Abstract
We present a discriminative approach to human action recognition. At the heart of our approach is the use of common spatial patterns (CSP), a spatial filter technique that transforms temporal feature data by using differences in variance between two classes. Such a transformation focusses on differences between classes, rather than on modelling each class individually. As a results, to distinguish between two classes, we can use simple distance metrics in the low-dimensional transformed space. The most likely class is found by pairwise evaluation of all discriminant functions. Our image representations are silhouette boundary gradients, spatially binned into cells. We achieve scores of approximately 96% on a standard action dataset, and show that reasonable results can be obtained when training on only a single subject. Future work is aimed at combining our approach with automatic human detection.
Original language | Undefined |
---|---|
Title of host publication | IEEE International Conference on Automatic Face and Gesture Recognition (FG 2008) |
Editors | J. Cohn, T.S Huang, Maja Pantic, N. Sebe |
Place of Publication | Los Alamitos |
Publisher | IEEE Computer Society Press |
Pages | 1-8 |
Number of pages | 8 |
ISBN (Print) | 978-1-4244-2154-1 |
DOIs | |
Publication status | Published - 17 Sep 2008 |
Event | 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008 - Amsterdam, Netherlands Duration: 17 Sep 2008 → 19 Sep 2008 Conference number: 8 |
Publication series
Name | |
---|---|
Publisher | IEEE Computer Society Press |
Number | 2008/14935 |
Conference
Conference | 8th IEEE International Conference on Automatic Face and Gesture Recognition, FG 2008 |
---|---|
Abbreviated title | FG |
Country/Territory | Netherlands |
City | Amsterdam |
Period | 17/09/08 → 19/09/08 |
Keywords
- EWI-14643
- HMI-CI: Computational Intelligence
- METIS-255025
- EC Grant Agreement nr.: FP6/033812
- IR-62627