[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-85662-en":3,"doc-seo-85662-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},85662,549758252649,"Ivy","https://ap-avatar.wpscdn.com/avatar/8000253669c5317157?_k=1778319167496531819",8,"Research & Report","Continual Test-Time Adaptation in Computer Vision","Deep neural networks excel when training and test data follow the same distribution, yet real deployments face continual distribution shifts. Continual Test-Time Adaptation (CTTA) adapts pretrained vision models to evolving, non-stationary target domains during inference without source data or labeled targets. It addresses catastrophic forgetting of source knowledge and error accumulation from noisy pseudo-labels. The survey defines CTTA formally, studies continual domain-shift patterns across protocols, and organizes methods into optimization, parameter-efficient, and architecture-based families.","arXiv :2607 .08164v2 [ cs .CV] 12 Jul 2026  \nContinual Test-Time Adaptation in Computer Vision: Methods, Benchmarks, and Future Directions  \nSarthak Kumar Maharana 1 Shambhavi Mishra2 ∗ Yunbei Zhang3 ∗ Shuaicheng Niu4  \nTaki Hasan Rafi5  \nJihun Hamm3 Marco Pedersoli2 Jose Dolz2  \nYunhui Guo 1  \n[sarthak.maharana@utdallas. edu](sarthak.maharana@utdallas. edu)[ ](sarthak.maharana@utdallas. edu)[shambhavi.mishra.1@etsmtl.net](shambhavi.mishra.1@etsmtl.net)[ ](shambhavi.mishra.1@etsmtl.net)[yzhang111@tulane. edu](yzhang111@tulane. edu)[ ](yzhang111@tulane. edu)[shuaicheng.niu@ntu. edu.sg](shuaicheng.niu@ntu. edu.sg)[takihr@hanyang. ac.kr](takihr@hanyang. ac.kr)[ ](takihr@hanyang. ac.kr)[jhamm3@tulane. edu](jhamm3@tulane. edu)  \n[marco.pedersoli@etsmtl. ca](marco.pedersoli@etsmtl. ca)[ ](marco.pedersoli@etsmtl. ca)[jose. dolz@etsmtl. ca](jose. dolz@etsmtl. ca)  \n[yunhui.guo@utdallas. edu](yunhui.guo@utdallas. edu)  \n1 The University of Texas at Dallas, USA  \n2 LIVIA ETS Montreal, ILLS International Laboratory on Learning Systems (ILLS), McGill - ÉTS-MILA - CNRS  \n- Université Paris-Saclay - CentraleSupélec  \n3 Tulane University, USA  \n4 Nanyang Technological University, Singapore  \n5 Hanyang University, South Korea  \nReviewed on OpenReview: [https: // openreview. net/ forum? id= mM3r03Xw1V](https: // openreview. net/ forum? id= mM3r03Xw1V)  \nAbstract  \nDeep neural nets achieve remarkable performance when training and test data share the same distribution, but this assumption frequently breaks in real-world deployment, where data undergoes continual distributional shifts. Continual Test-Time Adaptation (CTTA) addresses this challenge by adapting pretrained models to non-stationary target distributionson-the-fly, without access to source data or labeled targets, while mitigating two critical failure modes: catastrophic forgetting of source knowledge and error accumulation from noisy pseudo-labels over extended time horizons. In this comprehensive survey, we formally define the CTTA problem, analyze the diverse continual domain shift patterns that characterize different evaluation protocols, and propose a hierarchical taxonomy that categorizes existing methods into three families: optimization-based strategies (entropy minimization, pseudolabeling, parameter restoration), parameter-efficient methods (normalization layer adaptation, adaptive parameter selection), and architecture-based approaches (teacher-student frameworks, adapters, visual prompting, masked modeling) . We systematically review representative methods within each category and present comparative benchmarks and experimental results across standard evaluation settings. Finally, we discuss the limitations of current approaches and highlight emerging research directions, including the adaptation of foundation models and black-box systems, thereby providing a roadmap for future research in robust continual test-time adaptation. We encourage visiting our repository at  \n[https://github.com/sarthaxxxxx/Awesome-Continual-Test-Time-Adaptation](https://github.com/sarthaxxxxx/Awesome-Continual-Test-Time-Adaptation)  \n1 Introduction  \nIn recent years, deep neural networks have achieved remarkable performance across a wide range of tasks (Silver et al. , 2018 ; Russakovsky et al. , 2015 ; Jumper et al. , 2021), largely due to the assumption that the  \n∗ Equal contribution  \nFigure 1: Comparison of adaptation paradigms under distribution shift. Domain Generalization (DG)(Zhou et al. , 2022) trains on multiple source domains but keeps the model frozen at deployment. Domain Adaptation (DA) (Kouw & Loog, 2019) jointly trains on source and unlabeled target data, also resulting ina frozen deployment model. Test-Time Adaptation (TTA) (Wang et al. , 2021) trains only on source data but adapts to a single target domain during deployment. Continual Test-Time Adaptation (CTTA)(Wang et al. , 2022) trains on source only and continually adapts to a sequence of evolving target domainsat test time","cbCainWMari31Eaa","https://ap.wps.com/l/cbCainWMari31Eaa","pdf",1790705,1,48,"English","en",105,"# Introduction\n## Distributional Shifts and Deployment Challenges\n## Adaptation Paradigms (DG, DA, TTA, CTTA)","[{\"question\":\"What problem does Continual Test-Time Adaptation (CTTA) address?\",\"answer\":\"CTTA tackles continual distributional shifts during deployment by adapting a pretrained model to evolving target domains on the fly, without access to source data or labeled targets.\"},{\"question\":\"How does CTTA prevent the main failure modes during adaptation?\",\"answer\":\"CTTA mitigates catastrophic forgetting of source knowledge and reduces error accumulation caused by noisy pseudo-labels over long time horizons.\"},{\"question\":\"How are CTTA methods organized in the survey?\",\"answer\":\"The survey proposes a hierarchical taxonomy grouping methods into three families: optimization-based strategies, parameter-efficient methods, and architecture-based approaches.\"}]",1784205440,121,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"continual-test-time-adaptation-in-computer-vision","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/continual-test-time-adaptation-in-computer-vision/85662/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problem does Continual Test-Time Adaptation (CTTA) address?","Question",{"text":75,"@type":76},"CTTA tackles continual distributional shifts during deployment by adapting a pretrained model to evolving target domains on the fly, without access to source data or labeled targets.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does CTTA prevent the main failure modes during adaptation?",{"text":80,"@type":76},"CTTA mitigates catastrophic forgetting of source knowledge and reduces error accumulation caused by noisy pseudo-labels over long time horizons.",{"name":82,"@type":73,"acceptedAnswer":83},"How are CTTA methods organized in the survey?",{"text":84,"@type":76},"The survey proposes a hierarchical taxonomy grouping methods into three families: optimization-based strategies, parameter-efficient methods, and architecture-based approaches.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":45,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":45,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":45,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":45,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]