RAFTcorr: A Deep Learning Digital Image Correlation Framework with Operating-Boundary Characterization

RAFTcorr: A Deep Learning Digital Image Correlation Framework with Operating-Boundary Characterization | 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 RAFTcorr: A Deep Learning Digital Image Correlation Framework with Operating-Boundary Characterization Zixiang Tong, Lehu Bu, Qihang Shi, Runtian Du, Jin Yang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9452500/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 Background: Digital Image Correlation (DIC) has emerged as a powerful full-field deformation measurement technique in experimental mechanics. Conventional DIC methods rely on subset-based or finite-element-based formulations, which require prescribed deformation models and involve inherent trade-offs between spatial resolution and measurement uncertainty. These approaches also depend on user-defined parameters, such as subset size and step size, which can significantly influence the results. Recent advances in deep learning, particularly the Recurrent All-Pairs Field Transforms (RAFT) framework, enable dense, pixel-wise displacement estimation with reduced user intervention. However, the impact of model architecture on the practical operating boundaries of learning-based DIC remains insufficiently understood. Objective: In this work, we present RAFTcorr, an accessible deep learning-based DIC framework built upon the RAFT architecture, together with an integrated graphical user interface (GUI) for streamlined experimental workflows. The framework provides an end-to-end “input-to-result” pipeline for dense displacement and strain field measurement with minimal user intervention. In addition, we aim to systematically characterize the operating boundaries of RAFT-based DIC across a range of deformation conditions. Methods: A comprehensive benchmark suite is developed to evaluate performance under rigid-body translation, rotation, large finite deformation, complex geometries, and high-frequency deformation fields. Synthetic datasets with known ground truth are used for quantitative validation, while real experimental datasets are employed to assess practical applicability. The influence of speckle size, deformation magnitude, and motion type on measurement performance is systematically investigated. Results: The results reveal a strong coupling between speckle characteristics, deformation magnitude, and model performance, and identify distinct operating regimes in which RAFT-based DIC maintains sub-pixel accuracy or exhibits rapid degradation. These findings establish, for the first time, a quantitative characterization of the operating boundaries of learning-based DIC. Conclusions: This work provides both a practical research software framework for dense deformation measurement and a set of experimentally grounded guidelines for applying deep learning-based DIC in real-world scenarios. Digital Image Correlation Machine learning Recurrent All-Pairs Field Transforms Optical flow Full Text Additional Declarations The authors declare no competing interests. Supplementary Files S5casecavitationflowmagnitudestreamline.mp4 Video S5_Velocity_Streamlines_in_Cavitation_Flow S6casefoamfracturedisp.mp4 Video S6: Mode I Fracture Tracking in Foam Material S7casefoamfracturestrain.mp4 Video S7: Von Mises Strain in Fracture Test Suppvideodescription.pdf Supp_video_description S1Howtouseraftcorr.mp4 Video S1: GUI Demonstration and Workflow S2caseAluminumwithholedisplacement.mp4 Video S2: Full-Field Displacement in Uniaxial Tensile Test S3caseAluminumwithholevonMisesstrain.mp4 Video S3: Von Mises Strain Evolution (Uniaxial Tension) S4casecavitationflowdisplacement.mp4 Video S4: Velocity Field Tracking in Laser-Induced Cavitation 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-9452500","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":625184144,"identity":"34f117bf-59af-436a-a25b-5e1c123b4bb7","order_by":0,"name":"Zixiang Tong","email":"","orcid":"https://orcid.org/0009-0008-6807-0757","institution":"The University of Texas at Austin","correspondingAuthor":false,"prefix":"","firstName":"Zixiang","middleName":"","lastName":"Tong","suffix":""},{"id":625184145,"identity":"9ba2aff0-08de-403d-83d2-f0f52a143faf","order_by":1,"name":"Lehu Bu","email":"","orcid":"","institution":"The University of Texas at Austin","correspondingAuthor":false,"prefix":"","firstName":"Lehu","middleName":"","lastName":"Bu","suffix":""},{"id":625184146,"identity":"2eb2bbab-e59e-4f45-bd2d-d7f09992f050","order_by":2,"name":"Qihang Shi","email":"","orcid":"","institution":"The University of Texas at