A Bayesian Approach to Tracking Learning Detection

Tracking objects of interest in video sequences, referred in computer vision literature as video tracking or visual tracking, is an essential task for intelligent machines able to understand and react to the surrounding environment. This work investigates the problem of robust, long-term visual tracking of unknown objects in unconstrained environments. Such problem is affected by several challenging difficulties arising from fast camera movements, partial or total object occlusions and temporal disappearance. We describe a novel framework based on Tracking-Learning-Detection (TLD), that combine bayesian optimal filtering with pn on-line learning theory [12] to adapt target visual likelihood during tracking. We designed particle filtering algorithm for parameter inference and propose a solution that enables accurate and efficient tracking. The performance and the long-term stability are demonstrated and evaluated on a set of challenging video sequences usually employed to test tracking algorithms.