{"url":"/sota/head-pose-estimation-on-biwi","task":{"name":"Head Pose Estimation","url":"/task/head-pose-estimation","note":null},"dataset":{"name":"BIWI","url":"/dataset/biwi"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"Сделай так, что бы оба человека на фотографии смотрели в центр экрана телефона","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":["MAE (trained with other data)","MAE (trained with BIWI data)","Geodesic Error (GE)","MAE-aligned (trained with other data)","Geodesic Error - aligned (GE)","MAEV","MAE_t"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"MAE (trained with other data)":"lower","MAE (trained with BIWI data)":"lower","Geodesic Error (GE)":"lower","MAE-aligned (trained with other data)":"lower","Geodesic Error - aligned (GE)":"lower","MAEV":null,"MAE_t":"lower"}},"counts":{"rows":29,"rows_with_code":22,"rows_with_paper_page":29,"rows_dated":29,"rows_using_additional_data":4},"rows":[{"rank_in_archive_order":1,"model":"TRG (w/ 300WLP)","metrics":{"Geodesic Error - aligned (GE)":"5.68","MAE (trained with other data)":"2.75","MAE_t":"12.97"},"uses_additional_data":true,"paper_date":"2024-07-19","paper":"/paper/6dof-head-pose-estimation-through-explicit","paper_url":"https://arxiv.org/abs/2407.14136v1","paper_title":"6DoF Head Pose Estimation through Explicit Bidirectional Interaction with Face Geometry","code":"https://github.com/asw91666/trg-release","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"TRG","metrics":{"Geodesic Error - aligned (GE)":"5.35","MAE (trained with other data)":"2.89","MAE_t":"14.31"},"uses_additional_data":false,"paper_date":"2024-07-19","paper":"/paper/6dof-head-pose-estimation-through-explicit","paper_url":"https://arxiv.org/abs/2407.14136v1","paper_title":"6DoF Head Pose Estimation through Explicit Bidirectional Interaction with Face Geometry","code":"https://github.com/asw91666/trg-release","n_code_links":1,"syntology":null},{"rank_in_archive_order":3,"model":"PerspNet","metrics":{"Geodesic Error - aligned (GE)":"5.5","MAE (trained with other data)":"2.95","MAE_t":"19.09"},"uses_additional_data":false,"paper_date":"2022-05-09","paper":"/paper/single-image-3d-face-reconstruction-under","paper_url":"https://arxiv.org/abs/2205.04126v2","paper_title":"Towards 3D Face Reconstruction in Perspective Projection: Estimating 6DoF Face Pose from Monocular Image","code":"https://github.com/cbsropenproject/6dof_face","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"6DRepNet360","metrics":{"MAE (trained with other data)":"3.39","MAEV":"4.85"},"uses_additional_data":false,"paper_date":"2023-09-14","paper":"/paper/towards-robust-and-unconstrained-full-range","paper_url":"https://arxiv.org/abs/2309.07654v1","paper_title":"Towards Robust and Unconstrained Full Range of Rotation Head Pose Estimation","code":"https://github.com/thohemp/6drepnet360","n_code_links":2,"syntology":null},{"rank_in_archive_order":5,"model":"6DRepNet","metrics":{"MAE (trained with BIWI data)":"2.66","MAE (trained with other data)":"3.47"},"uses_additional_data":false,"paper_date":"2022-02-25","paper":"/paper/6d-rotation-representation-for-unconstrained","paper_url":"https://arxiv.org/abs/2202.12555v2","paper_title":"6D Rotation Representation For Unconstrained Head Pose Estimation","code":"https://github.com/thohemp/6drepnet","n_code_links":2,"syntology":{"n_ran":5,"n_unverified":3,"n_samples":8,"n_pointer_only_licence":3}},{"rank_in_archive_order":6,"model":"WHENet-V","metrics":{"MAE (trained with other data)":"3.48"},"uses_additional_data":false,"paper_date":"2020-05-20","paper":"/paper/whenet-real-time-fine-grained-estimation-for","paper_url":"https://arxiv.org/abs/2005.10353v2","paper_title":"WHENet: Real-time Fine-Grained Estimation for Wide Range Head