Reliable multi-object tracking dealing with occlusions for a smart camera

In this paper, a multi-object tracking system designed for a low cost embedded smart camera is proposed. Objects tracking constitutes a main step in video-surveillance applications. Because of the number of cameras used to cover a large area, surveillance applications are constrained by the cost of each node, the power efficiency of the system, the robustness of the tracking algorithm and the real-time processing. They require a reliable multi-object tracking algorithm that can run in a real-time on light computing architectures. In this paper, we propose a tracking pipeline designed for a fixed smart camera that can handle occlusions between objects. We show that the proposed pipeline reaches real-time processing on the RaspberryPi board equipped with the RaspiCam camera. The tracking quality of the proposed pipeline is evaluated on publicly available datatsets: PETS2009 and CAVIAR.