{"url":"/sota/audio-visual-active-speaker-detection-on-ava","task":{"name":"Audio-Visual Active Speaker Detection","url":"/task/audio-visual-active-speaker-detection","note":null},"dataset":{"name":"AVA-ActiveSpeaker","url":"/dataset/ava-activespeaker"},"category":"Computer Vision","categories":["Computer Vision","Time Series"],"category_note":null,"description":"Determine if and when each visible person in the video is speaking.","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["validation mean average precision"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"validation mean average precision":"higher"}},"counts":{"rows":20,"rows_with_code":13,"rows_with_paper_page":20,"rows_dated":20,"rows_using_additional_data":4},"rows":[{"rank_in_archive_order":1,"model":"LoCoNet+TalkNCE","metrics":{"validation mean average precision":"95.5%"},"uses_additional_data":false,"paper_date":"2023-09-21","paper":"/paper/talknce-improving-active-speaker-detection","paper_url":"https://arxiv.org/abs/2309.12306v1","paper_title":"TalkNCE: Improving Active Speaker Detection with Talk-Aware Contrastive Learning","code":"https://github.com/kaistmm/TalkNCE","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"LoCoNet + Laser","metrics":{"validation mean average precision":"95.3%"},"uses_additional_data":false,"paper_date":"2025-01-21","paper":"/paper/laser-lip-landmark-assisted-speaker-detection","paper_url":"https://arxiv.org/abs/2501.11899v1","paper_title":"LASER: Lip Landmark Assisted Speaker Detection for Robustness","code":"https://github.com/plnguyen2908/laser_asd","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"LoCoNet","metrics":{"validation mean average precision":"95.2%"},"uses_additional_data":false,"paper_date":"2023-01-19","paper":"/paper/loconet-long-short-context-network-for-active","paper_url":"https://arxiv.org/abs/2301.08237v2","paper_title":"LoCoNet: Long-Short Context Network for Active Speaker Detection","code":"https://github.com/sjtuwxz/loconet_asd","n_code_links":2,"syntology":null},{"rank_in_archive_order":4,"model":"SPELL+","metrics":{"validation mean average precision":"94.9%"},"uses_additional_data":false,"paper_date":"2022-07-15","paper":"/paper/learning-long-term-spatial-temporal-graphs","paper_url":"https://arxiv.org/abs/2207.07783v3","paper_title":"Learning Long-Term Spatial-Temporal Graphs for Active Speaker Detection","code":"https://github.com/sra2/spell","n_code_links":2,"syntology":null},{"rank_in_archive_order":5,"model":"UniCon+","metrics":{"validation mean average precision":"94.5%"},"uses_additional_data":false,"paper_date":"2022-06-22","paper":"/paper/unicon-ictcas-ucas-submission-to-the-ava","paper_url":"https://arxiv.org/abs/2206.10861v1","paper_title":"UniCon+: ICTCAS-UCAS Submission to the AVA-ActiveSpeaker Task at ActivityNet Challenge 2022","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":6,"model":"SPELL","metrics":{"validation mean average precision":"94.2%"},"uses_additional_data":false,"paper_date":"2022-07-15","paper":"/paper/learning-long-term-spatial-temporal-graphs","paper_url":"https://arxiv.org/abs/2207.07783v3","paper_title":"Learning Long-Term Spatial-Temporal Graphs for Active Speaker Detection","code":"https://github.com/sra2/spell","n_code_links":2,"syntology":null},{"rank_in_archive_order":7,"model":"EASEE-50","metrics":{"validation mean average precision":"94.1%"},"uses_additional_data":true,"paper_date":"2022-03-27","paper":"/paper/end-to-end-active-speaker-detection","paper_url":"https://arxiv.org/abs/2203.14250v2","paper_title":"End-to-End Active Speaker Detection","code":"https://github.com/fuankarion/end-to-end-asd","n_code_links":3,"syntology":{"n_ran":3,"n_unverified":2,"n_samples":5,"n_pointer_only_licence":5}},{"rank_in_archive_order":8,"model":"Light-ASD","metrics":{"validation mean average precision":"94.1%"},"uses_additional_data":false,"paper_date":"2023-03-08","paper":"/paper/a-light-weight-model-for-active-speaker","paper_url":"https://arxiv.org/abs/2303.04439v1","paper_title":"A Light Weight Model for Active Speaker Detection","code":"https://github.com/junhua-liao/light-asd","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":1,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"Extended UniCon","metrics":{"validation mean average precision":"93.6%"},"uses_additional_data":false,"paper_date":"2021-06-01","paper":"/paper/ictcas-ucas-tal-submission-to-the-ava","paper_url":"http://research.google.com/ava/2021/S1_ICTCAS-UCAS-TAL.pdf","paper_title":"ICTCAS-UCAS-TAL Submission to the AVA-ActiveSpeaker Task at ActivityNet Challenge 2021","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":10,"model":"ASDNet","metrics":{"validation mean average precision":"93.5%"},"uses_additional_data":false,"paper_date":"2021-06-07","paper":"/paper/how-to-design-a-three-stage-architecture-for","paper_url":"https://arxiv.org/abs/2106.03932v2","paper_title":"How to Design a Three-Stage Architecture for