A multi-target tracking algorithm in dense-crowded environments is proposed. Existing global optimisation based multi-target tracking assumes that each pedestrian's motion is independent and shows impressive results in sparse datasets. However, in semi-crowded environments, pedestrians often occlude and interact with each other, making tracking a challenging task. In this reported work, social group behaviour to mitigate ambiguities is considered using the social force model, which is widely used in crowd simulation applications, and the robustness of the proposed method compared to state-of-the-art multi-target tracking in more crowded scenarios is demonstrated.
展开▼
