Fast Upper Body Joint Tracking Using Kinect Pose Priors

Michael Burke, Joan Lasenby

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract / Description of output

Traditional approaches to upper body pose estimation using monocular vision rely on complex body models and a large variety of geometric constraints. We argue that this is not ideal and instead attempt to incorporate these constraints through priors obtained directly from training data, by fitting a Gaussian mixture model to a large dataset of recorded human body poses, tracked using a Kinect sensor. We combine this information with a random walk transition model to obtain an upper body model that can be viewed as a mixture of discrete Ornstein-Uhlenbeck processes, in that states behave as random walks, but drift towards a set of typically observed poses. The suggested model is designed with analytical tractability in mind and we show that the pose tracking can be Rao-Blackwellised using the mixture Kalman filter, allowing for computational efficiency while still incorporating bio-mechanical properties of the upper body.
Original languageEnglish
Title of host publicationProceedings of the 8th International Conference on Articulated Motion and Deformable Objects
Place of PublicationPalma de Mallorca, Spain
PublisherSpringer, Cham
Number of pages12
ISBN (Electronic)978-3-319-08849-5
ISBN (Print)978-3-319-08848-8
Publication statusPublished - 2014
Event8th International Conference of Articulated Motion and Deformable Objects - Palma de Mallorca, Spain
Duration: 16 Jul 201418 Jul 2014


Conference8th International Conference of Articulated Motion and Deformable Objects
Abbreviated titleAMDO 2014
CityPalma de Mallorca


Dive into the research topics of 'Fast Upper Body Joint Tracking Using Kinect Pose Priors'. Together they form a unique fingerprint.

Cite this