A Novel Hybrid Framework for Deepfake Detection

A Novel Hybrid Framework for Deepfake Detection | 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 A Novel Hybrid Framework for Deepfake Detection Samarth Dhol, Nishant Kanani, Diya Koyani, Devansh Javiya, Hemang Thakar, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6273520/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 Fast developments in artificial intelligence technology produced out- standing generative advancements that produced highly realistic deepfake media content which consistently outpaces existing method detection capabilities. The rising distribution of synthetic content leads to urgent threats for media authen- ticity because privacy and security issues remain even though detection systems need to be implemented immediately. The proposed combination of lightweight deepfake detection model solves present problems related to spatial-temporal fea- ture analysis and adaptive adversarial noise reduction and noise-resilient feature extraction methods.The Xception backbone operates with temporal attention to find inconsistencies between compressed video frames through the Celeb-DF-v1 and Celeb-DF-v2 datasets at real-time speeds for inference. Deepfake video de- tection on Celeb-DF-v2 achieved a top-tier success rate of 90% accuracy thus surpassing all current competing solutions by 5.7%. At the same time the pro- posed model maintained strong effectiveness when facing actual video defor- mation challenges and different dataset environments. The model shows great efficiency in combination with adaptability making it ideal for social media en- vironments where it defends against evolving synthetic media dangers at scale. Supported by future enhancement research we consider the limitations for im- proving attacks detection on previously unobserved adversary conditions. Deepfake Detection Hybrid Framework Spatial-Temporal Analysis Adversarial Noise Reduction Full Text Additional Declarations No competing interests reported. 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-6273520","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":439959126,"identity":"6b7062e2-3fcf-4311-932a-9b31288b3648","order_by":0,"name":"Samarth Dhol","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABBklEQVRIiWNgGAWjYBACfvbmxx8SKiTkGCQQggZghAtI9hwzk/hwxsYYpkWCoBaDGzlijDPb0hIbULXgAwdy2Jh52A6n989ufvaZ549dHQN78zYJhoI7OHUwNpxh/szDczh3xp1jxrN525IlGHiOlUkwGDzDqYWZsYdBmkficO4GiQRjZt4GZgkGiRwzoJbDOLUAXQXUYnA43UAi/TMzz596CQb5N/i18LDxCEjOSEhLMJDIMQZ5CmgLD34tEjxswEA+YGM440ZOMePctuOSbTxpxRYJeLTY33/8+EPiPwl5/hnpmxne/Knm52c/vPHGhz+4tWDxHYhIIEHDKBgFo2AUjAJMAAAtf0xRpIhDDgAAAABJRU5ErkJggg==","orcid":"","institution":"Chandubhai S. Patel Institute of Technology, CHARUSAT","correspondingAuthor":true,"prefix":"","firstName":"Samarth","middleName":"","lastName":"Dhol","suffix":""},{"id":439959127,"identity":"ea0fe050-2fc0-46fd-a3c6-b047bb056de5","order_by":1,"name":"Nishant Kanani","email":"","orcid":"","institution":"Chandubhai S. Patel Institute of Technology, CHARUSAT","correspondingAuthor":false,"prefix":"","firstName":"Nishant","middleName":"","lastName":"Kanani","suffix":""},{"id":439959128,"identity":"2c86bc12-80cf-41c3-a292-c37c5f6cec8a","order_by":2,"name":"Diya Koyani","email":"","orcid":"","institution":"Chandubhai S. Patel Institute of Technology, CHARUSAT","correspondingAuthor":false,"prefix":"","firstName":"Diya","middleName":"","lastName":"Koyani","suffix":""},{"id":439959129,"identity":"a73212f7-75c7-4384-b732-34dc3ccedb93","order_by":3,"name":"Devansh Javiya","email":"","orcid":"","institution":"Chandubhai S. Patel Institute of Technology, CHARUSAT","correspondingAuthor":false,"prefix":"","firstName":"Devansh","middleName":"","lastName":"Javiya","suffix":""},{"id":439959130,"identity":"26c776e1-5b34-4da7-ae5b-945fb5eedc00","order_by":4,"name":"Hemang Thakar","email":"","orcid":"","institution":"Chandubhai S. Patel Institute of Technology, CHARUSAT","correspondingAuthor":false,"prefix":"","firstName":"Hemang","middleName":"","lastName":"Thakar","suffix":""},{"id":439959131,"identity":"35bbc09e-9f9d-4166-bbe4-ce05a8d4e184","order_by":5,"name":"Amit Thakkar","email":"","orcid":"","institution":"Chandubhai S. Patel Institute of Technology, CHARUSAT","correspondingAuthor":false,"prefix":"","firstName":"Amit","middleName":"","lastName":"Thakkar","suffix":""}],"badges":[],"createdAt":"2025-03-21 03:08:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6273520/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6273520/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":88584908,"identity":"166dcf35-1aa9-473f-8b37-76922b60599e","added_by":"auto","created_at":"2025-08-08 03:53:52","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":562592,"visible":true,"origin":"","legend":"","description":"","filename":"ANovelHybridFrameworkforDeepfakeDetection.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6273520/v1_covered_fcaf6ac4-6ced-49bb-a82c-47478864b550.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"A Novel Hybrid Framework for Deepfake Detection","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"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":"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":"Deepfake Detection, Hybrid Framework, Spatial-Temporal Analysis, Adversarial Noise Reduction","lastPublishedDoi":"10.21203/rs.3.rs-6273520/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6273520/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eFast developments in artificial intelligence technology produced out- standing generative advancements that produced highly realistic deepfake media content which consistently outpaces existing method detection capabilities. The rising distribution of synthetic content leads to urgent threats for media authen- ticity because privacy and security issues remain even though detection systems need to be implemented immediately. The proposed combination of lightweight deepfake detection model solves present problems related to spatial-temporal fea- ture analysis and adaptive adversarial noise reduction and noise-resilient feature extraction methods.The Xception backbone operates with temporal attention to find inconsistencies between compressed video frames through the Celeb-DF-v1 and Celeb-DF-v2 datasets at real-time speeds for inference. Deepfake video de- tection on Celeb-DF-v2 achieved a top-tier success rate of 90% accuracy thus surpassing all current competing solutions by 5.7%. At the same time the pro- posed model maintained strong effectiveness when facing actual video defor- mation challenges and different dataset environments. The model shows great efficiency in combination with adaptability making it ideal for social media en- vironments where it defends against evolving synthetic media dangers at scale. Supported by future enhancement research we consider the limitations for im- proving attacks detection on previously unobserved adversary conditions.\u003c/p\u003e","manuscriptTitle":"A Novel Hybrid Framework for Deepfake Detection","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-17 17:38:31","doi":"10.21203/rs.3.rs-6273520/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":"c3bc5053-875f-4626-807c-84e4cf282b39","owner":[],"postedDate":"April 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-08-08T03:53:29+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-17 17:38:31","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6273520","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6273520","identity":"rs-6273520","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}