{"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/single-camera-pose-estimation-using-bayesian","title":"Single camera pose estimation using Bayesian filtering and Kinect motion priors","arxiv_id":"1405.5047","date":"2014-05-20","proceeding":null,"authors":["Michael Burke","Joan Lasenby"],"abstract":"Traditional approaches to upper body pose estimation using monocular vision\nrely on complex body models and a large variety of geometric constraints. We\nargue that this is not ideal and somewhat inelegant as it results in large\nprocessing burdens, and instead attempt to incorporate these constraints\nthrough priors obtained directly from training data. A prior distribution\ncovering the probability of a human pose occurring is used to incorporate\nlikely human poses. This distribution is obtained offline, by fitting a\nGaussian mixture model to a large dataset of recorded human body poses, tracked\nusing a Kinect sensor. We combine this prior information with a random walk\ntransition model to obtain an upper body model, suitable for use within a\nrecursive Bayesian filtering framework. Our model can be viewed as a mixture of\ndiscrete Ornstein-Uhlenbeck processes, in that states behave as random walks,\nbut drift towards a set of typically observed poses. This model is combined\nwith measurements of the human head and hand positions, using recursive\nBayesian estimation to incorporate temporal information. Measurements are\nobtained using face detection and a simple skin colour hand detector, trained\nusing the detected face. The suggested model is designed with analytical\ntractability in mind and we show that the pose tracking can be\nRao-Blackwellised using the mixture Kalman filter, allowing for computational\nefficiency while still incorporating bio-mechanical properties of the upper\nbody. In addition, the use of the proposed upper body model allows reliable\nthree-dimensional pose estimates to be obtained indirectly for a number of\njoints that are often difficult to detect using traditional object recognition\nstrategies. Comparisons with Kinect sensor results and the state of the art in\n2D pose estimation highlight the efficacy of the proposed approach.","url_abs":"http://arxiv.org/abs/1405.5047v2","url_pdf":"http://arxiv.org/pdf/1405.5047v2.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":"single-camera-pose-estimation-using-bayesian","repo_url":"https://github.com/mgb45/mkfbodytracker_pdaf","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"2d-pose-estimation","task_name":"2D Pose Estimation"},{"task_slug":"camera-pose-estimation","task_name":"Camera Pose Estimation"},{"task_slug":"computational-efficiency","task_name":"Computational Efficiency"},{"task_slug":"face-detection","task_name":"Face Detection"},{"task_slug":"object-recognition","task_name":"Object Recognition"},{"task_slug":"pose-estimation","task_name":"Pose Estimation"},{"task_slug":"pose-tracking","task_name":"Pose Tracking"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}