Austin","correspondingAuthor":false,"prefix":"","firstName":"Qihang","middleName":"","lastName":"Shi","suffix":""},{"id":625184147,"identity":"2d8841f8-3f75-4e75-a114-0ce9ffd75450","order_by":3,"name":"Runtian Du","email":"","orcid":"","institution":"The University of Texas at Austin","correspondingAuthor":false,"prefix":"","firstName":"Runtian","middleName":"","lastName":"Du","suffix":""},{"id":625184148,"identity":"0824438c-256f-4287-a2c7-cbc3878b5f2f","order_by":4,"name":"Jin Yang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABFElEQVRIiWNgGAWjYBACxmYogx9MGiDLHSCgRbIBpAikhY2AFjgwOABTREgLczvvAWaemjt2m4+fMf78oeAOg8H95mMSP/4wyPHdSMDhML4EZp5jz5K3nckxkzhg8IzB4BhbsmFvG4OxJE4tPAbMPGyHk80O5JgB/XIYqIXH8AFvA0PiBrxa/h1ONu5/Y/wBooX/w8E/fxjq8WrhbTtsZyCRYyABtYXxMQ8bQ4IBHi0H5/YdTpC48axM4ozBYR7JY2nGxrJtEoYzzzzAqsWw/4zhgzffDtvz9ydv/lDx57Ac3+HDzyTf/LGR5zuO3RbDBgaGQzwMDIkNUAEeKC2BVTkIyIMc94OBwR6nilEwCkbBKBgFABu3Yz18SB4NAAAAAElFTkSuQmCC","orcid":"https://orcid.org/0000-0002-5967-980X","institution":"The University of Michigan-Ann Arbor","correspondingAuthor":true,"prefix":"","firstName":"Jin","middleName":"","lastName":"Yang","suffix":""}],"badges":[],"createdAt":"2026-04-17 20:18:05","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-9452500/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9452500/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107868758,"identity":"5daabde7-3244-4453-a11f-793ce9971729","added_by":"auto","created_at":"2026-04-27 07:33:24","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":7100117,"visible":true,"origin":"","legend":"","description":"","filename":"RAFTcorr20260412.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1_covered_b61a01c8-5cc4-4eb4-a48d-dfec0467e97e.pdf"},{"id":107533092,"identity":"2e206a09-fbfe-4d60-ab47-95a1b83607d1","added_by":"auto","created_at":"2026-04-22 10:47:32","extension":"mp4","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":1053829,"visible":true,"origin":"","legend":"\u003cp\u003eVideo S5_Velocity_Streamlines_in_Cavitation_Flow\u003c/p\u003e","description":"","filename":"S5casecavitationflowmagnitudestreamline.mp4","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/6eeeb251b22a6b970b69213f.mp4"},{"id":107533093,"identity":"7d1fffa2-b6b4-4a21-a4f7-0ddcaab7b66f","added_by":"auto","created_at":"2026-04-22 10:47:32","extension":"mp4","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":2112678,"visible":true,"origin":"","legend":"\u003cp\u003eVideo S6: Mode I Fracture Tracking in Foam Material\u003c/p\u003e","description":"","filename":"S6casefoamfracturedisp.mp4","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/1f61763de38c04e759fa75fa.mp4"},{"id":107705471,"identity":"9f81de52-5463-44bb-89ff-90d910e2e5ed","added_by":"auto","created_at":"2026-04-24 09:13:00","extension":"mp4","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":491258,"visible":true,"origin":"","legend":"\u003cp\u003eVideo S7: Von Mises Strain in Fracture Test\u003c/p\u003e","description":"","filename":"S7casefoamfracturestrain.mp4","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/8ccc3ece6c7c5496bfdb176a.mp4"},{"id":107533095,"identity":"50e0cad1-a233-40f9-916a-90c8639a4b39","added_by":"auto","created_at":"2026-04-22 10:47:32","extension":"pdf","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":132778,"visible":true,"origin":"","legend":"\u003cp\u003eSupp_video_description\u003c/p\u003e","description":"","filename":"Suppvideodescription.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/b4164fac0c67c3ba09ba4bba.pdf"},{"id":107533096,"identity":"5a06c5fb-4583-459f-b5ce-3e19ae14236b","added_by":"auto","created_at":"2026-04-22 10:47:32","extension":"mp4","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":3393934,"visible":true,"origin":"","legend":"\u003cp\u003eVideo S1: GUI Demonstration and Workflow\u003c/p\u003e","description":"","filename":"S1Howtouseraftcorr.mp4","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/ab38e20daea8c95eaaf2e5a5.mp4"},{"id":107705828,"identity":"fd9ff2e7-5d93-40c9-ac3c-9b98ed739ec9","added_by":"auto","created_at":"2026-04-24 09:15:27","extension":"mp4","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":695175,"visible":true,"origin":"","legend":"\u003cp\u003eVideo S2: Full-Field Displacement in Uniaxial Tensile Test\u003c/p\u003e","description":"","filename":"S2caseAluminumwithholedisplacement.mp4","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/33d18fff158b7c3281aa2866.mp4"},{"id":107533097,"identity":"39b4e9df-084e-42c2-b409-31389ae9088e","added_by":"auto","created_at":"2026-04-22 10:47:32","extension":"mp4","order_by":7,"title":"","display":"","copyAsset":false,"role":"supplement","size":294070,"visible":true,"origin":"","legend":"\u003cp\u003eVideo S3: Von