{"url":"/sota/keypoint-detection-on-coco-test-challenge","task":{"name":"Keypoint Detection","url":"/task/keypoint-detection","note":null},"dataset":{"name":"COCO test-challenge","url":"/dataset/coco"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"**Keypoint Detection** is essential for analyzing and interpreting images in computer vision. It involves simultaneously detecting and localizing interesting points in an image. Keypoints, also known as interest points, are spatial locations or points in the image that define what is interesting or what stands out. They are invariant to image rotation, shrinkage, translation, distortion, etc. Keypoints examples are body joints, facial landmarks, or any other salient points in objects. Keypoints have uses in problems such as pose estimation, object detection and tracking, facial analysis, and augmented reality.\r\n\r\n<span style=\"color:grey; opacity: 0.6\">( Image credit: [PifPaf: Composite Fields for Human Pose Estimation](https://github.com/vita-epfl/openpifpaf); \"Learning to surf\" by fotologic, license: CC-BY-2.0 )</span>","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":["AR","ARM","AP","AP50","AP75","APL","AR50","AR75","ARL"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"AR":null,"ARM":null,"AP":"higher","AP50":null,"AP75":null,"APL":null,"AR50":null,"AR75":null,"ARL":null}},"counts":{"rows":8,"rows_with_code":7,"rows_with_paper_page":8,"rows_dated":8,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"4×RSN-50","metrics":{"AP":"77.1","AP50":"93.3","AP75":"83.6","APL":"82.6","AR":"82.6","AR50":"96.1","AR75":"88.2","ARL":"88.7","ARM":"78.0"},"uses_additional_data":false,"paper_date":"2020-03-09","paper":"/paper/learning-delicate-local-representations-for","paper_url":"https://arxiv.org/abs/2003.04030v3","paper_title":"Learning Delicate Local Representations for Multi-Person Pose Estimation","code":"https://github.com/open-mmlab/mmpose","n_code_links":4,"syntology":{"n_ran":2,"n_unverified":1,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":2,"model":"MSPN+*","metrics":{"AP":"76.4","AP50":"92.9","AP75":"82.6","APL":"88.6","AR":"82.2","AR50":"96","AR75":"87.7","ARL":"83.2","ARM":"77.5"},"uses_additional_data":false,"paper_date":"2019-01-01","paper":"/paper/rethinking-on-multi-stage-networks-for-human","paper_url":"https://arxiv.org/abs/1901.00148v4","paper_title":"Rethinking on Multi-Stage Networks for Human Pose Estimation","code":"https://github.com/open-mmlab/mmpose","n_code_links":7,"syntology":null},{"rank_in_archive_order":3,"model":"Simple Base+*","metrics":{"AP":"74.5","AP50":"90.9","AP75":"80.8","APL":"87.5","AR":"80.5","AR50":"95.1","AR75":"86.3","ARL":"82.9","ARM":"75.3"},"uses_additional_data":false,"paper_date":"2018-04-17","paper":"/paper/simple-baselines-for-human-pose-estimation","paper_url":"http://arxiv.org/abs/1804.06208v2","paper_title":"Simple Baselines for Human Pose Estimation and Tracking","code":"https://github.com/PaddlePaddle/PaddleDetection","n_code_links":27,"syntology":null},{"rank_in_archive_order":4,"model":"CPN+","metrics":{"AP":"72.1","AP50":"90.5","AP75":"78.9","APL":"84.7","AR":"78.7","AR50":"94.7","AR75":"84.8","ARL":"78.1","ARM":"74.3"},"uses_additional_data":false,"paper_date":"2017-11-20","paper":"/paper/cascaded-pyramid-network-for-multi-person","paper_url":"http://arxiv.org/abs/1711.07319v2","paper_title":"Cascaded Pyramid Network for Multi-Person Pose Estimation","code":"https://github.com/chenyilun95/tf-cpn","n_code_links":5,"syntology":null},{"rank_in_archive_order":5,"model":"Mask R-CNN*","metrics":{"AP":"68.9","AP50":"89.2","AP75":"75.2","APL":"82.6","AR":"75.4","AR50":"93.2","AR75":"81.2","ARL":"76.8","ARM":"70.2"},"uses_additional_data":false,"paper_date":"2017-03-20","paper":"/paper/mask-r-cnn","paper_url":"http://arxiv.org/abs/1703.06870v3","paper_title":"Mask R-CNN","code":"https://github.com/tensorflow/models/tree/master/official/vision","n_code_links":179,"syntology":{"n_ran":42,"n_unverified":98,"n_samples":140,"n_pointer_only_licence":23}},{"rank_in_archive_order":6,"model":"G-RMI*","metrics":{"AP":"69.1","AP50":"85.9","AP75":"75.2","APL":"82.4","AR":"75.1","AR50":"90.7","AR75":"80.7","ARL":"74.5","ARM":"69.7"},"uses_additional_data":false,"paper_date":"2017-01-06","paper":"/paper/towards-accurate-multi-person-pose-estimation","paper_url":"http://arxiv.org/abs/1701.01779v2","paper_title":"Towards Accurate Multi-person Pose Estimation in the Wild","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":7,"model":"Girdhar et al.","metrics":{"AR":"70.2","ARM":"60.7"},"uses_additional_data":false,"paper_date":"2017-12-26","paper":"/paper/detect-and-track-efficient-pose-estimation-in","paper_url":"http://arxiv.org/abs/1712.09184v2","paper_title":"Detect-and-Track: Efficient Pose Estimation in Videos","code":"https://github.com/facebookresearch/DetectAndTrack","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"Xiu et al.","metrics":{"AR":"67.5","ARM":"62.5"},"uses_additional_data":false,"paper_date":"2018-02-03","paper":"/paper/pose-flow-efficient-online-pose-tracking","paper_url":"http://arxiv.org/abs/1802.00977v2","paper_title":"Pose Flow: Efficient Online Pose Tracking","code":"https://github.com/YuliangXiu/PoseFlow","n_code_links":1,"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":44,"n_unverified":99,"n_samples":143,"n_pointer_only_licence":23,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":44,"n_unverified":99,"n_samples":143,"n_pointer_only_licence":23,"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"}}}