{"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/visual-keyword-spotting-with-attention","title":"Visual Keyword Spotting with Attention","arxiv_id":"2110.15957","date":"2021-10-29","proceeding":null,"authors":["K R Prajwal","Liliane Momeni","Triantafyllos Afouras","Andrew Zisserman"],"abstract":"In this paper, we consider the task of spotting spoken keywords in silent video sequences -- also known as visual keyword spotting. To this end, we investigate Transformer-based models that ingest two streams, a visual encoding of the video and a phonetic encoding of the keyword, and output the temporal location of the keyword if present. Our contributions are as follows: (1) We propose a novel architecture, the Transpotter, that uses full cross-modal attention between the visual and phonetic streams; (2) We show through extensive evaluations that our model outperforms the prior state-of-the-art visual keyword spotting and lip reading methods on the challenging LRW, LRS2, LRS3 datasets by a large margin; (3) We demonstrate the ability of our model to spot words under the extreme conditions of isolated mouthings in sign language videos.","url_abs":"https://arxiv.org/abs/2110.15957v1","url_pdf":"https://arxiv.org/pdf/2110.15957v1.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":"visual-keyword-spotting-with-attention","repo_url":"https://github.com/prajwalkr/transpotter","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"lip-reading","task_name":"Lip Reading"},{"task_slug":"visual-keyword-spotting","task_name":"Visual Keyword Spotting"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/visual-keyword-spotting-on-lrs2","task":"Visual Keyword Spotting","dataset":"LRS2","model":"Transpotter","rank_in_archive_order":1,"of":1,"metrics":{"Top-1 Accuracy":"65","Top-5 Accuracy":"87.1","mAP":"69.2","mAP IOU@0.5":"68.3"},"uses_additional_data":false},{"leaderboard":"/sota/visual-keyword-spotting-on-lrs3-ted","task":"Visual Keyword Spotting","dataset":"LRS3-TED","model":"Transpotter","rank_in_archive_order":1,"of":1,"metrics":{"Top-1 Accuracy":"52","Top-5 Accuracy":"77.1","mAP":"55.4","mAP IOU@0.5":"53.6"},"uses_additional_data":false},{"leaderboard":"/sota/visual-keyword-spotting-on-lrw","task":"Visual Keyword Spotting","dataset":"LRW","model":"Transpotter","rank_in_archive_order":1,"of":1,"metrics":{"Top-1 Accuracy":"85.8","Top-5 Accuracy":"99.6","mAP":"64.1"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2110.15957","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}