Weakly Supervised Semantic Segmentation Based on Subspace- Decoupled Representations and Cross-Layer CAM Structural Alignment

Weakly Supervised Semantic Segmentation Based on Subspace- Decoupled Representations and Cross-Layer CAM Structural Alignment | 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 Weakly Supervised Semantic Segmentation Based on Subspace- Decoupled Representations and Cross-Layer CAM Structural Alignment Kaiyang Liao, Junwen Pang, Yuanlin Zheng, Yunfei Tan This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9178163/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Weakly Supervised Semantic Segmentation (WSSS) aims to learn pixel-level semantic predictions using only image-level annotations. However, due to the absence of precise spatial supervision, the generated Class Activation Maps (CAMs) often highlight only the most discriminative regions of objects, resulting in incomplete object coverage and unstable cross-layer semantic responses. To address these challenges, we propose a token-level contrastive learning based framework for WSSS, which improves CAM localization quality by enhancing feature representation and enforcing cross-layer structural consistency. Specifically, we first introduce a multi-subspace token-level contrastive module, which decouples feature representations through a shared semantic backbone and multiple projection subspaces, thereby increasing the diversity and discriminability of the embedding space. Furthermore, we propose a cross-layer CAM structural alignment module that jointly constrains both the response intensity and spatial structural relationships of CAMs across different Transformer layers, leading to more stable semantic localization and improved spatial consistency of object regions. Extensive experiments on the PASCAL VOC 2012 and MS COCO 2014 benchmarks demonstrate that the proposed method consistently improves segmentation performance under an end-to-end training framework. In particular, it achieves 71.8% (val) and 72.3% (test) mIoU on VOC 2012, and 42.6% mIoU on the COCO 2014 validation set. Further ablation studies validate the effectiveness of each component. Overall, our method significantly enhances the completeness and structural stability of CAMs, providing an effective solution for representation learning and structural modeling in WSSS. Class Activation Maps (CAM) Token-level Contrastive Learning Multi-subspace Representation Learning Cross-layer Structural Alignment Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Editorial decision: Revision requested 10 Apr, 2026 Reviews received at journal 10 Apr, 2026 Reviewers agreed at journal 29 Mar, 2026 Reviews received at journal 28 Mar, 2026 Reviewers agreed at journal 28 Mar, 2026 Reviewers agreed at journal 28 Mar, 2026 Reviewers invited by journal 28 Mar, 2026 Editor assigned by journal 21 Mar, 2026 Submission checks completed at journal 21 Mar, 2026 First submitted to journal 20 Mar, 2026 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-9178163","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":615046220,"identity":"f15695d6-ace2-44ba-8ec7-a1cb9358f6de","order_by":0,"name":"Kaiyang Liao","email":"","orcid":"","institution":"Xi’an University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Kaiyang","middleName":"","lastName":"Liao","suffix":""},{"id":615046222,"identity":"6cc839ff-77ee-4737-ad8a-7c9f2176410f","order_by":1,"name":"Junwen Pang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA0ElEQVRIiWNgGAWjYBACNvaG5AcSP2x4+PmbDxCnhY/nwDMDy540GckZxxKI0yInkfhAooLtsI1BQ44BkQ6TSE4wuMFzmMeA4czHG28Y7OR0Gwhp4XmW8HCGRTqPOXPvZss5DMnGZgcIaWHPSTCW4LHmsWw4u02ah+FA4jaCWhjyP0j/YWPmMTiQ84xILRwJCRISbM4gLWxEauE5kGYg2ZPGAwxkY8s5BkT4Rb4dEpX2wKh8eONNhZ0cQS0oQIKHyKhB1kKqjlEwCkbBKBgRAAAnrEBwpc0wpQAAAABJRU5ErkJggg==","orcid":"","institution":"Xi’an University of Technology","correspondingAuthor":true,"prefix":"","firstName":"Junwen","middleName":"","lastName":"Pang","suffix":""},{"id":615046227,"identity":"bc7d23da-7b55-4461-bfe2-4eca0a7af1a1","order_by":2,"name":"Yuanlin Zheng","email":"","orcid":"","institution":"Xi’an University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Yuanlin","middleName":"","lastName":"Zheng","suffix":""},{"id":615046229,"identity":"b2a5d01f-ac1c-42ab-b03c-7bff578dd116","order_by":3,"name":"Yunfei Tan","email":"","orcid":"","institution":"Xi’an University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Yunfei","middleName":"","lastName":"Tan","suffix":""}],"badges":[],"createdAt":"2026-03-20 10:56:10","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9178163/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9178163/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":106093517,"identity":"9ba02966-7e86-4c23-9674-896500ab1f37","added_by":"auto","created_at":"2026-04-03 11:37:44","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":738548,"visible":true,"origin":"","legend":"","description":"","filename":"WeaklySupervisedSemanticSegmentationBasedonSubspaceDecoupledRepresentationsandCrossLayerCAMStructuralAlignment.