{"url":"/sota/action-recognition-on-epic-kitchens-100","task":{"name":"Action Recognition","url":"/task/action-recognition-in-videos","note":null},"dataset":{"name":"EPIC-KITCHENS-100","url":"/dataset/epic-kitchens-100"},"category":"Computer Vision","categories":["Computer Vision","Time Series"],"category_note":null,"description":"**Action Recognition** is a computer vision task that involves recognizing human actions in videos or images. The goal is to classify and categorize the actions being performed in the video or image into a predefined set of action classes.\r\n\r\nIn the video domain, it is an open question whether training an action classification network on a sufficiently large dataset, will give a similar boost in performance when applied to a different temporal task or dataset. The challenges of building video datasets has meant that most popular benchmarks for action recognition are small, having on the order of 10k videos. \r\n\r\nPlease note some benchmarks may be located in the [Action Classification](https://paperswithcode.com/task/action-classification) or [Video Classification](https://paperswithcode.com/task/video-classification) tasks, e.g. Kinetics-400.","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":["Action@1","Verb@1","Noun@1","GFLOPs"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Action@1":null,"Verb@1":null,"Noun@1":null,"GFLOPs":null}},"counts":{"rows":32,"rows_with_code":28,"rows_with_paper_page":32,"rows_dated":32,"rows_using_additional_data":15},"rows":[{"rank_in_archive_order":1,"model":"LLaVAction","metrics":{"Action@1":"58.3","Noun@1":"69","Verb@1":"76"},"uses_additional_data":true,"paper_date":"2025-03-24","paper":"/paper/llavaction-evaluating-and-training-multi","paper_url":"https://arxiv.org/abs/2503.18712v1","paper_title":"LLaVAction: evaluating and training multi-modal large language models for action recognition","code":"https://github.com/adaptivemotorcontrollab/llavaction","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"TIM","metrics":{"Action@1":"56.4","Noun@1":"66.4","Verb@1":"76.2"},"uses_additional_data":true,"paper_date":"2024-04-08","paper":"/paper/tim-a-time-interval-machine-for-audio-visual","paper_url":"https://arxiv.org/abs/2404.05559v2","paper_title":"TIM: A Time Interval Machine for Audio-Visual Action Recognition","code":"https://github.com/jacobchalk/tim","n_code_links":1,"syntology":{"n_ran":11,"n_unverified":1,"n_samples":12,"n_pointer_only_licence":12}},{"rank_in_archive_order":3,"model":"Avion (ViT-L)","metrics":{"Action@1":"54.4","Noun@1":"65.4","Verb@1":"73.0"},"uses_additional_data":true,"paper_date":"2023-09-28","paper":"/paper/training-a-large-video-model-on-a-single","paper_url":"https://arxiv.org/abs/2309.16669v1","paper_title":"Training a Large Video Model on a Single Machine in a Day","code":"https://github.com/zhaoyue-zephyrus/avion","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"M&M (WTS 60M)","metrics":{"Action@1":"53.6","Noun@1":"66.3","Verb@1":"72.0"},"uses_additional_data":true,"paper_date":"2022-06-20","paper":"/paper/m-m-mix-a-multimodal-multiview-transformer","paper_url":"https://arxiv.org/abs/2206.09852v1","paper_title":"M&M Mix: A Multimodal Multiview Transformer Ensemble","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"LVMAE","metrics":{"Action@1":"52.1","Noun@1":"61.8","Verb@1":"75.0"},"uses_additional_data":true,"paper_date":"2024-11-20","paper":"/paper/extending-video-masked-autoencoders-to-128-1","paper_url":"https://arxiv.org/abs/2411.13683v1","paper_title":"Extending Video Masked Autoencoders to 128 frames","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":6,"model":"TAdaFormer-L/14","metrics":{"Action@1":"51.8","Noun@1":"64.1","Verb@1":"71.7"},"uses_additional_data":true,"paper_date":"2023-08-10","paper":"/paper/temporally-adaptive-models-for-efficient","paper_url":"https://arxiv.org/abs/2308.05787v1","paper_title":"Temporally-Adaptive Models for Efficient Video Understanding","code":"https://github.com/alibaba-mmai-research/TAdaConv","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":7,"model":"LaViLa (TimeSformer-L)","metrics":{"Action@1":"51","Noun@1":"62.9","Verb@1":"72"},"uses_additional_data":true,"paper_date":"2022-12-08","paper":"/paper/learning-video-representations-from-large","paper_url":"https://arxiv.org/abs/2212.04501v1","paper_title":"Learning Video Representations from Large Language Models","code":"https://github.com/facebookresearch/lavila","n_code_links":3,"syntology":{"n_ran":13,"n_unverified":7,"n_samples":20,"n_pointer_only_licence":20}},{"rank_in_archive_order":8,"model":"MTV-B (WTS 60M)","metrics":{"Action@1":"50.5","Noun@1":"63.9","Verb@1":"69.9"},"uses_additional_data":true,"paper_date":"2022-01-12","paper":"/paper/multiview-transformers-for-video-recognition","paper_url":"https://arxiv.org/abs/2201.04288v4","paper_title":"Multiview Transformers for Video Recognition","code":"https://github.com/google-research/scenic","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"OMNIVORE (Swin-B, finetuned)","metrics":{"Action@1":"49.9","Noun@1":"61.7","Verb@1":"69.5"},"uses_additional_data":true,"paper_date":"2022-01-20","paper":"/paper/omnivore-a-single-model-for-many-visual","paper_url":"https://arxiv.org/abs/2201.08377v2","paper_title":"Omnivore: A Single Model for Many Visual Modalities","code":"https://github.com/towhee-io/towhee","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":1}},{"rank_in_archive_order":10,"model":"CAST(ViT-B/16)","metrics":{"Action@1":"49.3","Noun@1":"60.9","Verb@1":"72.5"},"uses_additional_data":false,"paper_date":"2023-11-30","paper":"/paper/cast-cross-attention-in-space-and-time-for-1","paper_url":"https://arxiv.org/abs/2311.18825v2","paper_title":"CAST: Cross-Attention in Space and Time for Video Action Recognition","code":"https://github.com/khu-vll/cast","n_code_links":1,"syntology":{"n_ran":9,"n_unverified":8,"n_samples":17,"n_pointer_only_licence":17}},{"rank_in_archive_order":11,"model":"TAdaConvNeXtV2-S","metrics":{"Action@1":"48.9","Noun@1":"60.2","Verb@1":"71.0"},"uses_additional_data":true,"paper_date":"2023-08-10","paper":"/paper/temporally-adaptive-models-for-efficient","paper_url":"https://arxiv.org/abs/2308.05787v1","paper_title":"Temporally-Adaptive Models for Efficient Video Understanding","code":"https://github.com/alibaba-mmai-research/TAdaConv","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":12,"model":"MeMViT-24","metrics":{"Action@1":"48.4","Noun@1":"60.3","Verb@1":"71.4"},"uses_additional_data":true,"paper_date":"2022-01-20","paper":"/paper/memvit-memory-augmented-multiscale-vision","paper_url":"https://arxiv.org/abs/2201.08383v2","paper_title":"MeMViT: Memory-Augmented Multiscale Vision Transformer for Efficient Long-Term Video Recognition","code":"https://github.com/facebookresearch/memvit","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"MMT","metrics":{"Action@1":"47.8","Noun@1":"61.0","Verb@1":"70.1"},"uses_additional_data":false,"paper_date":"2022-09-22","paper":"/paper/multiscale-multimodal-transformer-for","paper_url":"https://openreview.net/pdf?id=aqP3WFwMPbe","paper_title":"Multiscale Multimodal Transformer for Multimodal Action Recognition","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":14,"model":"MoViNet-A6","metrics":{"Action@1":"47.7","GFLOPs":"117x1","Noun@1":"57.3","Verb@1":"72.2"},"uses_additional_data":false,"paper_date":"2021-03-21","paper":"/paper/movinets-mobile-video-networks-for-efficient","paper_url":"https://arxiv.org/abs/2103.11511v2","paper_title":"MoViNets: Mobile Video Networks for Efficient Video Recognition","code":"https://github.com/tensorflow/models","n_code_links":3,"syntology":{"n_ran":9,"n_unverified":4,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":15,"model":"AVT","metrics":{"Action@1":"47.2","Noun@1":"59.3","Verb@1":"70.4"},"uses_additional_data":false,"paper_date":"2022-09-22","paper":"/paper/avt-audio-video-transformer-for-multimodal","paper_url":"https://openreview.net/pdf?id=yFuHxmSwGus","paper_title":"AVT: Audio-Video Transformer for Multimodal Action Recognition","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"ORViT Mformer-L (ORViT blocks)","metrics":{"Action@1":"45.7","Noun@1":"58.7","Verb@1":"68.4"},"uses_additional_data":false,"paper_date":"2021-10-13","paper":"/paper/object-region-video-transformers-1","paper_url":"https://arxiv.org/abs/2110.06915v3","paper_title":"Object-Region