{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/3d-human-pose-estimation-from-a-single-image","title":"3D Human Pose Estimation from a Single Image via Distance Matrix Regression","arxiv_id":"1611.09010","date":"2016-11-28","proceeding":"CVPR 2017 7","authors":["Francesc Moreno-Noguer"],"abstract":"This paper addresses the problem of 3D human pose estimation from a single\nimage. We follow a standard two-step pipeline by first detecting the 2D\nposition of the $N$ body joints, and then using these observations to infer 3D\npose. For the first step, we use a recent CNN-based detector. For the second\nstep, most existing approaches perform 2$N$-to-3$N$ regression of the Cartesian\njoint coordinates. We show that more precise pose estimates can be obtained by\nrepresenting both the 2D and 3D human poses using $N\\times N$ distance\nmatrices, and formulating the problem as a 2D-to-3D distance matrix regression.\nFor learning such a regressor we leverage on simple Neural Network\narchitectures, which by construction, enforce positivity and symmetry of the\npredicted matrices. The approach has also the advantage to naturally handle\nmissing observations and allowing to hypothesize the position of non-observed\njoints. Quantitative results on Humaneva and Human3.6M datasets demonstrate\nconsistent performance gains over state-of-the-art. Qualitative evaluation on\nthe images in-the-wild of the LSP dataset, using the regressor learned on\nHuman3.6M, reveals very promising generalization results.","url_abs":"http://arxiv.org/abs/1611.09010v1","url_pdf":"http://arxiv.org/pdf/1611.09010v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[],"tasks":[{"task_slug":"3d-human-pose-estimation","task_name":"3D Human Pose Estimation"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"},{"task_slug":null,"task_name":"Position"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-human-pose-estimation-on-humaneva-i","task":"3D Human Pose Estimation","dataset":"HumanEva-I","model":"EDM","rank_in_archive_order":21,"of":31,"metrics":{"Mean Reconstruction Error (mm)":"26.9"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1611.09010","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}