{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/rgnet-a-unified-retrieval-and-grounding","title":"RGNet: A Unified Clip Retrieval and Grounding Network for Long Videos","arxiv_id":"2312.06729","date":"2023-12-11","proceeding":null,"authors":["Tanveer Hannan","Md Mohaiminul Islam","Thomas Seidl","Gedas Bertasius"],"abstract":"Locating specific moments within long videos (20-120 minutes) presents a significant challenge, akin to finding a needle in a haystack. Adapting existing short video (5-30 seconds) grounding methods to this problem yields poor performance. Since most real life videos, such as those on YouTube and AR/VR, are lengthy, addressing this issue is crucial. Existing methods typically operate in two stages: clip retrieval and grounding. However, this disjoint process limits the retrieval module's fine-grained event understanding, crucial for specific moment detection. We propose RGNet which deeply integrates clip retrieval and grounding into a single network capable of processing long videos into multiple granular levels, e.g., clips and frames. Its core component is a novel transformer encoder, RG-Encoder, that unifies the two stages through shared features and mutual optimization. The encoder incorporates a sparse attention mechanism and an attention loss to model both granularity jointly. Moreover, we introduce a contrastive clip sampling technique to mimic the long video paradigm closely during training. RGNet surpasses prior methods, showcasing state-of-the-art performance on long video temporal grounding (LVTG) datasets MAD and Ego4D.","url_abs":"https://arxiv.org/abs/2312.06729v3","url_pdf":"https://arxiv.org/pdf/2312.06729v3.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"rgnet-a-unified-retrieval-and-grounding","repo_url":"https://github.com/tanveer81/rgnet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"rgnet-a-unified-retrieval-and-grounding","repo_url":"https://github.com/tanveer81/revisionllm","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"natural-language-moment-retrieval","task_name":"Natural Language Moment Retrieval"},{"task_slug":"natural-language-queries","task_name":"Natural Language Queries"},{"task_slug":"retrieval","task_name":"Retrieval"},{"task_slug":"text-retrieval","task_name":"Text Retrieval"},{"task_slug":"video-text-retrieval","task_name":"Video-Text Retrieval"}],"methods":[{"method_slug":"clip","method_name":"CLIP"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/natural-language-moment-retrieval-on-mad","task":"Natural Language Moment Retrieval","dataset":"MAD","model":"RGNet","rank_in_archive_order":3,"of":8,"metrics":{"R@1,IoU=0.1":"12.43","R@1,IoU=0.3":"9.48","R@1,IoU=0.5":"5.61","R@5,IoU=0.1":"25.12","R@5,IoU=0.3":"18.72","R@5,IoU=0.5":"10.86"},"uses_additional_data":false},{"leaderboard":"/sota/natural-language-queries-on-ego4d","task":"Natural Language Queries","dataset":"Ego4D","model":"RGNet","rank_in_archive_order":4,"of":10,"metrics":{"R@1 IoU=0.3":"20.63","R@1 IoU=0.5":"12.47","R@1 Mean(0.3 and 0.5)":"16.55","R@5 IoU=0.3":"41.67","R@5 IoU=0.5":"25.08"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2312.06729","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}