3D Gaussian-Driven SAM2 Repair Method]{3D Gaussian-Driven SAM2 Multi-View Fusion Detection and Triple-Constrained Symmetry Plane Generation Repair Method

3D Gaussian-Driven SAM2 Repair Method]{3D Gaussian-Driven SAM2 Multi-View Fusion Detection and Triple-Constrained Symmetry Plane Generation Repair Method | 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 3D Gaussian-Driven SAM2 Repair Method]{3D Gaussian-Driven SAM2 Multi-View Fusion Detection and Triple-Constrained Symmetry Plane Generation Repair Method Jiuyi Zhang, Jiaqi Ji, Sijia Feng, Huiying Ru This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8451382/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 Advancements in digital conservation have established precise damage localization and automated repair template extraction as critical needs in 3D artifact restoration. Conventional approaches, often limited by heavy manual intervention, high missed-detection rates from single-view analysis, and underutilization of symmetrical features, struggle to meet the demands of high-fidelity restoration. To address these challenges, this paper presents a novel method for 3D artifact damage detection and template extraction, which integrates Segment Anything Model 2 (SAM2) semantic segmentation with geometric symmetry constraints. Our framework begins with high-accuracy 3D reconstruction of the damaged artifact using 3D Gaussian splatting to generate a comprehensive, multi-view image set. SAM2 is then employed to produce damage masks across these views, which are projected into 3D space and integrated using camera parameters. Subsequently, a triple-verification strategy---combining PCA principal axis analysis, minimum symmetry error search, and voxel symmetry evaluation---is proposed to robustly identify the artifact's symmetry plane. This enables the extraction of a structural template from the intact side, which guides the repair of the damaged region. Experimental results demonstrate the efficacy of our method, achieving an average damage detection accuracy of 92.7\%, a symmetry plane detection error below 0.8 mm, and an 89.3\% fit between the extracted template and the damaged area, significantly outperforming traditional single-strategy detection and template extraction techniques. Requiring no large-scale training dataset and reducing reliance on manual input, this approach offers a practical, automated, and high-precision solution for 3D cultural heritage restoration. Computer Architecture and Engineering 3D artifact restoration 3D Gaussian splatting damage detection geometric symmetry verification SAM2 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-8451382","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":565610978,"identity":"9f1ea3ec-b50b-40a0-a310-7f41aa452fe3","order_by":0,"name":"Jiuyi Zhang","email":"","orcid":"","institution":"College of Information Engineering, Hebei University of Architecture","correspondingAuthor":false,"prefix":"","firstName":"Jiuyi","middleName":"","lastName":"Zhang","suffix":""},{"id":565611160,"identity":"871d36c0-dd2a-44dd-b63d-74ba244e262b","order_by":1,"name":"Jiaqi Ji","email":"","orcid":"","institution":"Hebei MinZu Normal University \u0026 The Technology Innovation Center of Cultural Tourism Big Data of Hebei Province","correspondingAuthor":false,"prefix":"","firstName":"Jiaqi","middleName":"","lastName":"Ji","suffix":""},{"id":565611161,"identity":"1c2e49d8-ce85-466b-9146-63dcddcff8e8","order_by":2,"name":"Sijia Feng","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3UlEQVRIie2PoQrCUBSGLww2w5HVE8S9wpGhGHyYycClgUkMCyYNHvFVBj6Apptmd6y4btBmEPFuzbLdKHi/duD/+P8jhMHwi4Cwyvsy6Tur6qKxlmL7mEkfTrWCWorA7tqacq0IDcXjjAhtO2KHh+lzjsLdbINGhc5M1zn0YgY5y1kNw+ycNisuqBa04yOG8gJKIYybFW+vYkBWBDhd5y8dRahhCIEVgGoptFookwsfT3JQ/VL0lN76i8fhoXy8Ew8cHuW316TvbnYtw0SHvk5oiVc4V42QwWAw/DUfkyxCl0PyipMAAAAASUVORK5CYII=","orcid":"","institution":"College of Innovation and Practice, Liaoning Technical University","correspondingAuthor":true,"prefix":"","firstName":"Sijia","middleName":"","lastName":"Feng","suffix":""},{"id":565611162,"identity":"11a9b82e-f743-49f2-86a5-6567f5348cb4","order_by":3,"name":"Huiying Ru","email":"","orcid":"","institution":"School of Science, Hebei University of Architecture","correspondingAuthor":false,"prefix":"","firstName":"Huiying","middleName":"","lastName":"Ru","suffix":""}],"badges":[],"createdAt":"2025-12-26 01:44:21","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-8451382/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8451382/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":99790715,"identity":"4146c57b-aefd-4a61-91c7-70029cb010c9","added_by":"auto","created_at":"2026-01-08 12:58:37","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":562993,"visible":true,"origin":"","legend":"","description":"","filename":"3DGaussianDrivenSAM2RepairMethod.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8451382/v1_covered_b0a59f7e-82a6-48bb-afea-e2e1dff45654.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e3D Gaussian-Driven SAM2 Repair Method]{3D Gaussian-Driven SAM2 Multi-View Fusion Detection and Triple-Constrained Symmetry Plane Generation Repair Method\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Hebei University of Architecture","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":"3D artifact restoration, 3D Gaussian splatting, damage detection, geometric symmetry verification, SAM2","lastPublishedDoi":"10.21203/rs.3.rs-8451382/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8451382/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAdvancements in digital conservation have established precise damage localization and automated repair template extraction as critical needs in 3D artifact restoration. Conventional approaches, often limited by heavy manual intervention, high missed-detection rates from single-view analysis, and underutilization of symmetrical features, struggle to meet the demands of high-fidelity restoration. To address these challenges, this paper presents a novel method for 3D artifact damage detection and template extraction, which integrates Segment Anything Model 2 (SAM2) semantic segmentation with geometric symmetry constraints. Our framework begins with high-accuracy 3D reconstruction of the damaged artifact using 3D Gaussian splatting to generate a comprehensive, multi-view image set. SAM2 is then employed to produce damage masks across these views, which are projected into 3D space and integrated using camera parameters. Subsequently, a triple-verification strategy---combining PCA principal axis analysis, minimum symmetry error search, and voxel symmetry evaluation---is proposed to robustly identify the artifact's symmetry plane. This enables the extraction of a structural template from the intact side, which guides the repair of the damaged region. Experimental results demonstrate the efficacy of our method, achieving an average damage detection accuracy of 92.7\\%, a symmetry plane detection error below 0.8 mm, and an 89.3\\% fit between the extracted template and the damaged area, significantly outperforming traditional single-strategy detection and template extraction techniques. Requiring no large-scale training dataset and reducing reliance on manual input, this approach offers a practical, automated, and high-precision solution for 3D cultural heritage restoration.\u003c/p\u003e","manuscriptTitle":"3D Gaussian-Driven SAM2 Repair Method]{3D Gaussian-Driven SAM2 Multi-View Fusion Detection and Triple-Constrained Symmetry Plane Generation Repair Method","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-01-05 10:21:40","doi":"10.21203/rs.3.rs-8451382/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":"cc2c9f1b-3a5f-4a21-ba54-42fc583d23b6","owner":[],"postedDate":"January 5th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":60217891,"name":"Computer Architecture and Engineering"}],"tags":[],"updatedAt":"2026-01-05T10:21:40+00:00","versionOfRecord":[],"versionCreatedAt":"2026-01-05 10:21:40","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8451382","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8451382","identity":"rs-8451382","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}