[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-327246-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-327246-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","longvqubench-benchmarking-long-term-video-quality-understanding-of-vision-language-models","LongVQUBench - Benchmarking Long-Term Video Quality Understanding of Vision-Language Models","","Evaluation of long-term video quality understanding remains an open challenge for large vision-language models (LVLMs). Existing video quality benchmarks emphasize short clips and isolated distortions, missing temporal continuity, cumulative degradation, and complex reasoning in long-duration content. LongVQUBench provides 1,200+ diverse videos and 1,500 questions, with three hierarchical levels (LQU, CQR, GQU) plus needle distortion QA (NDQA) to probe fine-grained detection and temporal quality reasoning.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/longvqubench-benchmarking-long-term-video-quality-understanding-of-vision-language-models/327246/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/longvqubench-benchmarking-long-term-video-quality-understanding-of-vision-language-models/327246.png","ImageObject",300,407,{"name":42,"@type":43},"Finn","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-11-05","2026-09-21",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",11,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does LongVQUBench address for LVLMs?","Question",{"text":63,"@type":64},"It targets long-term video quality understanding, where models must reason about perceptual fidelity, temporal coherence, and cumulative degradation across extended durations.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"What are the three evaluation levels in LongVQUBench?",{"text":68,"@type":64},"Local event quality understanding (LQU), cross-event quality reasoning (CQR), and global quality understanding (GQU), each increasing temporal and reasoning complexity.",{"name":70,"@type":61,"acceptedAnswer":71},"How does the NDQA paradigm work in LongVQUBench?",{"text":72,"@type":64},"Needle distortion QA sparsely inserts spatial or temporal artifacts with varying intensities, challenging models to detect and reason beyond coarse semantic cues.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},327246,1790003535,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":140,"language":141,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":12,"update_tm":145,"read_time":146,"preview_page_count":4},34359740700684,"https://ap-avatar.wpscdn.com/avatar/1f400023980c374ae676?_k=1777273430885731487","arXiv :2607 .0 1086v 1 [ cs .CV] 1 Jul 2026  \nLongVQUBench: Benchmarking Long-Term Video Quality Understanding of Vision-Language Models  \nArpita Nema, Hanwei Zhu, Xi Zhang, and Weisi Lin  \nNanyang Technological University, Singapore [arpita004@e.ntu.edu.sg](arpita004@e.ntu.edu.sg) , {hanwei.zhu, xi.zhang, [WSLin}@ntu.edu.sg](WSLin}@ntu.edu.sg)[ ](WSLin}@ntu.edu.sg)Project page: [https://longvqubench.github.io](https://longvqubench.github.io).  \nAbstract. The evaluation of long-term video quality understanding remains an open challenge for large vision-language models (LVLMs) . Existing video quality benchmarks predominantly focus on short clips and isolated distortions, overlooking the temporal continuity, cumulative degradation, and reasoning complexity inherent in long-duration content. To address these limitations, we present LongVQUBench, a comprehensive benchmark for long-term video quality understanding. LongVQUBench contains over 1 ,200 diverse videos spanning movies, documentaries, surveillance footage, egocentric recordings, and animated content, accompanied by 1 ,500 multiple-choice and open-ended questions for validation and testing. To assess perceptual reasoning across different temporal scopes, we introduce three progressively complex evaluation levels: (i) local event quality understanding (LQU) for analyzing localized distortions; (ii) cross-event quality reasoning (CQR) for integrating multiple degraded events; and (iii) global quality understanding (GQU) for holistic perceptual evaluation over extended durations. Furthermore, a needle distortion question-answering (NDQA) paradigm is embedded across all three levels, where spatial or temporal artifacts are sparsely inserted to probe fine-grained detection and reasoning capabilities. Extensive experiments on 14 state-of-the-art LVLMs reveal significant performance degradation with increasing video length and reasoning depth, highlighting their limited capacity for long-range temporal integration and perceptual attribution. We envision LongVQUBench as a foundational step toward the systematic, hierarchical, and explainable evaluation of LVLMs’long-term video quality understanding.  \nKeywords: Long-term video quality understanding · Video quality benchmark · Large vision-language models  \n1 Introduction  \nThe rapid progress of large vision-language models (LVLMs) has revolutionized multimodal understanding by integrating visual perception with linguistic reasoning. Recent models such as GPT-5 [46], LLaVA-OneVision-1.5 [2], and  \n2 A. Nema et al.  \nQwen2.5-VL [5] exhibit strong capabilities in fine-grained recognition, contextaware image description [78], and short-and long-term video question answering [14, 57 , 75], signaling a transition from perception-driven analysis to semantic comprehension. Nevertheless, long-term video quality understanding (LVQU) remains underexplored. Unlike conventional video understanding tasks centered on events and semantics, LVQU requires reasoning over perceptual fidelity, temporal coherence, and cumulative degradation across extended time spans. This isan essential ability for assessing the stability and human-perceived quality of real-world long-form videos, yet still beyond the reach of current LVLMs.  \nVideo quality assessment (VQA) provides a complementary foundation for perceptual modeling by quantifying human judgments of distortions such as compression artifacts [34, 44 , 54 , 60 , 68 , 70], transmission errors [23, 42 , 45 , 64], and temporal flicker [9, 10 , 76] . Classical datasets, including LIVE-VQA [45], KoNViD-1k [20], and YouTube-UGC [52], consist mainly of short clips with controlled distortions, which are valuable for low-level modeling but insufficient for capturing the temporal continuity and contextual dynamics of long videos. In contrast, LVLM-oriented benchmarks have emphasized semantic video understanding, such as action recognition and multimodal reasoning [14, 16] . Although LongVideoBench [57] extends e","cbCaiarSalR4WuJS","https://ap.wps.com/l/cbCaiarSalR4WuJS","pdf",5269573,37,"English","# Introduction\n## LongVQUBench overview and dataset scope\n## Hierarchical evaluation levels: LQU, CQR, GQU\n## Needle distortion question-answering (NDQA)\n## Experimental findings on LVLM performance","[{\"question\":\"What problem does LongVQUBench address for LVLMs?\",\"answer\":\"It targets long-term video quality understanding, where models must reason about perceptual fidelity, temporal coherence, and cumulative degradation across extended durations.\"},{\"question\":\"What are the three evaluation levels in LongVQUBench?\",\"answer\":\"Local event quality understanding (LQU), cross-event quality reasoning (CQR), and global quality understanding (GQU), each increasing temporal and reasoning complexity.\"},{\"question\":\"How does the NDQA paradigm work in LongVQUBench?\",\"answer\":\"Needle distortion QA sparsely inserts spatial or temporal artifacts with varying intensities, challenging models to detect and reason beyond coarse semantic cues.\"}]","LongVQUBench - Benchmarking Long-Term Video Quality Understanding of Vision-Language Models | PDF",1789976804,93]