Fig. 4From: Robust object tracking via online discriminative appearance modelingTracking results on motocross video sequence for frames of 2, 7, 10, 17, and 27. a The standard mean shift method [18]. b The DF method [17]. c The proposed methodBack to article page