{"url":"/sota/action-recognition-on-h2o-2-hands-and-objects","task":{"name":"Action Recognition","url":"/task/action-recognition-in-videos","note":null},"dataset":{"name":"H2O  (2 Hands and Objects)","url":"/dataset/h2o-dataset"},"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":["Actions Top-1","RGB","Hand Pose","Object Pose","Object Label"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Actions Top-1":"higher","RGB":null,"Hand Pose":null,"Object Pose":null,"Object Label":null}},"counts":{"rows":11,"rows_with_code":10,"rows_with_paper_page":11,"rows_dated":11,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"HandFormer-B/21x8","metrics":{"Actions Top-1":"93.39","Hand Pose":"3D","Object Label":"No","Object Pose":"No","RGB":"Yes"},"uses_additional_data":false,"paper_date":"2024-03-14","paper":"/paper/on-the-utility-of-3d-hand-poses-for-action","paper_url":"https://arxiv.org/abs/2403.09805v2","paper_title":"On the Utility of 3D Hand Poses for Action Recognition","code":"https://github.com/s-shamil/HandFormer","n_code_links":1,"syntology":{"n_ran":8,"n_unverified":5,"n_samples":13,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"SHARP","metrics":{"Actions Top-1":"91.73","Hand Pose":"3D","Object Label":"Yes","Object Pose":"2D","RGB":"No"},"uses_additional_data":false,"paper_date":"2024-08-19","paper":"/paper/sharp-segmentation-of-hands-and-arms-by-range","paper_url":"https://arxiv.org/abs/2408.10037v1","paper_title":"SHARP: Segmentation of Hands and Arms by Range using Pseudo-Depth for Enhanced Egocentric 3D Hand Pose Estimation and Action Recognition","code":"https://github.com/wiktormucha/SHARP","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"EffHandEgoNet","metrics":{"Actions Top-1":"91.32","Hand Pose":"2D","Object Label":"Yes","Object Pose":"2D","RGB":"No"},"uses_additional_data":false,"paper_date":"2024-04-14","paper":"/paper/in-my-perspective-in-my-hands-accurate","paper_url":"https://arxiv.org/abs/2404.09308v2","paper_title":"In My Perspective, In My Hands: Accurate Egocentric 2D Hand Pose and Action Recognition","code":"https://github.com/wiktormucha/effhandegonet","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":1,"n_samples":5,"n_pointer_only_licence":5}},{"rank_in_archive_order":4,"model":"H2OTR","metrics":{"Actions Top-1":"90.90","Hand Pose":"3D (est.)","Object Label":"Yes (est.)","Object Pose":"Yes (est.)","RGB":"Yes"},"uses_additional_data":false,"paper_date":"2023-01-01","paper":"/paper/transformer-based-unified-recognition-of-two","paper_url":"http://openaccess.thecvf.com//content/CVPR2023/html/Cho_Transformer-Based_Unified_Recognition_of_Two_Hands_Manipulating_Objects_CVPR_2023_paper.html","paper_title":"Transformer-Based Unified Recognition of Two Hands Manipulating Objects","code":"https://github.com/chohoseong/Transformer-based-Unified-Recognition-of-Two-Hands-Manipulating-Objects","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"ISTA-Net","metrics":{"Actions Top-1":"89.09","Hand Pose":"3D","Object Label":"No","Object Pose":"Yes","RGB":"No"},"uses_additional_data":false,"paper_date":"2023-07-14","paper":"/paper/interactive-spatiotemporal-token-attention","paper_url":"https://arxiv.org/abs/2307.07469v1","paper_title":"Interactive Spatiotemporal Token Attention Network for Skeleton-based General Interactive Action Recognition","code":"https://github.com/Necolizer/ISTA-Net","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"HTT","metrics":{"Actions Top-1":"86.36","Hand Pose":"3D (est.)","Object Label":"Yes (est.)","Object Pose":"No","RGB":"Yes"},"uses_additional_data":false,"paper_date":"2022-09-20","paper":"/paper/hierarchical-temporal-transformer-for-3d-hand","paper_url":"https://arxiv.org/abs/2209.09484v4","paper_title":"Hierarchical Temporal Transformer for 3D Hand Pose Estimation and Action Recognition from Egocentric RGB Videos","code":"https://github.com/fylwen/htt","n_code_links":1,"syntology":{"n_ran":7,"n_unverified":1,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":7,"model":"RGBPoseConv3D","metrics":{"Actions Top-1":"83.47","Hand Pose":"2D","Object Label":"No","Object Pose":"No","RGB":"Yes"},"uses_additional_data":false,"paper_date":"2021-04-28","paper":"/paper/revisiting-skeleton-based-action-recognition","paper_url":"https://arxiv.org/abs/2104.13586v2","paper_title":"Revisiting Skeleton-based Action Recognition","code":"https://github.com/open-mmlab/mmaction2","n_code_links":4,"syntology":null},{"rank_in_archive_order":8,"model":"TA-GCN","metrics":{"Actions Top-1":"79.25","Hand Pose":"3D","Object Label":"No","Object Pose":"Yes","RGB":"No"},"uses_additional_data":false,"paper_date":"2021-04-22","paper":"/paper/h2o-two-hands-manipulating-objects-for-first","paper_url":"https://arxiv.org/abs/2104.11181v2","paper_title":"H2O: Two Hands Manipulating Objects for First Person Interaction Recognition","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":9,"model":"SlowFast","metrics":{"Actions Top-1":"77.69","Hand Pose":"No","Object Label":"No","Object Pose":"No","RGB":"Yes"},"uses_additional_data":false,"paper_date":"2018-12-10","paper":"/paper/slowfast-networks-for-video-recognition","paper_url":"https://arxiv.org/abs/1812.03982v3","paper_title":"SlowFast Networks for Video Recognition","code":"https://github.com/facebookresearch/SlowFast","n_code_links":15,"syntology":{"n_ran":0,"n_unverified":10,"n_samples":10,"n_pointer_only_licence":0}},{"rank_in_archive_order":10,"model":"ST-GCN","metrics":{"Actions Top-1":"73.86","Hand Pose":"3D","Object Label":"No","Object Pose":"Yes","RGB":"No"},"uses_additional_data":false,"paper_date":"2018-01-23","paper":"/paper/spatial-temporal-graph-convolutional-networks-1","paper_url":"http://arxiv.org/abs/1801.07455v2","paper_title":"Spatial Temporal Graph Convolutional Networks for Skeleton-Based Action Recognition","code":"https://github.com/open-mmlab/mmskeleton","n_code_links":24,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"MS-G3D","metrics":{"Actions Top-1":"50.83","Hand Pose":"3D","Object Label":"No","Object Pose":"No","RGB":"No"},"uses_additional_data":false,"paper_date":"2020-03-31","paper":"/paper/disentangling-and-unifying-graph-convolutions","paper_url":"https://arxiv.org/abs/2003.14111v2","paper_title":"Disentangling and Unifying Graph Convolutions for Skeleton-Based Action Recognition","code":"https://github.com/kennymckormick/pyskl","n_code_links":3,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"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 6,885 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":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"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":7,"rows_with_any_sample_ran":6,"distinct_papers_with_graph_line":7,"distinct_papers_with_any_sample_ran":6,"samples_over_distinct_papers":{"n_ran":25,"n_unverified":17,"n_samples":42,"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":25,"n_unverified":17,"n_samples":42,"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"}}}