{"url":"/sota/keypoint-detection-on-coco","task":{"name":"Keypoint Detection","url":"/task/keypoint-detection","note":null},"dataset":{"name":"COCO (Common Objects in Context)","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":["Test AP","Validation AP","FPS"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Test AP":"higher","Validation AP":"higher","FPS":null}},"counts":{"rows":24,"rows_with_code":24,"rows_with_paper_page":24,"rows_dated":24,"rows_using_additional_data":5},"rows":[{"rank_in_archive_order":1,"model":"4xRSN-50(384×288)","metrics":{"Test AP":"78.6"},"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":"Poseur(384x288)","metrics":{"Test AP":"78.3","Validation AP":"79.6"},"uses_additional_data":false,"paper_date":"2022-01-19","paper":"/paper/poseur-direct-human-pose-regression-with","paper_url":"https://arxiv.org/abs/2201.07412v2","paper_title":"Poseur: Direct Human Pose Regression with Transformers","code":"https://github.com/aim-uofa/poseur","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"EvoPose2D-L(512x384)","metrics":{"Test AP":"76.8","Validation AP":"77.5"},"uses_additional_data":false,"paper_date":"2020-11-17","paper":"/paper/evopose2d-pushing-the-boundaries-of-2d-human","paper_url":"https://arxiv.org/abs/2011.08446v2","paper_title":"EvoPose2D: Pushing the Boundaries of 2D Human Pose Estimation using Accelerated Neuroevolution with Weight Transfer","code":"https://github.com/wmcnally/evopose2d","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":4,"model":"PoseFix(384x288)","metrics":{"Test AP":"76.7","Validation AP":"77.3"},"uses_additional_data":false,"paper_date":"2018-12-10","paper":"/paper/posefix-model-agnostic-general-human-pose","paper_url":"http://arxiv.org/abs/1812.03595v3","paper_title":"PoseFix: Model-agnostic General Human Pose Refinement Network","code":"https://github.com/mks0601/PoseFix_RELEASE","n_code_links":1,"syntology":null},{"rank_in_archive_order":5,"model":"DarkPose(384x288)","metrics":{"Test AP":"76.2"},"uses_additional_data":false,"paper_date":"2019-10-14","paper":"/paper/distribution-aware-coordinate-representation","paper_url":"https://arxiv.org/abs/1910.06278v1","paper_title":"Distribution-Aware Coordinate Representation for Human Pose Estimation","code":"https://github.com/PaddlePaddle/PaddleDetection","n_code_links":6,"syntology":{"n_ran":5,"n_unverified":4,"n_samples":9,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"MSPN(384x288)","metrics":{"Test AP":"76.1"},"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":7,"model":"AggPose(256x192)","metrics":{"Test AP":"75.7","Validation AP":"76.4"},"uses_additional_data":false,"paper_date":"2022-05-11","paper":"/paper/aggpose-deep-aggregation-vision-transformer","paper_url":"https://arxiv.org/abs/2205.05277v2","paper_title":"AggPose: Deep Aggregation Vision Transformer for Infant Pose Estimation","code":"https://github.com/szar-lab/aggpose","n_code_links":1,"syntology":{"n_ran":5,"n_unverified":2,"n_samples":7,"n_pointer_only_licence":7}},{"rank_in_archive_order":8,"model":"MIPNet(384x288)","metrics":{"Test AP":"75.7","Validation AP":"76.3"},"uses_additional_data":false,"paper_date":"2021-01-27","paper":"/paper/multi-hypothesis-pose-networks-rethinking-top","paper_url":"https://arxiv.org/abs/2101.11223v3","paper_title":"Multi-Instance Pose Networks: Rethinking Top-Down Pose Estimation","code":"https://github.com/rawalkhirodkar/MIPNet","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":9,"model":"HRNet-48(384x288)","metrics":{"Test AP":"75.5","Validation AP":"76.3"},"uses_additional_data":false,"paper_date":"2019-02-25","paper":"/paper/deep-high-resolution-representation-learning","paper_url":"http://arxiv.org/abs/1902.09212v1","paper_title":"Deep High-Resolution Representation Learning for Human Pose Estimation","code":"https://github.com/open-mmlab/mmdetection","n_code_links":39,"syntology":{"n_ran":8,"n_unverified":17,"n_samples":25,"n_pointer_only_licence":0}},{"rank_in_archive_order":10,"model":"TransPose(256x192)","metrics":{"Test AP":"75.0","Validation AP":"75.8"},"uses_additional_data":false,"paper_date":"2020-12-28","paper":"/paper/transpose-towards-explainable-human-pose","paper_url":"https://arxiv.org/abs/2012.14214v5","paper_title":"TransPose: Keypoint Localization via Transformer","code":"https://github.com/yangsenius/TransPose","n_code_links":1,"syntology":null},{"rank_in_archive_order":11,"model":"AlphaPose","metrics":{"FPS":"23","Test AP":"73.3"},"uses_additional_data":true,"paper_date":"2016-12-01","paper":"/paper/rmpe-regional-multi-person-pose-estimation","paper_url":"http://arxiv.org/abs/1612.00137v5","paper_title":"RMPE: Regional Multi-person Pose Estimation","code":"https://github.com/MVIG-SJTU/AlphaPose","n_code_links":14,"syntology":null},{"rank_in_archive_order":12,"model":"CPN+","metrics":{"Test AP":"73.0"},"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":13,"model":"PNFS","metrics":{"Test AP":"70.9"},"uses_additional_data":false,"paper_date":"2019-09-16","paper":"/paper/pose-neural-fabrics-search","paper_url":"https://arxiv.org/abs/1909.07068v4","paper_title":"Pose Neural Fabrics Search","code":"https://github.com/yangsenius/Pose-Neural-Fabrics-Search","n_code_links":2,"syntology":null},{"rank_in_archive_order":14,"model":"PersonLab","metrics":{"Test