MSConvAttn-YOLO: A Multi-Scale Convolutional Attention Network for High-Accuracy and Real-Time Landslide Detection in Remote Sensing Imagery

MSConvAttn-YOLO: A Multi-Scale Convolutional Attention Network for High-Accuracy and Real-Time Landslide Detection in Remote Sensing Imagery | 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 Method Article MSConvAttn-YOLO: A Multi-Scale Convolutional Attention Network for High-Accuracy and Real-Time Landslide Detection in Remote Sensing Imagery Yan Zhang, Ruifeng Li, Fan Zhang, Zhibing Wang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9392600/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 5 You are reading this latest preprint version Abstract Landslides pose a persistent threat to social stability and regional economic development. Utilizing remote sensing for rapid and precise landslide identification and loss assessment is critical for formulating effective disaster mitigation decisions. Although deep learning-based object detection methods have made significant progress, existing approaches still face challenges in complex remote sensing scenarios, such as insufficient adaptability to multi-scale features and the difficulty in balancing detection accuracy with computational efficiency. These limitations hinder their practical deployment in emergency response. To bridge the gap between algorithmic capability and operational requirements for disaster reduction, this paper proposes a Multi-Scale Convolutional Attention module (MSConvAttn) and integrates it into the YOLOv13 architecture. This module enhances the model's ability to synergistically model both local details and global contextual information of landslides by integrating multi-scale static and dynamic convolutional paths, combined with an adaptive weighted fusion mechanism and a triple-attention structure. Experimental results demonstrate that the proposed method achieves a mean Average Precision at 50% Intersection over Union (mAP50) of 81.1%. Compared to the baseline model, it achieves a 4.8% improvement in mAP50 with an increase of only 0.65 million parameters. Furthermore, compared to the original attention mechanism, the proposed module yields a performance gain of 3.1% with merely a 0.21 million parameter increase. Cross-regional zero-shot evaluation indicates that the model maintains excellent detection stability and real-time inference capability even in unseen geographical environments. This study provides an effective solution for accurate landslide identification in complex scenarios, demonstrating strong theoretical value and broad application prospects in supporting emergency response and disaster mitigation decisions. Landslide Multi-scale Adaptive Weighted Fusion Mechanism Triple Attention Structure Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 07 May, 2026 Editor invited by journal 29 Apr, 2026 Editor assigned by journal 15 Apr, 2026 Submission checks completed at journal 15 Apr, 2026 First submitted to journal 12 Apr, 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-9392600","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Method Article","associatedPublications":[],"authors":[{"id":636080626,"identity":"97a90133-fc97-4ff0-87b6-3f7fde7b6865","order_by":0,"name":"Yan Zhang","email":"","orcid":"","institution":"Guilin University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Yan","middleName":"","lastName":"Zhang","suffix":""},{"id":636080628,"identity":"27f2eda7-bd78-49bc-aa8b-a2a1c0042ce5","order_by":1,"name":"Ruifeng Li","email":"","orcid":"","institution":"Guilin University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Ruifeng","middleName":"","lastName":"Li","suffix":""},{"id":636080633,"identity":"3b7649a2-53d3-4833-a7a2-1823546f9052","order_by":2,"name":"Fan Zhang","email":"","orcid":"","institution":"Guilin University of Technology","correspondingAuthor":false,"prefix":"","firstName":"Fan","middleName":"","lastName":"Zhang","suffix":""},{"id":636080637,"identity":"9ad6bc8c-ce33-40f1-affd-2cc8f33823a5","order_by":3,"name":"Zhibing Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3ElEQVRIiWNgGAWjYLACxgaJBDYg9SChooY0LcwGD84cI1oLQwKQYpN82MJMWLXB8bOHX/7cYZHHJ91+rSKxgY2Bv707Ab+WM3lpFpJnJIrZZM6U3UjcIcMgcebsBrxazA7kmBkYtkkktknkpN1IPMPGYCCRS0DL+TdmBolQLQWJbcxEaLmRY/zgIFhL+jEGorTY33hjxtjYBvSLRA6zRMKZYzwE/SLZn2P88WdbXZ78jPSHH39U1Mjxt/fi1wIEbBIQmscATBJSDgLMHyA0+wNiVI+CUTAKRsEIBABzwEtdOXf4HgAAAABJRU5ErkJggg==","orcid":"","institution":"Guilin University of Technology","correspondingAuthor":true,"prefix":"","firstName":"Zhibing","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2026-04-12 08:08:16","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9392600/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9392600/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109405611,"identity":"940a661b-6386-4d9f-973b-fe7ad79b0485","added_by":"auto","created_at":"2026-05-17 13:19:23","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2627558,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscripts.