Conditional Activation-Free Network for Robust Multi-Weather Image Restoration

Conditional Activation-Free Network for Robust Multi-Weather Image Restoration | 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 Conditional Activation-Free Network for Robust Multi-Weather Image Restoration Rahul Shyam This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9599438/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 Adverse weather conditions including haze, rain, and snow have severe impacts on image quality and adversely affect the performances of vision-based surveillance systems. Conventional approaches tend to target single weather scenarios with a lack of generalization in real-world settings. This paper introduces cNAFNet, a conditional activation-free network that performs well in multi-weather image restoration tasks. First, FiLM blocks are used for weather-wise feature conditioning. Second, a residual architecture with skip connection is utilized for learning structure-preserving features. Third, a novel hybrid Charbonnier-SSIM loss function is introduced to train the model with the goal of achieving both high pixel and perceptual similarity. Experimental results on the WeatherBench dataset prove that the proposed approach performs the best among various conventional and SOTA methods with 24.00 dB PSNR, 0.787 SSIM, and 0.255 LPIPS values. Further visual analysis shows that the proposed model works better than other methods on real-world driving scenes under adverse weather conditions. Artificial Intelligence and Machine Learning Image Restoration Adverse Weather Removal cNAFNet FiLM Conditioning Residual Learning Hybrid Loss Function Intelligent Transportation Systems 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-9599438","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":633595052,"identity":"a4a578c5-971c-402f-a8d5-e292564e5203","order_by":0,"name":"Rahul Shyam","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA/UlEQVRIiWNgGAWjYFACHgaGBwwMCQzHGxIfMBjARA3w6ABpSQChMwceGzAYGJCi5UbiMwkCSiHAvP3sMYkEhjt5fAeS06oLCv7ImzMwP/zAUHAHpxaZM3lpQC3PiiUPHEu7PcPAwHBnA5uxBIPBM5xaJBhyzIBaDiduONiTdpvHwIBxwwEGM6ADD+PWwv8GquUw/7dioBb7DQfYv+HXIgGz5RhDGjNQS+KGAzwEbJF4Y2yRYPAsceYZhmRpHgPj5J3NPMUSCXgdlmN440PFncS++w8SP/P8kbPdzt6+8cOHP7i1QIDBASQ2MwM4pggBZC2EVY+CUTAKRsEIAwAJBFXKayP5awAAAABJRU5ErkJggg==","orcid":"","institution":"NIT, Arunachal Pradesh","correspondingAuthor":true,"prefix":"","firstName":"Rahul","middleName":"","lastName":"Shyam","suffix":""}],"badges":[],"createdAt":"2026-05-03 11:21:48","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-9599438/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9599438/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108476752,"identity":"bec5b45b-f54b-453e-acf9-5ad658108412","added_by":"auto","created_at":"2026-05-05 07:03:05","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":707996,"visible":true,"origin":"","legend":"","description":"","filename":"ConditionalActivationFreeNetworkforRobustMultiWeatherImageRestoration.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9599438/v1_covered_3832c034-6fe0-411e-bf41-c3a1972b764a.pdf"}],"financialInterests":"The authors declare no competing interests.","formattedTitle":"\u003cp\u003e\u003cstrong\u003eConditional Activation-Free Network for Robust Multi-Weather Image Restoration\u003c/strong\u003e\u003c/p\u003e","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"National Institute of Technology Arunachal Pradesh","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":"Image Restoration, Adverse Weather Removal, cNAFNet, FiLM Conditioning, Residual Learning, Hybrid Loss Function, Intelligent Transportation Systems","lastPublishedDoi":"10.21203/rs.3.rs-9599438/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9599438/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAdverse weather conditions including haze, rain, and snow have severe impacts on image quality and adversely affect the performances of vision-based surveillance systems. Conventional approaches tend to target single weather scenarios with a lack of generalization in real-world settings. This paper introduces cNAFNet, a conditional activation-free network that performs well in multi-weather image restoration tasks. First, FiLM blocks are used for weather-wise feature conditioning. Second, a residual architecture with skip connection is utilized for learning structure-preserving features. Third, a novel hybrid Charbonnier-SSIM loss function is introduced to train the model with the goal of achieving both high pixel and perceptual similarity. Experimental results on the WeatherBench dataset prove that the proposed approach performs the best among various conventional and SOTA methods with 24.00 dB PSNR, 0.787 SSIM, and 0.255 LPIPS values. Further visual analysis shows that the proposed model works better than other methods on real-world driving scenes under adverse weather conditions.\u003c/p\u003e","manuscriptTitle":"Conditional Activation-Free Network for Robust Multi-Weather Image Restoration","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-05 07:02:56","doi":"10.21203/rs.3.rs-9599438/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":"79140a84-020f-4391-80e0-2d16f3ac0ec7","owner":[],"postedDate":"May 5th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67435639,"name":"Artificial Intelligence and Machine Learning"}],"tags":[],"updatedAt":"2026-05-05T07:02:56+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-05 07:02:56","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9599438","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9599438","identity":"rs-9599438","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}