Video Transformers","code":"https://github.com/eladb3/orvit","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":2,"n_samples":7,"n_pointer_only_licence":7}},{"rank_in_archive_order":17,"model":"TempAgg","metrics":{"Action@1":"45.26","Noun@1":"53.35","Verb@1":"66"},"uses_additional_data":false,"paper_date":"2021-06-06","paper":"/paper/technical-report-temporal-aggregate","paper_url":"https://arxiv.org/abs/2106.03152v2","paper_title":"Technical Report: Temporal Aggregate Representations","code":"https://github.com/dibschat/tempAgg","n_code_links":1,"syntology":null},{"rank_in_archive_order":18,"model":"MoViNet-A5","metrics":{"Action@1":"44.5","GFLOPs":"74.9x1","Noun@1":"55.1","Verb@1":"69.1"},"uses_additional_data":false,"paper_date":"2021-03-21","paper":"/paper/movinets-mobile-video-networks-for-efficient","paper_url":"https://arxiv.org/abs/2103.11511v2","paper_title":"MoViNets: Mobile Video Networks for Efficient Video Recognition","code":"https://github.com/tensorflow/models","n_code_links":3,"syntology":{"n_ran":9,"n_unverified":4,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"Mformer-HR","metrics":{"Action@1":"44.5","Noun@1":"58.5","Verb@1":"67.0"},"uses_additional_data":true,"paper_date":"2021-06-09","paper":"/paper/keeping-your-eye-on-the-ball-trajectory","paper_url":"https://arxiv.org/abs/2106.05392v2","paper_title":"Keeping Your Eye on the Ball: Trajectory Attention in Video Transformers","code":"https://github.com/facebookresearch/xformers","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":1,"n_samples":2,"n_pointer_only_licence":2}},{"rank_in_archive_order":20,"model":"GSF","metrics":{"Action@1":"44.48","Noun@1":"53.18","Verb@1":"69.06"},"uses_additional_data":true,"paper_date":"2022-03-16","paper":"/paper/gate-shift-fuse-for-video-action-recognition","paper_url":"https://arxiv.org/abs/2203.08897v3","paper_title":"Gate-Shift-Fuse for Video Action Recognition","code":"https://github.com/swathikirans/gsf","n_code_links":2,"syntology":null},{"rank_in_archive_order":21,"model":"MoViNet-A4","metrics":{"Action@1":"44.4","GFLOPs":"42.2x1","Noun@1":"56.2","Verb@1":"68.8"},"uses_additional_data":false,"paper_date":"2021-03-21","paper":"/paper/movinets-mobile-video-networks-for-efficient","paper_url":"https://arxiv.org/abs/2103.11511v2","paper_title":"MoViNets: Mobile Video Networks for Efficient Video Recognition","code":"https://github.com/tensorflow/models","n_code_links":3,"syntology":{"n_ran":9,"n_unverified":4,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"Mformer-L","metrics":{"Action@1":"44.1","Noun@1":"57.6","Verb@1":"67.1"},"uses_additional_data":true,"paper_date":"2021-06-09","paper":"/paper/keeping-your-eye-on-the-ball-trajectory","paper_url":"https://arxiv.org/abs/2106.05392v2","paper_title":"Keeping Your Eye on the Ball: Trajectory Attention in Video Transformers","code":"https://github.com/facebookresearch/xformers","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":1,"n_samples":2,"n_pointer_only_licence":2}},{"rank_in_archive_order":23,"model":"ViViT-L/16x2 Fact. encoder","metrics":{"Action@1":"44.0","Noun@1":"56.8","Verb@1":"66.4"},"uses_additional_data":false,"paper_date":"2021-03-29","paper":"/paper/2103-15691","paper_url":"https://arxiv.org/abs/2103.15691v2","paper_title":"ViViT: A Video Vision Transformer","code":"https://github.com/google-research/scenic","n_code_links":10,"syntology":{"n_ran":14,"n_unverified":7,"n_samples":21,"n_pointer_only_licence":1}},{"rank_in_archive_order":24,"model":"MBT","metrics":{"Action@1":"43.4","Noun@1":"58","Verb@1":"64.8"},"uses_additional_data":false,"paper_date":"2021-06-30","paper":"/paper/attention-bottlenecks-for-multimodal-fusion","paper_url":"https://arxiv.org/abs/2107.00135v3","paper_title":"Attention Bottlenecks for Multimodal Fusion","code":"https://github.com/google-research/scenic/tree/main/scenic/projects/mbt","n_code_links":1,"syntology":null},{"rank_in_archive_order":25,"model":"Mformer","metrics":{"Action@1":"43.1","Noun@1":"56.5","Verb@1":"66.7"},"uses_additional_data":true,"paper_date":"2021-06-09","paper":"/paper/keeping-your-eye-on-the-ball-trajectory","paper_url":"https://arxiv.org/abs/2106.05392v2","paper_title":"Keeping