Group interaction and group tracking for video-surveillance in underground railway stations

In this paper we propose an approach to recognize behaviors of groups of people in the subway. Violent behavior or vandalism per- formed by a group can be detected in order to alert subway security. The proposed system is composed of 3 main layers: the detection of people in the video, the detection and tracking of groups among the detected individuals and the detection of events and scenarios of interest based on tracked actors (groups). The main focus of this paper are the group tracking and event detection layers.