[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-83092-en":3,"doc-seo-83092-105":30,"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":21,"is_downloadable":21,"audit_status":21,"page_count":22,"language":23,"language_code":24,"site_id":25,"html_lang":24,"table_of_contents":26,"faqs":27,"seo_title":13,"seo_description":14,"update_tm":28,"read_time":29},83092,1099514067415,"Rowan","https://ap-avatar.wpscdn.com/avatar/100002539d78ffe74a7?x-image-process=image/resize,m_fixed,w_180,h_180&k=1779092875211072502",8,"Research & Report","Temporal Modeling of Optically Variable Devices in Identity Documents","Robust remote identity verification depends on analyzing faint, transparent security features—Optically Variable Devices (OVDs), or holograms—within user-captured videos under uncontrolled conditions. Existing methods often process frames independently, ignore OVD temporal dynamics, and remain vulnerable to swapping attacks, while also failing to verify specific OVD types. Limited annotation economics prevents supervised training. This work proposes two self-supervised sequence-modeling approaches for dynamic OVD verification in open-set scenarios without attack samples, achieving state-of-the-art results on public datasets and highlighting temporal modeling and anomaly detection as essential defenses in industrial pipelines.","arXiv :2607 .06408v 1 [ cs .CV] 7 Jul 2026  \nTemporal Modeling of Optically Variable Devices in Identity Documents  \nGlen Pouliquen 1 ,2, Joseph Chazalon2, Guillaume Chiron 1, Oriol Ramos Terrades3, Thierry Géraud2, and Ahmad Montaser Awal 1  \n1 IDnow Research Center, Cesson-Sévigné, France  \n[name.surname@idnow.io](name.surname@idnow.io)  \n2 EPITA Research Lab. (LRE), Le Kremlin-Bicêtre, France [name.surname@epita.fr](name.surname@epita.fr)  \n3 Computer Vision Center (CVC), Barcelona, Spain  \n[oriolrt@cvc.uab.cat](oriolrt@cvc.uab.cat)  \nAbstract. Robust remote verification of identity documents relies on analyzing faint, transparent security features like Optically Variable Devices (OVDs), or “holograms”, within user-captured videos under uncontrolled conditions. Current systems, however, face critical limitations:  \nexisting methods often treat video frames in isolation, neglecting the intrinsic dynamic nature of OVDs and leaving systems vulnerable to swapping attacks, or focus on general holographic presence and lack the ability to verify specific OVD types. Moreover, the economic infeasibility of frame-by-frame video annotation makes supervised training impractical. In this work, we introduce two novel approaches for verifying the dynamic behavior of transparent OVDs protecting the holder’s portrait, specifically designed for open-set scenarios where attack types are unknown during training. We demonstrate that these approaches can be trained without any attack samples in a self-supervised setting, surpassing previous state-of-the-art methods on public datasets while adhering strictly to industrial constraints. Our results confirm that modeling temporal dynamics is essential for defeating sophisticated attacks under realistic conditions, and underscores the promise of sequence modeling and anomaly detection for OVD verification. Code is available at [https://github.com/EPITAResearchLab/pouliquen.26.icdar](https://github.com/EPITAResearchLab/pouliquen.26.icdar).  \nKeywords: Identity Documents · OVD Verification · Fraud Detection  \n· Sequence Modeling · Anomaly Detection  \n1 Introduction  \nThis paper addresses the automated verification of Optically Variable Devices (OVDs) on identity documents captured with commodity smartphones in remote Know Your Customer (KYC) scenarios. Such scenarios are increasingly common in various applications, such as opening a bank account. The overall process typically involves the user capturing a short video of their identity document, which is then analyzed by an automated system to verify its authenticity. Other  \n2 Pouliquen et al.  \nsecurity measures are integrated into the process, such as liveness checks to ensure the user is present and prevents the use of static images, as well as safeguards against digital content injection, but are beyond the scope of this work. Ultimately, this global KYC process consists in binding an online identity to a physical document, which serves as the primary source of truth for the holder’s identity.  \nOVDs, often referred to as holograms, are critical visible security features, particularly when they overlap and protect the photo portrait linking the document to its holder. While current industrial pipelines rely heavily on attack detection (e.g., identifying screen recaptures or photocopies), they provide limited guarantees that the expected OVD is actually present and behaves as specified. Furthermore, attack scenarios are inherently unpredictable, and training samples for such attacks are scarce. his data scarcity severely restricts the performance of existing approaches in realistic settings, as demonstrated by the recent MIDVDynAttack dataset study [15], which highlights that dynamic attacks should primarily be used for evaluation rather than training.  \nIn this work, we adopt a complementary perspective based on model verification rather than specific attack detection. We aim to verify that the OVD in the portrait region is consistent with its expe","cbCaifWrX4bQk0O8","https://ap.wps.com/l/cbCaifWrX4bQk0O8","pdf",1622856,2,1,17,"English","en",105,"# Introduction\n## Problem in remote KYC and OVD verification\n## Limitations of existing frame-level or attack-dependent methods\n## Proposed approaches and contributions","[{\"question\":\"Why is temporal modeling important for Optically Variable Device (OVD) verification?\",\"answer\":\"Because OVD security depends on dynamic changes, while many existing approaches analyze frames in isolation and thus miss essential behavior. Modeling temporal dynamics helps defeat sophisticated attacks under realistic capture conditions.\"},{\"question\":\"What are the two main approaches introduced in this work?\",\"answer\":\"A discriminative span-based classifier trained from synthetically corrupted sequences, and a generative masked sequence model that uses reconstruction error in an embedding space as an anomaly score for dynamic OVD verification.\"},{\"question\":\"How does the method handle open-set scenarios where attack types are unknown during training?\",\"answer\":\"It is trained in a self-supervised setting without any attack samples, and the evaluation demonstrates improved robustness on datasets such as MIDV-Holo and MIDV-DynAttack while adhering to industrial constraints.\"}]",1784185157,43,{"code":4,"msg":31,"data":32},"ok",{"site_id":25,"language":24,"slug":33,"title":13,"keywords":34,"description":14,"schema_data":35,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":28},"temporal-modeling-of-optically-variable-devices-in-identity-documents","",{"@graph":36,"@context":85},[37,53,68],{"@type":38,"itemListElement":39},"BreadcrumbList",[40,44,47,50],{"item":41,"name":42,"@type":43,"position":21},"https://docshare.wps.com","Home","ListItem",{"item":45,"name":46,"@type":43,"position":20},"https://docshare.wps.com/document/","Document",{"item":48,"name":12,"@type":43,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":43,"position":52},"https://docshare.wps.com/document/temporal-modeling-of-optically-variable-devices-in-identity-documents/83092/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":24,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":41,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-24","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},"Why is temporal modeling important for Optically Variable Device (OVD) verification?","Question",{"text":75,"@type":76},"Because OVD security depends on dynamic changes, while many existing approaches analyze frames in isolation and thus miss essential behavior. Modeling temporal dynamics helps defeat sophisticated attacks under realistic capture conditions.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"What are the two main approaches introduced in this work?",{"text":80,"@type":76},"A discriminative span-based classifier trained from synthetically corrupted sequences, and a generative masked sequence model that uses reconstruction error in an embedding space as an anomaly score for dynamic OVD verification.",{"name":82,"@type":73,"acceptedAnswer":83},"How does the method handle open-set scenarios where attack types are unknown during training?",{"text":84,"@type":76},"It is trained in a self-supervised setting without any attack samples, and the evaluation demonstrates improved robustness on datasets such as MIDV-Holo and MIDV-DynAttack while adhering to industrial constraints.","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":25},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,123,128,131,135],{"id":21,"doc_module":4,"doc_module_name":46,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":20,"doc_module":4,"doc_module_name":46,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":46,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":46,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":46,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":46,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":46,"category_name":12,"show_sort_weight":121,"slug":122},30,"research-report",{"id":124,"doc_module":4,"doc_module_name":46,"category_name":125,"show_sort_weight":126,"slug":127},9,"Religion & Spirituality",20,"religion-spirituality",{"id":126,"doc_module":4,"doc_module_name":46,"category_name":129,"show_sort_weight":126,"slug":130},"World Cup","world-cup",{"id":132,"doc_module":4,"doc_module_name":46,"category_name":133,"show_sort_weight":132,"slug":134},10,"Lifestyle","lifestyle",{"id":136,"doc_module":4,"doc_module_name":46,"category_name":137,"show_sort_weight":106,"slug":138},19,"General","general"]