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9392600/v1_covered_a018f2d9-6e2b-4ce4-be32-50947746e037.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"MSConvAttn-YOLO: A Multi-Scale Convolutional Attention Network for High-Accuracy and Real-Time Landslide Detection in Remote Sensing Imagery","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":"discover-geoscience","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Geoscience](https://www.springer.com/journal/44288)","snPcode":"44288","submissionUrl":"https://submission.nature.com/new-submission/44288","title":"Discover Geoscience","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Landslide, Multi-scale, Adaptive Weighted Fusion Mechanism, Triple Attention Structure","lastPublishedDoi":"10.21203/rs.3.rs-9392600/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9392600/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eLandslides pose a persistent threat to social stability and regional economic development. Utilizing remote sensing for rapid and precise landslide identification and loss assessment is critical for formulating effective disaster mitigation decisions. Although deep learning-based object detection methods have made significant progress, existing approaches still face challenges in complex remote sensing scenarios, such as insufficient adaptability to multi-scale features and the difficulty in balancing detection accuracy with computational efficiency. These limitations hinder their practical deployment in emergency response. To bridge the gap between algorithmic capability and operational requirements for disaster reduction, this paper proposes a Multi-Scale Convolutional Attention module (MSConvAttn) and integrates it into the YOLOv13 architecture. This module enhances the model's ability to synergistically model both local details and global contextual information of landslides by integrating multi-scale static and dynamic convolutional paths, combined with an adaptive weighted fusion mechanism and a triple-attention structure. Experimental results demonstrate that the proposed method achieves a mean Average Precision at 50% Intersection over Union (mAP50) of 81.1%. Compared to the baseline model, it achieves a 4.8% improvement in mAP50 with an increase of only 0.65\u0026nbsp;million parameters. Furthermore, compared to the original attention mechanism, the proposed module yields a performance gain of 3.1% with merely a 0.21\u0026nbsp;million parameter increase. Cross-regional zero-shot evaluation indicates that the model maintains excellent detection stability and real-time inference capability even in unseen geographical environments. This study provides an effective solution for accurate landslide identification in complex scenarios, demonstrating strong theoretical value and broad application prospects in supporting emergency response and disaster mitigation decisions.\u003c/p\u003e","manuscriptTitle":"MSConvAttn-YOLO: A Multi-Scale Convolutional Attention Network for High-Accuracy and Real-Time Landslide Detection in Remote Sensing Imagery","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-15 11:28:11","doi":"10.21203/rs.3.rs-9392600/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-05-07T09:36:54+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-29T09:08:47+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-16T02:25:44+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-04-16T02:24:52+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Geoscience","date":"2026-04-12T07:52:37+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-geoscience","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"Learn more about [Discover Geoscience](https://www.springer.com/journal/44288)","snPcode":"44288","submissionUrl":"https://submission.nature.com/new-submission/44288","title":"Discover Geoscience","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"8cef3faa-3291-4617-8100-fde2c9474fc7","owner":[],"postedDate":"May 15th, 2026","published":true,"recentEditorialEvents":[{"type":"reviewersInvited","content":"40","date":"2026-05-07T09:36:54+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-29T09:08:47+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-15T11:28:11+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-15 11:28:11","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9392600","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9392600","identity":"rs-9392600","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}