The MUCATAR project is aimed at designing principled methods (models and algorthms) for the
simultaneous tracking of people and recognition of their activity using multiple cameras as input, in
a probabilistic framework (Sequential Monte Carlo methods and graphical models).  Specific issues
addressed inlcude:

1.  the development of novel data fusion mechanisms to improve tracking;

2.  the development of mixed-state models (which combine continuous motion parameters and
     discrete labels in a joint distribution) for tracking and recognition;

3.  the development of new sampling methods for improving tracking efficiency and performance;

4.  data collection and development of performance evaluation procedures.


This work is funded funded by the Swiss National Center of Competence in Research (NCCR) on
Interactive Multimodal Information Management (IM)2.

This site is maintained by Kevin Smith.