Mises Strain Evolution (Uniaxial Tension)\u003c/p\u003e","description":"","filename":"S3caseAluminumwithholevonMisesstrain.mp4","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/c6754f4da1c582789cfdc54b.mp4"},{"id":107533099,"identity":"a07eece4-8dbc-44e6-a685-0657b5300ac2","added_by":"auto","created_at":"2026-04-22 10:47:32","extension":"mp4","order_by":8,"title":"","display":"","copyAsset":false,"role":"supplement","size":1200564,"visible":true,"origin":"","legend":"\u003cp\u003eVideo S4: Velocity Field Tracking in Laser-Induced Cavitation\u003c/p\u003e","description":"","filename":"S4casecavitationflowdisplacement.mp4","url":"https://assets-eu.researchsquare.com/files/rs-9452500/v1/11e5e43443f4b2bec2640ad5.mp4"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003eRAFTcorr: A Deep Learning Digital Image Correlation Framework with Operating-Boundary Characterization\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"The University of Texas at Austin","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":true,"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":"Digital Image Correlation, Machine learning, Recurrent All-Pairs Field Transforms, Optical flow","lastPublishedDoi":"10.21203/rs.3.rs-9452500/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9452500/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground: \u003c/strong\u003eDigital Image Correlation (DIC) has emerged as a powerful full-field deformation measurement technique in experimental mechanics. Conventional DIC methods rely on subset-based or finite-element-based formulations, which require prescribed deformation models and involve inherent trade-offs between spatial resolution and measurement uncertainty. These approaches also depend on user-defined parameters, such as subset size and step size, which can significantly influence the results. Recent advances in deep learning, particularly the Recurrent All-Pairs Field Transforms (RAFT) framework, enable dense, pixel-wise displacement estimation with reduced user intervention. However, the impact of model architecture on the practical operating boundaries of learning-based DIC remains insufficiently understood.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eObjective: \u003c/strong\u003eIn this work, we present RAFTcorr, an accessible deep learning-based DIC framework built upon the RAFT architecture, together with an integrated graphical user interface (GUI) for streamlined experimental workflows. The framework provides an end-to-end “input-to-result” pipeline for dense displacement and strain field measurement with minimal user intervention. In addition, we aim to systematically characterize the operating boundaries of RAFT-based DIC across a range of deformation conditions.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods: \u003c/strong\u003eA comprehensive benchmark suite is developed to evaluate performance under rigid-body translation, rotation, large finite deformation, complex geometries, and high-frequency deformation fields. Synthetic datasets with known ground truth are used for quantitative validation, while real experimental datasets are employed to assess practical applicability. The influence of speckle size, deformation magnitude, and motion type on measurement performance is systematically investigated.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eThe results reveal a strong coupling between speckle characteristics, deformation magnitude, and model performance, and identify distinct operating regimes in which RAFT-based DIC maintains sub-pixel accuracy or exhibits rapid degradation. These findings establish, for the first time, a quantitative characterization of the operating boundaries of learning-based DIC.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e This work provides both a practical research software framework for dense deformation measurement and a set of experimentally grounded guidelines for applying deep learning-based DIC in real-world scenarios.\u003c/p\u003e","manuscriptTitle":"RAFTcorr: A Deep Learning Digital Image Correlation Framework with Operating-Boundary Characterization","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-22 10:47:27","doi":"10.21203/rs.3.rs-9452500/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":"c0908b8a-81f6-4717-abe7-6d039c836d81","owner":[],"postedDate":"April 22nd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-22T10:47:27+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-22 10:47:27","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9452500","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9452500","identity":"rs-9452500","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}