SPS-UNet:A Super-pixel Sampling UNet for Extracting Buildings from High-resolution Satellite Images

SPS-UNet:A Super-pixel Sampling UNet for Extracting Buildings from High-resolution Satellite Images | 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 SPS-UNet:A Super-pixel Sampling UNet for Extracting Buildings from High-resolution Satellite Images Qiuquan Zhao, Jianyuan Li This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3824243/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 21 Jan, 2025 Read the published version in The Visual Computer → Version 1 posted 9 You are reading this latest preprint version Abstract Fully convolutional networks in general perform well in automatically extracting buildings from high-resolution satellite images. However, we find experimentally that a variety of such methods have low robustness when extracting noisy buildings, which leads to intuitively unreasonable results such as broken segmentations or inaccurate boundaries. In this paper, we propose a super-pixel sampling UNet (SPS-UNet) for tackling this problem, which acts the MobileNetV2 as the backbone and replaces the traditional down-sampling operators with a new learnable super-pixel sampling module (SPSM). We also introduce an additional entropy loss item in the training phase to enhance the certainty of prediction results. Experimental results over two public datasets show that: (1) SPS-UNet outperforms the competing methods in terms of both segmentation accuracies and the robustness to noisy buildings; (2) by plugging SPSM in a variety of existing fully convolutional networks and replacing the traditional rule-based down-sampling operators, the semantic segmentation results can be consistently improved. Code can be downloaded from https://github.com/1193639809ZD/SPSNet fully convolutional networks the automatic extraction of buildings high resolution images super-pixel sampling Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 21 Jan, 2025 Read the published version in The Visual Computer → Version 1 posted Editorial decision: Revision requested 25 Jun, 2024 Reviews received at journal 17 Apr, 2024 Reviewers agreed at journal 01 Apr, 2024 Reviews received at journal 04 Feb, 2024 Reviewers agreed at journal 16 Jan, 2024 Reviewers invited by journal 15 Jan, 2024 Editor assigned by journal 01 Jan, 2024 Submission checks completed at journal 01 Jan, 2024 First submitted to journal 30 Dec, 2023 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-3824243","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":264786101,"identity":"28f922c5-a325-48ca-a528-d34cf61c00db","order_by":0,"name":"Qiuquan Zhao","email":"","orcid":"","institution":"Zhejiang University","correspondingAuthor":false,"prefix":"","firstName":"Qiuquan","middleName":"","lastName":"Zhao","suffix":""},{"id":264786103,"identity":"7411fefe-41cc-41ef-a18c-3e81195a3fe9","order_by":1,"name":"Jianyuan Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAArklEQVRIiWNgGAWjYBACA2aGBIaPDRCOBNFaGGeSpgWImXlJ08LO8EzadkedvMEB5oO3eRjs8ohxWJp07hk2ww0H2JKteRiSi4nU0saTYHCAx0yah+FAYgNRWizbJIBa+L+RoIWxzQBkCxvRWpIte9sSDGceZjO2nGOQTFiLff+ZxBs/2+rk+Y43P7zxpsKOsBYGBp4USHQwgy0lrB4I2A9/IErdKBgFo2AUjFwAADbxMnjRhQB+AAAAAElFTkSuQmCC","orcid":"","institution":"Hangzhou City University","correspondingAuthor":true,"prefix":"","firstName":"Jianyuan","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2023-12-30 14:44:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3824243/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3824243/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s00371-024-03787-4","type":"published","date":"2025-01-21T15:58:09+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":74859128,"identity":"63d77d80-0892-4505-a4db-d538fb9387ed","added_by":"auto","created_at":"2025-01-27 16:13:33","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1020630,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3824243/v1_covered_0d4e5f6f-e25a-4622-aeb0-a67613a6286f.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"SPS-UNet:A Super-pixel Sampling UNet for Extracting Buildings from High-resolution Satellite Images","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"the-visual-computer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"tvcj","sideBox":"Learn more about [The Visual Computer](http://link.springer.com/journal/371)","snPcode":"371","submissionUrl":"https://submission.nature.com/new-submission/371/3","title":"The Visual Computer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false},"keywords":"fully convolutional networks, the automatic extraction of buildings, high resolution images, super-pixel sampling","lastPublishedDoi":"10.21203/rs.3.rs-3824243/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3824243/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFully convolutional networks in general perform well in automatically extracting buildings from high-resolution satellite images. However, we find experimentally that a variety of such methods have low robustness when extracting noisy buildings, which leads to intuitively unreasonable results such as broken segmentations or inaccurate boundaries. In this paper, we propose a super-pixel sampling UNet (SPS-UNet) for tackling this problem, which acts the MobileNetV2 as the backbone and replaces the traditional down-sampling operators with a new learnable super-pixel sampling module (SPSM). We also introduce an additional entropy loss item in the training phase to enhance the certainty of prediction results. Experimental results over two public datasets show that: (1) SPS-UNet outperforms the competing methods in terms of both segmentation accuracies and the robustness to noisy buildings; (2) by plugging SPSM in a variety of existing fully convolutional networks and replacing the traditional rule-based down-sampling operators, the semantic segmentation results can be consistently improved. Code can be downloaded from \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/1193639809ZD/SPSNet\u003c/span\u003e\u003cspan address=\"https://github.com/1193639809ZD/SPSNet\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e","manuscriptTitle":"SPS-UNet:A Super-pixel Sampling UNet for Extracting Buildings from High-resolution Satellite Images","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-01-03 10:30:00","doi":"10.21203/rs.3.rs-3824243/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2024-06-25T04:14:26+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-04-17T10:42:55+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"7bf6332b-c999-4496-a424-abc8906da977","date":"2024-04-01T04:42:32+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-02-04T12:50:06+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"bc10a279-b84d-4894-8c2c-9ae396b6fd23","date":"2024-01-16T06:53:55+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-01-15T15:13:24+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-01-02T02:59:53+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-01-01T16:41:56+00:00","index":"","fulltext":""},{"type":"submitted","content":"The Visual Computer","date":"2023-12-30T14:39:16+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"the-visual-computer","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"tvcj","sideBox":"Learn more about [The Visual Computer](http://link.springer.com/journal/371)","snPcode":"371","submissionUrl":"https://submission.nature.com/new-submission/371/3","title":"The Visual Computer","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Springer Hybrid","inReviewEnabled":true,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"4179f2b7-2d23-42ea-9e34-97914c60d484","owner":[],"postedDate":"January 3rd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-01-27T16:10:25+00:00","versionOfRecord":{"articleIdentity":"rs-3824243","link":"https://doi.org/10.1007/s00371-024-03787-4","journal":{"identity":"the-visual-computer","isVorOnly":false,"title":"The Visual Computer"},"publishedOn":"2025-01-21 15:58:09","publishedOnDateReadable":"January 21st, 2025"},"versionCreatedAt":"2024-01-03 10:30:00","video":"","vorDoi":"10.1007/s00371-024-03787-4","vorDoiUrl":"https://doi.org/10.1007/s00371-024-03787-4","workflowStages":[]},"version":"v1","identity":"rs-3824243","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3824243","identity":"rs-3824243","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}