Developed a supervised tracking system based on Particle Filter and Multiple Instance Learning (MIL)
the orignal MIL code refer to http://vision.ucsd.edu/~bbabenko/project_miltrack.shtml
My Work:
Used Particle filter as the sample method to find the best observation from the
candidate prediction.
Updated the appearance model by online MIL.
Haar-like feature was used as the weak classifier in the process
the experimental analysis using matlab is also included in this project