[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-82522-en":3,"doc-seo-82522-105":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},82522,687197100911,"Himbo","https://ap-avatar.wpscdn.com/avatar/a000239b6f1da00475?x-image-process=image/resize,m_fixed,w_180,h_180&k=1782698725881665579",8,"Research & Report","VideoSearch-R1 Iterative Video Retrieval and Reasoning via Soft Query Refinement","VideoSearch-R1 addresses the gap between video corpus retrieval and fine-grained, query-conditioned reasoning such as temporal grounding. Existing methods treat retrieval as a fixed preprocessing step, so mistakes early in retrieval cannot be corrected later, causing downstream failures. The proposed agentic framework iteratively interacts with a video search engine and introduces Soft Query Refinement (SQR), refining query tokens in a continuous latent space rather than rewriting discrete text queries. Training uses Group Relative Policy Optimization (GRPO) with task-level rewards from retrieval and downstream grounding, achieving state-of-the-art results on VCMR.","arXiv :2607 .00446v 1 [ cs .CV] 1 Jul 2026  \nVideoSearch-R1: Iterative Video Retrieval and Reasoning via Soft Query Refinement  \nSeohyun Lee 1 ∗, Seoung Choi 1 ∗, Dohwan Ko2 ∗, Jongha Kim2, and  \nHyunwoo J. Kim 1†  \n1 KAIST, Daejeon, Republic of Korea  \n{seohyunlee, choisw0823, [hyunwoojkim}@kaist.ac.kr](hyunwoojkim}@kaist.ac.kr)  \n2 Korea University, Seoul, Republic of Korea  \n{ikodoh, [jonghakim}@korea.ac.kr](jonghakim}@korea.ac.kr)  \nAbstract. As video corpora continue to expand in both scale and task complexity, there is increasing demand for approaches that retrieve relevant videos from large-scale corpora (inter-video reasoning) and subsequently perform fine-grained, query-conditioned tasks (intra-video reasoning) within the retrieved content, such as temporal grounding. However, existing approaches typically treat retrieval as a preprocessing step, and consequently, when the initial retrieval fails, there is no mechanism to refine the search, leading to the failure of subsequent fine-grained intra-video reasoning. Moreover, while recent agentic frameworks have advanced video understanding, they typically assume that the queryrelevant video is already given, focusing exclusively on intra-video reasoning tasks. To address these limitations, we propose VideoSearch-R1 , an agentic framework for iterative video retrieval and reasoning through multi-turn interaction with a video search engine. Specifically, we introduce Soft Query Refinement (SQR) to refine search query tokens in a continuous latent space rather than rewriting queries in the discrete text space, enabling more efficient and fine-grained adjustments. SQR and its reasoning process are trained using Group Relative Policy Optimization (GRPO), guided by task-level reward signals derived from retrieval and downstream tasks. Building upon this, VideoSearch-R1 achieves stateof-the-art performance across three datasets on Video Corpus Moment Retrieval (VCMR), iteratively retrieving videos from large-scale corpora, refining search queries, and performing precise query-conditioned temporal grounding within the retrieved content. Our analyses show that SQR effectively refines the original query, requiring significantly fewer generated tokens than explicit text-level query refinement. Code and model [checkpoints are publicly available at](checkpoints are publicly available at mlvlab.github.io/VideoSearch-R1)[ mlvlab.github.io/VideoSearch-R1](checkpoints are publicly available at mlvlab.github.io/VideoSearch-R1) .  \nKeywords: Reinforcement Learning · Agentic AI · Video Retrieval  \n1 Introduction  \nWith the rapid growth of large-scale video corpora, recent studies have focused  \non efficiently  and accurately retrieving relevant videos given a user query [8, 14,∗ Equal contribution. † Corresponding author.  \n2 S. Lee et al.  \nFig. 1: An illustrative example of VideoSearch-R1 . As an agentic AI system, VideoSearch-R1 enables multi-turn interaction through iterative video retrieval and reasoning, leveraging an external video search engine. This pipeline unifies corpus-level inter-video reasoning (e.g ., video retrieval) with intra-video reasoning (e.g ., temporal grounding) grounded in the retrieved video.  \n15,19–22,41,43] . Although these approaches achieve strong performance on standard video-level retrieval benchmarks through inter-video reasoning, identifying the correct video alone is insufficient for real-world applications. In practice, users require not only coarse inter-video reasoning but also query-specific intra-video reasoning within the retrieved video. For example, beyond identifying a relevant video, a system may need to conduct fine-grained reasoning, such as localizing the exact timestamp of a described event, extracting temporally grounded evidence, or performing question answering over the retrieved video.  \nHowever, existing pipelines [10, 47, 48] treat inter-video retrieval as a preprocessing stage prior to intra-video reasoning: inter-video retrieval mo","cbCaih58RrdHUJDv","https://ap.wps.com/l/cbCaih58RrdHUJDv","pdf",20543789,1,29,"English","en",105,"# Introduction\n## Problem: decoupled retrieval and intra-video reasoning\n## Proposed solution: iterative agentic retrieval-reasoning\n## Related work: retrieval-augmented and video agentic systems","[{\"question\":\"What limitation in existing video systems motivates VideoSearch-R1?\",\"answer\":\"Existing pipelines typically run inter-video retrieval once as preprocessing and then perform intra-video reasoning independently. When initial retrieval fails, there is no mechanism to refine the search, so downstream fine-grained reasoning such as temporal grounding can fail.\"},{\"question\":\"How does Soft Query Refinement (SQR) improve iterative retrieval?\",\"answer\":\"SQR refines search query tokens in a continuous latent space instead of rewriting queries in the discrete text space. This enables more efficient and fine-grained adjustments during multi-turn interaction.\"},{\"question\":\"How is VideoSearch-R1 trained and what signals guide learning?\",\"answer\":\"SQR and the overall reasoning process are trained using Group Relative Policy Optimization (GRPO). Training is guided by task-level reward signals derived from the retrieval step and downstream tasks like temporal grounding.\"}]",1784181196,73,{"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},"videosearch-r1-iterative-video-retrieval-and-reasoning-via-soft-query-refinement","",{"@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/videosearch-r1-iterative-video-retrieval-and-reasoning-via-soft-query-refinement/82522/",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 limitation in existing video systems motivates VideoSearch-R1?","Question",{"text":75,"@type":76},"Existing pipelines typically run inter-video retrieval once as preprocessing and then perform intra-video reasoning independently. When initial retrieval fails, there is no mechanism to refine the search, so downstream fine-grained reasoning such as temporal grounding can fail.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How does Soft Query Refinement (SQR) improve iterative retrieval?",{"text":80,"@type":76},"SQR refines search query tokens in a continuous latent space instead of rewriting queries in the discrete text space. This enables more efficient and fine-grained adjustments during multi-turn interaction.",{"name":82,"@type":73,"acceptedAnswer":83},"How is VideoSearch-R1 trained and what signals guide learning?",{"text":84,"@type":76},"SQR and the overall reasoning process are trained using Group Relative Policy Optimization (GRPO). Training is guided by task-level reward signals derived from the retrieval step and downstream tasks like temporal grounding.","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"]