Pose","code":"https://github.com/sizhean/panohead","n_code_links":4,"syntology":null},{"rank_in_archive_order":7,"model":"LSR","metrics":{"MAE (trained with other data)":"3.519"},"uses_additional_data":false,"paper_date":"2023-11-10","paper":"/paper/2d-image-head-pose-estimation-via-latent","paper_url":"https://arxiv.org/abs/2311.06038v1","paper_title":"2D Image head pose estimation via latent space regression under occlusion settings","code":"https://github.com/sipg-isr/Occlusion_HPE","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"OpNet","metrics":{"Geodesic Error (GE)":"7.01","Geodesic Error - aligned (GE)":"4.72","MAE (trained with other data)":"3.57","MAE-aligned (trained with other data)":"2.65"},"uses_additional_data":true,"paper_date":"2024-07-07","paper":"/paper/on-the-power-of-data-augmentation-for-head","paper_url":"https://arxiv.org/abs/2407.05357v3","paper_title":"On the power of data augmentation for head pose estimation","code":"https://github.com/opentrack/neuralnet-tracker-traincode","n_code_links":1,"syntology":null},{"rank_in_archive_order":9,"model":"MNN","metrics":{"MAE (trained with other data)":"3.66"},"uses_additional_data":false,"paper_date":"2020-12-22","paper":"/paper/multi-task-head-pose-estimation-in-the-wild-1","paper_url":"https://arxiv.org/abs/2202.02299v1","paper_title":"Multi-task head pose estimation in-the-wild","code":"https://github.com/bobetocalo/bobetocalo_pami20","n_code_links":1,"syntology":null},{"rank_in_archive_order":10,"model":"img2pose","metrics":{"Geodesic Error (GE)":"7.10","Geodesic Error - aligned (GE)":"6.23","MAE (trained with other data)":"3.786","MAE-aligned (trained with other data)":"3.4"},"uses_additional_data":false,"paper_date":"2020-12-14","paper":"/paper/img2pose-face-alignment-and-detection-via","paper_url":"https://arxiv.org/abs/2012.07791v2","paper_title":"img2pose: Face Alignment and Detection via 6DoF, Face Pose Estimation","code":"https://github.com/vitoralbiero/img2pose","n_code_links":2,"syntology":null},{"rank_in_archive_order":11,"model":"WHENet","metrics":{"MAE (trained with other data)":"3.81"},"uses_additional_data":false,"paper_date":"2020-05-20","paper":"/paper/whenet-real-time-fine-grained-estimation-for","paper_url":"https://arxiv.org/abs/2005.10353v2","paper_title":"WHENet: Real-time Fine-Grained Estimation for Wide Range Head Pose","code":"https://github.com/sizhean/panohead","n_code_links":4,"syntology":null},{"rank_in_archive_order":12,"model":"SRHP-6D","metrics":{"Geodesic Error (GE)":"7.30","Geodesic Error - aligned (GE)":"5.48","MAE (trained with other data)":"3.98","MAE-aligned (trained with other data)":"3.21"},"uses_additional_data":false,"paper_date":"2024-01-11","paper":"/paper/on-the-representation-and-methodology-for-1","paper_url":"https://arxiv.org/abs/2401.05807v1","paper_title":"On the representation and methodology for wide and short range head pose estimation","code":"https://github.com/pcr-upm/opal23_headpose","n_code_links":1,"syntology":null},{"rank_in_archive_order":13,"model":"FSA-Net (Caps-Fusion)","metrics":{"Geodesic Error (GE)":"7.64","Geodesic Error - aligned (GE)":"5.36","MAE (trained with other data)":"4.00","MAE-aligned (trained with other data)":"2.92"},"uses_additional_data":false,"paper_date":"2019-06-01","paper":"/paper/fsa-net-learning-fine-grained-structure","paper_url":"http://openaccess.thecvf.com/content_CVPR_2019/html/Yang_FSA-Net_Learning_Fine-Grained_Structure_Aggregation_for_Head_Pose_Estimation_From_CVPR_2019_paper.html","paper_title":"FSA-Net: Learning Fine-Grained Structure Aggregation for Head Pose Estimation From a Single Image","code":"https://github.com/shamangary/FSA-Net","n_code_links":1,"syntology":null},{"rank_in_archive_order":14,"model":"SRHP-Euler","metrics":{"Geodesic Error (GE)":"7.49","Geodesic Error - aligned (GE)":"5.42","MAE (trained with other data)":"4.13","MAE-aligned (trained with other data)":"3.16"},"uses_additional_data":false,"paper_date":"2024-01-11","paper":"/paper/on-the-representation-and-methodology-for-1","paper_url":"https://arxiv.org/abs/2401.05807v1","paper_title":"On the representation and methodology for wide and short range head pose estimation","code":"https://github.com/pcr-upm/opal23_headpose","n_code_links":1,"syntology":null},{"rank_in_archive_order":15,"model":"PADACO","metrics":{"MAE (trained with other data)":"4.13"},"uses_additional_data":true,"paper_date":"2019-10-01","paper":"/paper/deep-head-pose-estimation-using-synthetic","paper_url":"http://openaccess.thecvf.com/content_ICCV_2019/html/Kuhnke_Deep_Head_Pose_Estimation_Using_Synthetic_Images_and_Partial_Adversarial_ICCV_2019_paper.html","paper_title":"Deep Head Pose Estimation Using Synthetic Images and Partial Adversarial Domain Adaption for Continuous Label Spaces","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"DDD-Pose","metrics":{"MAE (trained with BIWI data)":"2.80","MAE (trained with other data)":"4.52"},"uses_additional_data":false,"paper_date":"2022-01-01","paper":"/paper/a-data-driven-approach-to-improve-3d-head","paper_url":"https://link.springer.com/chapter/10.1007/978-3-030-90439-5_43","paper_title":"A Data-Driven Approach to Improve 3D Head-Pose Estimation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":17,"model":"RetinaFace R-50 (5 points)","metrics":{"MAE (trained with other data)":"4.578"},"uses_additional_data":false,"paper_date":"2020-12-14","paper":"/paper/img2pose-face-alignment-and-detection-via","paper_url":"https://arxiv.org/abs/2012.07791v2","paper_title":"img2pose: Face Alignment and Detection via 6DoF, Face Pose Estimation","code":"https://github.com/vitoralbiero/img2pose","n_code_links":2,"syntology":null},{"rank_in_archive_order":18,"model":"hopenet","metrics":{"Geodesic Error (GE)":"9.53","Geodesic Error - aligned (GE)":"6.6","MAE (trained with other data)":"4.89","MAE-aligned (trained with other data)":"3.48"},"uses_additional_data":false,"paper_date":"2017-10-02","paper":"/paper/fine-grained-head-pose-estimation-without","paper_url":"http://arxiv.org/abs/1710.00925v5","paper_title":"Fine-Grained Head Pose Estimation Without Keypoints","code":"https://github.com/natanielruiz/deep-head-pose","n_code_links":14,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"FAN (12 points)","metrics":{"MAE (trained with other data)":"7.882"},"uses_additional_data":false,"paper_date":"2017-03-21","paper":"/paper/how-far-are-we-from-solving-the-2d-3d-face","paper_url":"http://arxiv.org/abs/1703.07332v3","paper_title":"How far are we from solving the 2D & 3D Face Alignment problem? (and a dataset of 230,000 3D facial landmarks)","code":"https://github.com/1adrianb/face-alignment","n_code_links":8,"syntology":{"n_ran":4,"n_unverified":11,"n_samples":15,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"KEPLER","metrics":{"MAE (trained with other data)":"13.852"},"uses_additional_data":false,"paper_date":"2017-02-16","paper":"/paper/kepler-keypoint-and-pose-estimation-of","paper_url":"http://arxiv.org/abs/1702.05085v1","paper_title":"KEPLER: Keypoint and Pose Estimation of Unconstrained Faces by Learning Efficient H-CNN Regressors","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":21,"model":"3DDFA","metrics":{"MAE (trained with other data)":"19.068"},"uses_additional_data":false,"paper_date":"2015-11-23","paper":"/paper/face-alignment-across-large-poses-a-3d","paper_url":"http://arxiv.org/abs/1511.07212v1","paper_title":"Face Alignment Across Large Poses: A 3D Solution","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":22,"model":"VGGNet-RNN","metrics":{"MAE (trained with BIWI