Perceiving Systems, Computer Vision

Efficient Non-linear Markov Models for Human Motion

2014

Conference Paper

ps


Dynamic Bayesian networks such as Hidden Markov Models (HMMs) are successfully used as probabilistic models for human motion. The use of hidden variables makes them expressive models, but inference is only approximate and requires procedures such as particle filters or Markov chain Monte Carlo methods. In this work we propose to instead use simple Markov models that only model observed quantities. We retain a highly expressive dynamic model by using interactions that are nonlinear and non-parametric. A presentation of our approach in terms of latent variables shows logarithmic growth for the computation of exact loglikelihoods in the number of latent states. We validate our model on human motion capture data and demonstrate state-of-the-art performance on action recognition and motion completion tasks.

Author(s): Andreas M. Lehrmann and Peter V. Gehler and Sebastian Nowozin
Book Title: Proceedings IEEE Conf. on Computer Vision and Pattern Recognition (CVPR)
Pages: 1314-1321
Year: 2014
Month: June
Publisher: IEEE

Department(s): Perceiving Systems
Research Project(s): Pose and Motion Priors
Bibtex Type: Conference Paper (inproceedings)
Paper Type: Conference

DOI: 10.1109/CVPR.2014.171
Event Name: IEEE International Conference on Computer Vision and Pattern Recognition
Event Place: Columbus, Ohio, USA
Attachments: pdf

BibTex

@inproceedings{lehrmann14motion,
  title = {Efficient Non-linear Markov Models for Human Motion},
  author = {Lehrmann, Andreas M. and Gehler, Peter V. and Nowozin, Sebastian},
  booktitle = { Proceedings IEEE Conf. on Computer Vision and Pattern Recognition (CVPR)},
  pages = {1314-1321},
  publisher = {IEEE},
  month = jun,
  year = {2014},
  doi = {10.1109/CVPR.2014.171},
  month_numeric = {6}
}