Particle Filter Tracking With Online Multiple Instance Learning

This paper addresses the problem of object tracking by learning a discriminative classifier to separate the object from its background. The online-learned clas-sifier is used to adaptively model object ’ appearance and its background. To solve the typical problem of er-roneous training examples generated during tracking, an online multiple instance learning (MIL) algorithm is used by allowing false positive examples. In addi-tion, particle filter is applied to make best use of the learned classifier and help to generate a better repre-sentative set of training examples for the online MIL learning. The effectiveness of the proposed algorithm is demonstrated in some challenging environments for human tracking. 1