RS-YOLO: A YOLO-Based Method for Small Object Detection in Remote Sensing

RS-YOLO: A YOLO-Based Method for Small Object Detection in Remote Sensing | 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 RS-YOLO: A YOLO-Based Method for Small Object Detection in Remote Sensing Dun Ao, Wentao Zhao, Yu Zhang, Fei Lei This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4416811/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 Since the objects are usually small and densely distributed, the current mainstream algorithms have poor performance in the task of object detection in remote sensing, and are prone to produce missed detection and false detection. Moreover, remote sensing images(RSI) contain complex background noise, which further exacerbates this phenomenon. To settle these problems, this paper proposes a novel small object detection model in remote sensing, named RS-YOLO. In order to suppress the background noise and enhance the attention to small objects, by integrating Biformer, a dynamic attention mechanism, we construct an improved backbone network. Then, an efficient neck structure based on Gather-and-Distribute(GD) mechanism is introduced to enhance the feature fusion capability of the model. In addition, in order to fully utilize shallow features, we simplify the deep layer structure to reduce its competitiveness in the label assignment process. The strategy simultaneously improves the accuracy and speed of the model. Moreover, we combine Inner-IoU and CIoU loss functions as the new bounding box regression(BBR) loss function, which enhances the localization performance and robustness of the model, and speeds up the training time. Finally, on AI-TOD and DIOR datasets, the mean Average Precision (mAP) of our proposed RS-YOLO reaches 56.3% and 74.2% respectively, which exceeds the baseline model and other mainstream algorithms. small object detection remote sensing Biformer Gather-and-Distribute mechanism Inner-IoU 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-4416811","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":305791911,"identity":"06084e84-b164-47cb-b86d-6abae2f6b5d0","order_by":0,"name":"Dun Ao","email":"","orcid":"","institution":"Beijing University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Dun","middleName":"","lastName":"Ao","suffix":""},{"id":305791912,"identity":"fae14f3f-e646-4808-a95d-97388f426608","order_by":1,"name":"Wentao Zhao","email":"","orcid":"","institution":"Beijing University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Wentao","middleName":"","lastName":"Zhao","suffix":""},{"id":305791914,"identity":"c2858fe6-1007-4d35-a9ae-0d47cadcfbbf","order_by":2,"name":"Yu Zhang","email":"","orcid":"","institution":"Beijing University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Zhang","suffix":""},{"id":305791915,"identity":"8b5add6c-69f1-40ab-be37-314e5e114eb0","order_by":3,"name":"Fei Lei","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0ElEQVRIie3NsQrCMBDG8S8UdIl2zaSvcM6Kz9Ii6KaC4JyjoEsfIODLVDq41VXQweILdHRQMCK4JqNg/stxcD8OCIV+sC6EbkCjROj32vIgLQg2oKklkTeBMECZAN6knbFulsd5ZBYFmnWJeKcdRO6ZDZ1Xwkzsv6qEuhQOotLtTdI5ZUuizqYEqcRB+jVnD6o+5OlFlOAMVHyI8CIyZc5pshL5jfZ5NZPq5CBx+1Dr+2M8H2zT+npfD3uxcZBvAw0UdkrPe1vf/zQUCoX+rRfMfkOK5627iAAAAABJRU5ErkJggg==","orcid":"","institution":"Beijing University of Technology","correspondingAuthor":true,"prefix":"","firstName":"Fei","middleName":"","lastName":"Lei","suffix":""}],"badges":[],"createdAt":"2024-05-14 06:12:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4416811/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4416811/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":57312034,"identity":"5611f35e-fa9d-4a1c-9b8f-895d2a5e9573","added_by":"auto","created_at":"2024-05-29 03:08:49","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2618186,"visible":true,"origin":"","legend":"","description":"","filename":"RSYOLOAYOLOBasedMethodforSmallObjectDetectioninRemoteSensing.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4416811/v1_covered_5aaf4e47-e044-40ef-93d7-fba775e02859.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"RS-YOLO: A YOLO-Based Method for Small Object Detection in Remote Sensing","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":"small object detection, remote sensing, Biformer, Gather-and-Distribute mechanism, Inner-IoU","lastPublishedDoi":"10.21203/rs.3.rs-4416811/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4416811/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Since the objects are usually small and densely distributed, the current mainstream algorithms have poor performance in the task of object detection in remote sensing, and are prone to produce missed detection and false detection. Moreover, remote sensing images(RSI) contain complex background noise, which further exacerbates this phenomenon. To settle these problems, this paper proposes a novel small object detection model in remote sensing, named RS-YOLO. In order to suppress the background noise and enhance the attention to small objects, by integrating Biformer, a dynamic attention mechanism, we construct an improved backbone network. Then, an efficient neck structure based on Gather-and-Distribute(GD) mechanism is introduced to enhance the feature fusion capability of the model. In addition, in order to fully utilize shallow features, we simplify the deep layer structure to reduce its competitiveness in the label assignment process. The strategy simultaneously improves the accuracy and speed of the model. Moreover, we combine Inner-IoU and CIoU loss functions as the new bounding box regression(BBR) loss function, which enhances the localization performance and robustness of the model, and speeds up the training time. Finally, on AI-TOD and DIOR datasets, the mean Average Precision (mAP) of our proposed RS-YOLO reaches 56.3\\% and 74.2\\% respectively, which exceeds the baseline model and other mainstream algorithms.","manuscriptTitle":"RS-YOLO: A YOLO-Based Method for Small Object Detection in Remote Sensing","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-05-24 11:39:38","doi":"10.21203/rs.3.rs-4416811/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":"709a1cc9-70f6-4f10-98d1-9167a898126e","owner":[],"postedDate":"May 24th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-05-29T03:08:26+00:00","versionOfRecord":[],"versionCreatedAt":"2024-05-24 11:39:38","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4416811","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4416811","identity":"rs-4416811","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}