Your Eye on the Ball: Trajectory Attention in Video Transformers","code":"https://github.com/facebookresearch/xformers","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":1,"n_samples":2,"n_pointer_only_licence":2}},{"rank_in_archive_order":26,"model":"MoViNet-A2","metrics":{"Action@1":"41.2","GFLOPs":"7.59x1","Noun@1":"52.3","Verb@1":"67.1"},"uses_additional_data":false,"paper_date":"2021-03-21","paper":"/paper/movinets-mobile-video-networks-for-efficient","paper_url":"https://arxiv.org/abs/2103.11511v2","paper_title":"MoViNets: Mobile Video Networks for Efficient Video Recognition","code":"https://github.com/tensorflow/models","n_code_links":3,"syntology":{"n_ran":9,"n_unverified":4,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":27,"model":"TSM","metrics":{"Action@1":"37.39"},"uses_additional_data":false,"paper_date":"2020-06-23","paper":"/paper/rescaling-egocentric-vision","paper_url":"https://arxiv.org/abs/2006.13256v4","paper_title":"Rescaling Egocentric Vision","code":"https://github.com/epic-kitchens/epic-kitchens-100-annotations","n_code_links":7,"syntology":{"n_ran":9,"n_unverified":9,"n_samples":18,"n_pointer_only_licence":0}},{"rank_in_archive_order":28,"model":"SlowFast","metrics":{"Action@1":"36.81"},"uses_additional_data":false,"paper_date":"2020-06-23","paper":"/paper/rescaling-egocentric-vision","paper_url":"https://arxiv.org/abs/2006.13256v4","paper_title":"Rescaling Egocentric Vision","code":"https://github.com/epic-kitchens/epic-kitchens-100-annotations","n_code_links":7,"syntology":{"n_ran":9,"n_unverified":9,"n_samples":18,"n_pointer_only_licence":0}},{"rank_in_archive_order":29,"model":"MoViNet-A0","metrics":{"Action@1":"36.8","GFLOPs":"1.74x1","Noun@1":"47.4","Verb@1":"64.8"},"uses_additional_data":false,"paper_date":"2021-03-21","paper":"/paper/movinets-mobile-video-networks-for-efficient","paper_url":"https://arxiv.org/abs/2103.11511v2","paper_title":"MoViNets: Mobile Video Networks for Efficient Video Recognition","code":"https://github.com/tensorflow/models","n_code_links":3,"syntology":{"n_ran":9,"n_unverified":4,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":30,"model":"TBN","metrics":{"Action@1":"35.55"},"uses_additional_data":false,"paper_date":"2020-06-23","paper":"/paper/rescaling-egocentric-vision","paper_url":"https://arxiv.org/abs/2006.13256v4","paper_title":"Rescaling Egocentric Vision","code":"https://github.com/epic-kitchens/epic-kitchens-100-annotations","n_code_links":7,"syntology":{"n_ran":9,"n_unverified":9,"n_samples":18,"n_pointer_only_licence":0}},{"rank_in_archive_order":31,"model":"TRN","metrics":{"Action@1":"35.28"},"uses_additional_data":false,"paper_date":"2020-06-23","paper":"/paper/rescaling-egocentric-vision","paper_url":"https://arxiv.org/abs/2006.13256v4","paper_title":"Rescaling Egocentric Vision","code":"https://github.com/epic-kitchens/epic-kitchens-100-annotations","n_code_links":7,"syntology":{"n_ran":9,"n_unverified":9,"n_samples":18,"n_pointer_only_licence":0}},{"rank_in_archive_order":32,"model":"TSN","metrics":{"Action@1":"33.57"},"uses_additional_data":false,"paper_date":"2020-06-23","paper":"/paper/rescaling-egocentric-vision","paper_url":"https://arxiv.org/abs/2006.13256v4","paper_title":"Rescaling Egocentric Vision","code":"https://github.com/epic-kitchens/epic-kitchens-100-annotations","n_code_links":7,"syntology":{"n_ran":9,"n_unverified":9,"n_samples":18,"n_pointer_only_licence":0}}],"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 6,821 of the 9,623 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":9623,"papers_checked":6821,"papers_extracted_not_yet_verified":65,"boards_without_verdict":29,"papers_not_yet_extracted":2737},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+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":21,"rows_with_any_sample_ran":21,"distinct_papers_with_graph_line":10,"distinct_papers_with_any_sample_ran":10,"samples_over_distinct_papers":{"n_ran":75,"n_unverified":39,"n_samples":114,"n_pointer_only_licence":60,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":151,"n_unverified":93,"n_samples":244,"n_pointer_only_licence":64,"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"}}}