Multiple objects monitoring based on 3D information from multiple cameras

This paper presents an approach for monitoring multiple objects in a room using multiple cameras. The proposed system tracks the objects when they have been picked and placed to a new location in the room. In this approach, a 3D room model is used to analyze the human position as well as targeted objects in the room. The system consists of several modules, namely, 3D room model initialization, foreground extraction, human detection and removal, object detection, object monitoring and management, and background modeling. Experimental results show that the proposed system can accurately monitor objects with reasonable time delay. The limitation and the encountered issues are also reported and discussed.