AP":"66.5"},"uses_additional_data":false,"paper_date":"2018-03-22","paper":"/paper/personlab-person-pose-estimation-and-instance","paper_url":"http://arxiv.org/abs/1803.08225v1","paper_title":"PersonLab: Person Pose Estimation and Instance Segmentation with a Bottom-Up, Part-Based, Geometric Embedding Model","code":"https://github.com/google-coral/project-posenet","n_code_links":3,"syntology":null},{"rank_in_archive_order":15,"model":"Pose-AE","metrics":{"Test AP":"62.8"},"uses_additional_data":false,"paper_date":"2016-11-16","paper":"/paper/associative-embedding-end-to-end-learning-for","paper_url":"http://arxiv.org/abs/1611.05424v2","paper_title":"Associative Embedding: End-to-End Learning for Joint Detection and Grouping","code":"https://github.com/open-mmlab/mmpose","n_code_links":5,"syntology":{"n_ran":0,"n_unverified":8,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":16,"model":"Sapiens-2B","metrics":{"Validation AP":"82.2"},"uses_additional_data":true,"paper_date":"2024-08-22","paper":"/paper/sapiens-foundation-for-human-vision-models","paper_url":"https://arxiv.org/abs/2408.12569v3","paper_title":"Sapiens: Foundation for Human Vision Models","code":"https://github.com/facebookresearch/sapiens","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":11,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":17,"model":"Sapiens-1B","metrics":{"Validation AP":"82.1"},"uses_additional_data":true,"paper_date":"2024-08-22","paper":"/paper/sapiens-foundation-for-human-vision-models","paper_url":"https://arxiv.org/abs/2408.12569v3","paper_title":"Sapiens: Foundation for Human Vision Models","code":"https://github.com/facebookresearch/sapiens","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":11,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":18,"model":"Sapiens-0.6B","metrics":{"Validation AP":"81.2"},"uses_additional_data":true,"paper_date":"2024-08-22","paper":"/paper/sapiens-foundation-for-human-vision-models","paper_url":"https://arxiv.org/abs/2408.12569v3","paper_title":"Sapiens: Foundation for Human Vision Models","code":"https://github.com/facebookresearch/sapiens","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":11,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"Sapiens-0.3B","metrics":{"Validation AP":"79.6"},"uses_additional_data":true,"paper_date":"2024-08-22","paper":"/paper/sapiens-foundation-for-human-vision-models","paper_url":"https://arxiv.org/abs/2408.12569v3","paper_title":"Sapiens: Foundation for Human Vision Models","code":"https://github.com/facebookresearch/sapiens","n_code_links":2,"syntology":{"n_ran":0,"n_unverified":11,"n_samples":11,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"UDP-Pose-PSA(384x288)","metrics":{"Validation AP":"79.5"},"uses_additional_data":false,"paper_date":"2021-07-02","paper":"/paper/polarized-self-attention-towards-high-quality-1","paper_url":"https://arxiv.org/abs/2107.00782v2","paper_title":"Polarized Self-Attention: Towards High-quality Pixel-wise Regression","code":"https://github.com/PaddlePaddle/PaddleSeg","n_code_links":6,"syntology":null},{"rank_in_archive_order":21,"model":"HRNet-32","metrics":{"Validation AP":"75.8"},"uses_additional_data":false,"paper_date":"2019-02-25","paper":"/paper/deep-high-resolution-representation-learning","paper_url":"http://arxiv.org/abs/1902.09212v1","paper_title":"Deep High-Resolution Representation Learning for Human Pose Estimation","code":"https://github.com/open-mmlab/mmdetection","n_code_links":39,"syntology":{"n_ran":8,"n_unverified":17,"n_samples":25,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"ResNet-50","metrics":{"Validation AP":"72.2"},"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":23,"model":"Pose Residual Network","metrics":{"Validation AP":"69.6"},"uses_additional_data":false,"paper_date":"2018-07-11","paper":"/paper/multiposenet-fast-multi-person-pose","paper_url":"http://arxiv.org/abs/1807.04067v1","paper_title":"MultiPoseNet: Fast Multi-Person Pose Estimation using Pose Residual Network","code":"https://github.com/salihkaragoz/pose-residual-network-pytorch","n_code_links":4,"syntology":null},{"rank_in_archive_order":24,"model":"Mask R-CNN + NL blocks (4 in head, 1 in backbone)","metrics":{"Validation AP":"66.5"},"uses_additional_data":false,"paper_date":"2017-11-21","paper":"/paper/non-local-neural-networks","paper_url":"http://arxiv.org/abs/1711.07971v3","paper_title":"Non-local Neural Networks","code":"https://github.com/facebookresearch/detectron","n_code_links":32,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}}],"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 6,795 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":6795,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2785},"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":13,"rows_with_any_sample_ran":8,"distinct_papers_with_graph_line":9,"distinct_papers_with_any_sample_ran":7,"samples_over_distinct_papers":{"n_ran":25,"n_unverified":44,"n_samples":69,"n_pointer_only_licence":11,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":33,"n_unverified":94,"n_samples":127,"n_pointer_only_licence":11,"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"}}}