SFNet: Visual Attention-Based Slender objects Focus Network for Self-Supervised Monocular Depth Estimation

SFNet: Visual Attention-Based Slender objects Focus Network for Self-Supervised Monocular Depth Estimation | 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 SFNet: Visual Attention-Based Slender objects Focus Network for Self-Supervised Monocular Depth Estimation YimingShao, DakeZhou, XinYang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4415776/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 Self-supervised learning has recently gained significant attention in monocular depth estimation (MDE) community, as it uses image sequences rather than ground-truth depth to provide training signal. In traffic environments, slender objects such as signboards and traffic signals carry important traffic indication information, but they occupy a small area in the image. Current MDE methods are difficult to accurately predict their complete shapes. To address this problem, we propose a Slender objects Focus Network (SFNet) for self-supervised MDE. The SFNet uses a multiscale encoder-decoder architecture. In the encoder branch, we devise a novel Global-Local Collaboration Module (GLCM), which consists of an Edge-Texture Perception Module (ETPM) and a Transformer Block. GLCM effectively captures local detailed and global scene structure information, enabling the encoder to better understand and predict the shape of slender objects. In the decoder branch, to improve the fine-grained prediction of image texture details, we propose the Multiscale Feature Aggregation Module (MFAM), which adaptively integrates the dense features at different scales and depths. Moreover, we design the Pyramid Feature Integration Unit (PFIU) to further leverage the rich semantic information embedded in the encoder’s high-level features. Extensive experiments demonstrate the excellent performance of our method. SFNet achieves state-of-the-art performance on the KITTI dataset, while qualitative results show its advantages in predicting slender objects. In addition, results on the Make3D dataset demonstrate the good generalization ability of SFNet. Our codes and models are available at https://github.com/YIMings139/SFNet . Artificial Intelligence and Machine Learning Monocular Depth estimation Slender objects prediction Self-supervised learning Attention mechanism Full Text Additional Declarations The authors declare no competing interests. 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-4415776","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":302104180,"identity":"763f0e8b-e50c-4399-a867-0ccdd6015a07","order_by":0,"name":"YimingShao","email":"","orcid":"https://orcid.org/0009-0005-0465-1198","institution":"Nanjing University of Aeronautics and Astronautics","correspondingAuthor":false,"prefix":"","firstName":"","middleName":"","lastName":"YimingShao","suffix":""},{"id":302104181,"identity":"e21a8e48-555a-4289-9068-f635caa6444b","order_by":1,"name":"DakeZhou","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyUlEQVRIiWNgGAWjYDCCMwxsQNIGymMjXksa6VoOk6CF78zxZw8+tp3PMzh++AHDh7LDDPyzG/BrkTzbkG44s+12scGZNAPGGecOM0jcOYBfi8F5hmPSvG23EzccSDBg5m07zGAgkUBIC2MbUMu5xA3nn39g/kuUlrPNbEAtBxI33MgxYGYkRovkmWNskjPOJSfOvPGm4GDPuXQeiRsEtPCdSX8m8aHMLrHvfPrGBz/KrOX4ZxDQggIOADEPCepHwSgYBaNgFOACAAlgSGvVk2ghAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0003-2156-8319","institution":"Nanjing University of Aeronautics and Astronautics","correspondingAuthor":true,"prefix":"","firstName":"","middleName":"","lastName":"DakeZhou","suffix":""},{"id":302104182,"identity":"1857cc83-e40e-4c93-96c1-bd514ce9fee4","order_by":2,"name":"XinYang","email":"","orcid":"","institution":"Nanjing University of Aeronautics and Astronautics","correspondingAuthor":false,"prefix":"","firstName":"","middleName":"","lastName":"XinYang","suffix":""}],"badges":[],"createdAt":"2024-05-14 01:44:18","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":false,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-4415776/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4415776/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":56548804,"identity":"afaed6a1-485c-475f-8379-34426cabc177","added_by":"auto","created_at":"2024-05-15 15:43:06","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1087096,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript5.1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4415776/v1_covered_e3307765-6978-4331-a9ec-57436e02abd8.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eSFNet: Visual Attention-Based Slender objects Focus Network for Self-Supervised Monocular Depth Estimation\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Nanjing University of Aeronautics and Astronautics","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":"Monocular Depth estimation, Slender objects prediction, Self-supervised learning, Attention mechanism","lastPublishedDoi":"10.21203/rs.3.rs-4415776/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4415776/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eSelf-supervised learning has recently gained significant attention in monocular depth estimation (MDE) community, as it uses image sequences rather than ground-truth depth to provide training signal. In traffic environments, slender objects such as signboards and traffic signals carry important traffic indication information, but they occupy a small area in the image. Current MDE methods are difficult to accurately predict their complete shapes. To address this problem, we propose a Slender objects Focus Network (SFNet) for self-supervised MDE. The SFNet uses a multiscale encoder-decoder architecture. In the encoder branch, we devise a novel Global-Local Collaboration Module (GLCM), which consists of an Edge-Texture Perception Module (ETPM) and a Transformer Block. GLCM effectively captures local detailed and global scene structure information, enabling the encoder to better understand and predict the shape of slender objects. In the decoder branch, to improve the fine-grained prediction of image texture details, we propose the Multiscale Feature Aggregation Module (MFAM), which adaptively integrates the dense features at different scales and depths. Moreover, we design the Pyramid Feature Integration Unit (PFIU) to further leverage the rich semantic information embedded in the encoder\u0026rsquo;s high-level features. Extensive experiments demonstrate the excellent performance of our method. SFNet achieves state-of-the-art performance on the KITTI dataset, while qualitative results show its advantages in predicting slender objects. In addition, results on the Make3D dataset demonstrate the good generalization ability of SFNet. Our codes and models are available at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/YIMings139/SFNet\u003c/span\u003e\u003cspan address=\"https://github.com/YIMings139/SFNet\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e.\u003c/p\u003e","manuscriptTitle":"SFNet: Visual Attention-Based Slender objects Focus Network for Self-Supervised Monocular Depth Estimation","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-05-15 15:42:45","doi":"10.21203/rs.3.rs-4415776/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":"34d61fa8-c73d-4dd3-9d68-aa9238601125","owner":[],"postedDate":"May 15th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":31874553,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2024-05-15T15:42:45+00:00","versionOfRecord":[],"versionCreatedAt":"2024-05-15 15:42:45","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4415776","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4415776","identity":"rs-4415776","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}