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9178163/v1_covered_43a3e8cd-ce06-4d87-b341-31188e3a97eb.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Weakly Supervised Semantic Segmentation Based on Subspace- Decoupled Representations and Cross-Layer CAM Structural Alignment","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"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":"signal-image-and-video-processing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sivp","sideBox":"Learn more about [Signal, Image and Video Processing](http://link.springer.com/journal/11760)","snPcode":"11760","submissionUrl":"https://submission.nature.com/new-submission/11760/3","title":"Signal, Image and Video Processing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"Class Activation Maps (CAM), Token-level Contrastive Learning, Multi-subspace Representation Learning, Cross-layer Structural Alignment","lastPublishedDoi":"10.21203/rs.3.rs-9178163/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9178163/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eWeakly Supervised Semantic Segmentation (WSSS) aims to learn pixel-level semantic predictions using only image-level annotations. However, due to the absence of precise spatial supervision, the generated Class Activation Maps (CAMs) often highlight only the most discriminative regions of objects, resulting in incomplete object coverage and unstable cross-layer semantic responses. To address these challenges, we propose a token-level contrastive learning based framework for WSSS, which improves CAM localization quality by enhancing feature representation and enforcing cross-layer structural consistency. Specifically, we first introduce a multi-subspace token-level contrastive module, which decouples feature representations through a shared semantic backbone and multiple projection subspaces, thereby increasing the diversity and discriminability of the embedding space. Furthermore, we propose a cross-layer CAM structural alignment module that jointly constrains both the response intensity and spatial structural relationships of CAMs across different Transformer layers, leading to more stable semantic localization and improved spatial consistency of object regions. Extensive experiments on the PASCAL VOC 2012 and MS COCO 2014 benchmarks demonstrate that the proposed method consistently improves segmentation performance under an end-to-end training framework. In particular, it achieves 71.8% (val) and 72.3% (test) mIoU on VOC 2012, and 42.6% mIoU on the COCO 2014 validation set. Further ablation studies validate the effectiveness of each component. Overall, our method significantly enhances the completeness and structural stability of CAMs, providing an effective solution for representation learning and structural modeling in WSSS.\u003c/p\u003e","manuscriptTitle":"Weakly Supervised Semantic Segmentation Based on Subspace- Decoupled Representations and Cross-Layer CAM Structural Alignment","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-02 05:14:06","doi":"10.21203/rs.3.rs-9178163/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-04-11T02:35:33+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-10T08:47:24+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"155325278869717002148361287010290425090","date":"2026-03-29T14:10:29+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-03-29T02:12:14+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"325323842782389323815813920689347058221","date":"2026-03-29T01:59:03+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"167617492440008000502994793973796915512","date":"2026-03-29T01:57:56+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-29T01:54:23+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-22T02:01:25+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-22T02:01:15+00:00","index":"","fulltext":""},{"type":"submitted","content":"Signal, Image and Video Processing","date":"2026-03-20T10:40:27+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"signal-image-and-video-processing","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"sivp","sideBox":"Learn more about [Signal, Image and Video Processing](http://link.springer.com/journal/11760)","snPcode":"11760","submissionUrl":"https://submission.nature.com/new-submission/11760/3","title":"Signal, Image and Video Processing","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"em","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"66ef839b-8064-48fa-829d-363d773284fe","owner":[],"postedDate":"April 2nd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-11T15:53:16+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-02 05:14:06","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9178163","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9178163","identity":"rs-9178163","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}