{"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/coarse-to-fine-volumetric-prediction-for","title":"Coarse-to-Fine Volumetric Prediction for Single-Image 3D Human Pose","arxiv_id":"1611.07828","date":"2016-11-23","proceeding":"CVPR 2017 7","authors":["Georgios Pavlakos","Xiaowei Zhou","Konstantinos G. Derpanis","Kostas Daniilidis"],"abstract":"This paper addresses the challenge of 3D human pose estimation from a single\ncolor image. Despite the general success of the end-to-end learning paradigm,\ntop performing approaches employ a two-step solution consisting of a\nConvolutional Network (ConvNet) for 2D joint localization and a subsequent\noptimization step to recover 3D pose. In this paper, we identify the\nrepresentation of 3D pose as a critical issue with current ConvNet approaches\nand make two important contributions towards validating the value of end-to-end\nlearning for this task. First, we propose a fine discretization of the 3D space\naround the subject and train a ConvNet to predict per voxel likelihoods for\neach joint. This creates a natural representation for 3D pose and greatly\nimproves performance over the direct regression of joint coordinates. Second,\nto further improve upon initial estimates, we employ a coarse-to-fine\nprediction scheme. This step addresses the large dimensionality increase and\nenables iterative refinement and repeated processing of the image features. The\nproposed approach outperforms all state-of-the-art methods on standard\nbenchmarks achieving a relative error reduction greater than 30% on average.\nAdditionally, we investigate using our volumetric representation in a related\narchitecture which is suboptimal compared to our end-to-end approach, but is of\npractical interest, since it enables training when no image with corresponding\n3D groundtruth is available, and allows us to present compelling results for\nin-the-wild images.","url_abs":"http://arxiv.org/abs/1611.07828v2","url_pdf":"http://arxiv.org/pdf/1611.07828v2.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":[{"paper_slug":"coarse-to-fine-volumetric-prediction-for","repo_url":"https://github.com/geopavlakos/c2f-vol-demo","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"coarse-to-fine-volumetric-prediction-for","repo_url":"https://github.com/geopavlakos/c2f-vol-train","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"coarse-to-fine-volumetric-prediction-for","repo_url":"https://github.com/strawberryfg/c2f-3dhm-human-caffe","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"torch","reach":{"status":"unanswered"}},{"paper_slug":"coarse-to-fine-volumetric-prediction-for","repo_url":"https://github.com/thuml/ContextWM","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"3d-human-pose-estimation","task_name":"3D Human Pose Estimation"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-human-pose-estimation-on-humaneva-i","task":"3D Human Pose Estimation","dataset":"HumanEva-I","model":"c2f-vol","rank_in_archive_order":18,"of":31,"metrics":{"Mean Reconstruction Error (mm)":"24.3"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1611.07828","atlas_url":"https://app.syntology.ai/?focus=1611.07828","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}