Abstract

In this paper, we address the problem of model-free on-line object tracking based on color representations. According to the findings of recent benchmark evaluations, suchtrackers often tend to drift towards regions which exhibit asimilar appearance compared to the object of interest. Toovercome this limitation, we propose an efficient discriminative object model which allows us to identify potentially distracting regions in advance. Furthermore, we exploit thisknowledge to adapt the object representation beforehand sothat distractors are suppressed and the risk of drifting issignificantly reduced. We evaluate our approach on recentonline tracking benchmark datasets demonstrating state-ofthe-art results. In particular, our approach performs favorably both in terms of accuracy and robustness compared to recent tracking algorithms. Moreover, the proposed approach allows for an efficient implementation to enable on-line object tracking in real-time.