Bayesian Non-Parametric Inference for Manifold Based MoCap Representation

Fabrizio Natola, Valsamis Ntouskos, Marta Sanzari, Fiora Pirri

ICCV 2015

doi:10.1109/ICCV.2015.523 /iccv/2015/natola2015iccv-bayesian/

Abstract

We propose a novel approach to human action recognition, with motion capture data (MoCap), based on grouping sub-body parts. By representing configurations of actions as manifolds, joint positions are mapped on a subspace via principal geodesic analysis. The reduced space is still highly informative and allows for classification based on a non-parametric Bayesian approach, generating behaviors for each sub-body part. Having partitioned the set of joints, poses relative to a sub-body part are exchangeable, given a specified prior and can elicit, in principle, infinite behaviors. The generation of these behaviors is specified by a Dirichlet process mixture. We show with several experiments that the recognition gives very promising results, outperforming methods requiring temporal alignment.

PDF ICCV Semantic Scholar

Cite

Text

Natola et al. "Bayesian Non-Parametric Inference for Manifold Based MoCap Representation." International Conference on Computer Vision, 2015. doi:10.1109/ICCV.2015.523

Markdown

[Natola et al. "Bayesian Non-Parametric Inference for Manifold Based MoCap Representation." International Conference on Computer Vision, 2015.](https://mlanthology.org/iccv/2015/natola2015iccv-bayesian/) doi:10.1109/ICCV.2015.523

BibTeX

@inproceedings{natola2015iccv-bayesian,
  title     = {{Bayesian Non-Parametric Inference for Manifold Based MoCap Representation}},
  author    = {Natola, Fabrizio and Ntouskos, Valsamis and Sanzari, Marta and Pirri, Fiora},
  booktitle = {International Conference on Computer Vision},
  year      = {2015},
  doi       = {10.1109/ICCV.2015.523},
  url       = {https://mlanthology.org/iccv/2015/natola2015iccv-bayesian/}
}