Object tracking with online discriminative sub-instance learning
Peng Tian · 2015
For object tracking under complex scenes, this paper proposes an improved multi-instance target tracking algorithm. The algorithm is based on the binary classification. The most pivotal step of this algorithm is to correct and confirm the target location by describing the sample by color characteristic after the target location is orientated by the binary classification. The experiment results show the proposed algorithm realizes the robustness of the target tracking in a certain extent.