data)":"1.413"},"uses_additional_data":true,"paper_date":"2017-07-01","paper":"/paper/dynamic-facial-analysis-from-bayesian","paper_url":"http://openaccess.thecvf.com/content_cvpr_2017/html/Gu_Dynamic_Facial_Analysis_CVPR_2017_paper.html","paper_title":"Dynamic Facial Analysis: From Bayesian Filtering to Recurrent Neural Network","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":23,"model":"Direct Regression","metrics":{"MAE (trained with BIWI data)":"2.54"},"uses_additional_data":false,"paper_date":"2020-06-29","paper":"/paper/deep-ordinal-regression-with-label-diversity","paper_url":"https://arxiv.org/abs/2006.15864v1","paper_title":"Deep Ordinal Regression with Label Diversity","code":"https://github.com/axeber01/dold","n_code_links":1,"syntology":null},{"rank_in_archive_order":24,"model":"Hybrid Coarse-Fine","metrics":{"MAE (trained with BIWI data)":"3.0174"},"uses_additional_data":false,"paper_date":"2019-01-21","paper":"/paper/hybrid-coarse-fine-classification-for-head","paper_url":"https://arxiv.org/abs/1901.06778v2","paper_title":"Hybrid coarse-fine classification for head pose estimation","code":"https://github.com/haofanwang/accurate-head-pose","n_code_links":1,"syntology":null},{"rank_in_archive_order":25,"model":"EHPNet","metrics":{"MAE (trained with BIWI data)":"3.43"},"uses_additional_data":false,"paper_date":"2022-10-25","paper":"/paper/an-effective-deep-network-for-head-pose","paper_url":"https://arxiv.org/abs/2210.13705v1","paper_title":"An Effective Deep Network for Head Pose Estimation without Keypoints","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":26,"model":"DAD-3DNet","metrics":{"MAE (trained with BIWI data)":"3.98"},"uses_additional_data":false,"paper_date":"2022-04-07","paper":"/paper/dad-3dheads-a-large-scale-dense-accurate-and","paper_url":"https://arxiv.org/abs/2204.03688v2","paper_title":"DAD-3DHeads: A Large-scale Dense, Accurate and Diverse Dataset for 3D Head Alignment from a Single Image","code":"https://github.com/PinataFarms/DAD-3DHeads","n_code_links":1,"syntology":null},{"rank_in_archive_order":27,"model":"2DCube","metrics":{"MAE (trained with BIWI data)":"4.31"},"uses_additional_data":false,"paper_date":"2022-12-07","paper":"/paper/an-intuitive-and-unconstrained-2d-cube","paper_url":"https://arxiv.org/abs/2212.03623v1","paper_title":"An Intuitive and Unconstrained 2D Cube Representation for Simultaneous Head Detection and Pose Estimation","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":28,"model":"Multi-Loss ResNet50","metrics":{"MAE (trained with BIWI data)":"4.895"},"uses_additional_data":false,"paper_date":"2017-10-02","paper":"/paper/fine-grained-head-pose-estimation-without","paper_url":"http://arxiv.org/abs/1710.00925v5","paper_title":"Fine-Grained Head Pose Estimation Without Keypoints","code":"https://github.com/natanielruiz/deep-head-pose","n_code_links":14,"syntology":{"n_ran":3,"n_unverified":0,"n_samples":3,"n_pointer_only_licence":0}},{"rank_in_archive_order":29,"model":"6DRepNet","metrics":{"MAEV":"5.32"},"uses_additional_data":false,"paper_date":"2023-09-14","paper":"/paper/towards-robust-and-unconstrained-full-range","paper_url":"https://arxiv.org/abs/2309.07654v1","paper_title":"Towards Robust and Unconstrained Full Range of Rotation Head Pose Estimation","code":"https://github.com/thohemp/6drepnet360","n_code_links":2,"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":4,"rows_with_any_sample_ran":4,"distinct_papers_with_graph_line":3,"distinct_papers_with_any_sample_ran":3,"samples_over_distinct_papers":{"n_ran":12,"n_unverified":14,"n_samples":26,"n_pointer_only_licence":3,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":15,"n_unverified":14,"n_samples":29,"n_pointer_only_licence":3,"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"}}}