V. Bellandi, M. Anisetti, and F. Beverina (Italy)
Face tracking, Candide-3, Feature Extraction, Iris detect, FACS,
Nowadays there are not any reliable systems able to codify a video streaming according to the FACS. The few system that are able to do it works on frontal, or however not natu ral postures. The system we propose, takes advantage from a robust 3D tracking based on a 3D template, in order to extract some features correlated to the facial expressions. Moreover, this system provides for each feature a reliabil ity level according to the presence of some occlusions, or due to some errors in tracking that could hide or make the reconstruction on the template, become critical.
Important Links:
Go Back
