A Lightweight Adapter for Efficient Fine-Tuning in Computer Vision

A Lightweight Adapter for Efficient Fine-Tuning in Computer Vision | 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 A Lightweight Adapter for Efficient Fine-Tuning in Computer Vision Kim Huong Tran, Ba Ty Dang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8843187/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 Pretrained vision backbones are becoming increasingly large, making full fine-tuning expensive in both training time and storage, especially when adapting a single backbone to many tasks or data domains. In the setting of Parameter-Efficient Fine-Tuning (PEFT), we propose DT1D-Adapter , a lightweight adapter that can be plugged into both convolutional networks (ConvNets) and Transformer-based models by operating on spatial features of size \((H\times W)\) . DT1D-Adapter performs axial filtering via depthwise 1D convolution along the height and/or width, where dilation enlarges the receptive field with minimal parameter increase. To control the parameter budget, the filters are parameterized by symmetric, group-shared coefficients, optionally complemented with lightweight channel mixing using grouped \((1\times 1)\) projections and a small-initialized scalar residual gate to stabilize optimization under limited trainable parameters. Experiments on multiple image classification datasets show that DT1D-Adapter provides a strong accuracy--parameter trade-off, remaining competitive with common PEFT baselines such as SSF, BitFit, and VPT, and demonstrating notable efficiency compared with convolution-based adapters (e.g., Conv-Adapter) and residual-branch variants (e.g., Residual Adapters).We further report a simple video-stream inference benchmark, indicating that DT1D-Adapter remains compatible with real-time deployment on common GPU platforms. parameter-efficient fine-tuning adapter Resnet Vision Transformers 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-8843187","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":591373878,"identity":"299d229d-46e7-4c39-9d37-11116e9ab48a","order_by":0,"name":"Kim Huong Tran","email":"data:image/png;base64,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","orcid":"","institution":"Hanoi University of Science and Technology (HUST)","correspondingAuthor":true,"prefix":"","firstName":"Kim","middleName":"Huong","lastName":"Tran","suffix":""},{"id":591373879,"identity":"55e19f49-2d06-47e9-8b56-2876226f0759","order_by":1,"name":"Ba Ty Dang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Ba","middleName":"Ty","lastName":"Dang","suffix":""}],"badges":[],"createdAt":"2026-02-10 16:24:04","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8843187/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8843187/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102966401,"identity":"1bbe84d6-40b8-46f3-8872-c2b5b42fb092","added_by":"auto","created_at":"2026-02-19 04:36:23","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":328240,"visible":true,"origin":"","legend":"","description":"","filename":"AlightweightAdapterTranHuongauthor.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8843187/v1_covered_a9f43e66-f3d3-4972-9a92-991c50492221.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Lightweight Adapter for Efficient Fine-Tuning in Computer Vision","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"parameter-efficient fine-tuning, adapter, Resnet, Vision Transformers","lastPublishedDoi":"10.21203/rs.3.rs-8843187/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8843187/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003ePretrained vision backbones are becoming increasingly large, making full fine-tuning expensive in both training time and storage, especially when adapting a single backbone to many tasks or data domains. In the setting of Parameter-Efficient Fine-Tuning (PEFT), we propose \u003cb\u003eDT1D-Adapter\u003c/b\u003e, a lightweight adapter that can be plugged into both convolutional networks (ConvNets) and Transformer-based models by operating on spatial features of size \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((H\\times W)\\)\u003c/span\u003e\u003c/span\u003e. \u003cb\u003eDT1D-Adapter\u003c/b\u003e performs axial filtering via depthwise 1D convolution along the height and/or width, where dilation enlarges the receptive field with minimal parameter increase. To control the parameter budget, the filters are parameterized by symmetric, group-shared coefficients, optionally complemented with lightweight channel mixing using grouped \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\((1\\times 1)\\)\u003c/span\u003e\u003c/span\u003e projections and a small-initialized scalar residual gate to stabilize optimization under limited trainable parameters. Experiments on multiple image classification datasets show that \u003cb\u003eDT1D-Adapter\u003c/b\u003e provides a strong accuracy--parameter trade-off, remaining competitive with common PEFT baselines such as SSF, BitFit, and VPT, and demonstrating notable efficiency compared with convolution-based adapters (e.g., Conv-Adapter) and residual-branch variants (e.g., Residual Adapters).We further report a simple video-stream inference benchmark, indicating that DT1D-Adapter remains compatible with real-time deployment on common GPU platforms.\u003c/p\u003e","manuscriptTitle":"A Lightweight Adapter for Efficient Fine-Tuning in Computer Vision","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-17 05:32:32","doi":"10.21203/rs.3.rs-8843187/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":"9c8198f3-b380-4c18-a3a8-1f9655001047","owner":[],"postedDate":"February 17th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-19T04:14:31+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-17 05:32:32","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8843187","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8843187","identity":"rs-8843187","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}