Audio-Visual Active Speaker Detection in the Wild","code":"https://github.com/okankop/ASDNet","n_code_links":1,"syntology":null},{"rank_in_archive_order":11,"model":"GSCMIA","metrics":{"validation mean average precision":"92.86%"},"uses_additional_data":false,"paper_date":"2022-12-01","paper":"/paper/audio-visual-activity-guided-cross-modal","paper_url":"https://arxiv.org/abs/2212.00539v1","paper_title":"Audio-Visual Activity Guided Cross-Modal Identity Association for Active Speaker Detection","code":"https://github.com/rash1993/movie-asd","n_code_links":1,"syntology":null},{"rank_in_archive_order":12,"model":"TalkNet","metrics":{"validation mean average precision":"92.3%"},"uses_additional_data":false,"paper_date":"2021-06-01","paper":"/paper/nus-hlt-report-for-activitynet-challenge-2021","paper_url":"http://research.google.com/ava/2021/S3_NUS_Report_AVA_ActiveSpeaker_2021.pdf","paper_title":"NUS-HLT Report for ActivityNet Challenge 2021 AVA (Speaker)","code":"https://github.com/TaoRuijie/TalkNet_ASD","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"UniCon","metrics":{"validation mean average precision":"92.0%"},"uses_additional_data":false,"paper_date":"2021-08-05","paper":"/paper/unicon-unified-context-network-for-robust","paper_url":"https://arxiv.org/abs/2108.02607v1","paper_title":"UniCon: Unified Context Network for Robust Active Speaker Detection","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":14,"model":"SA-uncertainty Fusion","metrics":{"validation mean average precision":"91.9%"},"uses_additional_data":false,"paper_date":"2021-06-07","paper":"/paper/active-speaker-detection-as-a-multi-objective","paper_url":"https://arxiv.org/abs/2106.03821v2","paper_title":"Active Speaker Detection as a Multi-Objective Optimization with Uncertainty-based Multimodal Fusion","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":15,"model":"VTP (visual only)","metrics":{"validation mean average precision":"89.2%"},"uses_additional_data":true,"paper_date":"2021-10-14","paper":"/paper/sub-word-level-lip-reading-with-visual","paper_url":"https://arxiv.org/abs/2110.07603v2","paper_title":"Sub-word Level Lip Reading With Visual Attention","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"MAAS-TAN","metrics":{"validation mean average precision":"88.8%"},"uses_additional_data":false,"paper_date":"2021-01-11","paper":"/paper/maas-multi-modal-assignation-for-active","paper_url":"https://arxiv.org/abs/2101.03682v2","paper_title":"MAAS: Multi-modal Assignation for Active Speaker Detection","code":"https://github.com/fuankarion/maas","n_code_links":1,"syntology":null},{"rank_in_archive_order":17,"model":"VGG-{LSTM+TCN} (ensemble)","metrics":{"validation mean average precision":"87.8%"},"uses_additional_data":true,"paper_date":"2019-06-25","paper":"/paper/naver-at-activitynet-challenge-2019-task-b","paper_url":"https://arxiv.org/abs/1906.10555v1","paper_title":"Naver at ActivityNet Challenge 2019 -- Task B Active Speaker Detection (AVA)","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":18,"model":"Active Speakers in Context","metrics":{"validation mean average precision":"87.1%"},"uses_additional_data":false,"paper_date":"2020-05-20","paper":"/paper/active-speakers-in-context","paper_url":"https://arxiv.org/abs/2005.09812v1","paper_title":"Active Speakers in Context","code":"https://github.com/fuankarion/active-speakers-context","n_code_links":1,"syntology":null},{"rank_in_archive_order":19,"model":"MAAS-LAN","metrics":{"validation mean average precision":"85.1%"},"uses_additional_data":false,"paper_date":"2021-01-11","paper":"/paper/maas-multi-modal-assignation-for-active","paper_url":"https://arxiv.org/abs/2101.03682v2","paper_title":"MAAS: Multi-modal Assignation for Active Speaker Detection","code":"https://github.com/fuankarion/maas","n_code_links":1,"syntology":null},{"rank_in_archive_order":20,"model":"3D-ResNet-GRU","metrics":{"validation mean average precision":"84.0%"},"uses_additional_data":true,"paper_date":"2019-06-01","paper":"/paper/multi-task-learning-for-audio-visual-active","paper_url":"http://research.google.com/ava/2019/Multi_Task_Learning_for_Audio_Visual_Active_Speaker_Detection.pdf","paper_title":"Multi-Task Learning for Audio Visual Active Speaker Detection","code":null,"n_code_links":0,"syntology":null}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 6,264 of the 9,581 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9581,"papers_checked":6264,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":3316},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":2,"rows_with_any_sample_ran":2,"distinct_papers_with_graph_line":2,"distinct_papers_with_any_sample_ran":2,"samples_over_distinct_papers":{"n_ran":8,"n_unverified":3,"n_samples":11,"n_pointer_only_licence":5,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":8,"n_unverified":3,"n_samples":11,"n_pointer_only_licence":5,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}