SEMI-SUPERVISED ENSEMBLE TRACKING

In this paper, we propose a semi-supervised ensemble track-ing approach under the framework of particle filter. The parti-cle filter is used not only for object searching, but also for un-labelled sample generation. By adopting the semi-supervised learning technology, these unlabelled samples which are gen-erated online are utilized to progressively modify the clas-sifier and make the ensemble tracker to be more robust to environment changing. On the other hand, utilizing semi-supervised learning technology can avoid the drifting phe-nomenons which are often encountered when using super-vised learning. Finally, the performance of the proposed ap-proach is evaluated using real visual tracking examples. Index Terms — Semi-supervised learning, ensemble track-ing, visual tracking 1