A collaborative surveillance system for role identification

Many works of conventional surveillance have focused on people tracking, behavior or event detection, gait or face based recognition, etc. However, role identification is also very important in video surveillance but usually paid less attention. In this paper, we propose a collaborative multi-camera system to identify people with specific roles using a causal network to form a best identification result from the evidences known so far. Not only visual features but also spatio-temporal features are used in our method, as well as some object specific features. Collaborative multiple cameras benefit locating the position of moving objects and overcoming occlusions. Experimental results demonstrates the effectiveness of our method.