de5c6c4.html"},{"id":99688607,"identity":"e439cdef-68ee-47a2-aa61-11051cda7e69","added_by":"auto","created_at":"2026-01-07 09:55:46","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":512153,"visible":true,"origin":"","legend":"","description":"","filename":"FeatureFusionUnitsforFinegrainedImageCategorization.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7717226/v1_covered_f11b5edf-e775-43e9-b933-d099c09ff021.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Feature Fusion Units for Fine-grained Image Categorization","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":"Fine-grained image, categorization, transformer, computer vision, feature weight unit","lastPublishedDoi":"10.21203/rs.3.rs-7717226/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7717226/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFine-grained image categorization aims to categorize subclasses by processing detailed features, which is still a critical problem to be solved in computer version due to the small differences between subclasses. The traditional methods are usually to find features by manual annotation, using specific sliding Windows, using different thresholds and other methods. These methods are not only costly, but also ineffective. In computer version, by calculating attention scores between parts of the picture multiple times and weighting them, the transformer greatly improves the accuracy of categorization. In this paper, we propose a feature weight units. Specifically, transformer is used as the backbone to capture image feature(these features are called patches in transformer), and then all patches are weighted by our feature weight unit. The computal result of feature fusion unit represents the importance of the patch should to be forced on. To verify the effectiveness of our method, we conducted experiments on the CUB-200-2011 and stanford-dog datasets.\u003c/p\u003e","manuscriptTitle":"Feature Fusion Units for Fine-grained Image Categorization","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-13 13:28:27","doi":"10.21203/rs.3.rs-7717226/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":"0593ef24-a557-49e5-bb35-0bf112fe3c25","owner":[],"postedDate":"November 13th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-01-07T09:55:25+00:00","versionOfRecord":[],"versionCreatedAt":"2025-11-13 13:28:27","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7717226","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7717226","identity":"rs-7717226","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}