Video analytics for surveillance camera networks

Abstract

International usage and interest in Closed-Circuit Television (CCTV) for surveillance of public spaces has proved to be an effective in forensic, or reactive response to crime and terrorism. In the ideal scenario, it would be useful to detect events, in real-time or close to real-time, in order to mitigate possible harm. However, it is an issue to adequately monitor the video feeds with security guards. In this paper, we address the key problems in the existing surveillance system, followed by some discussions on the integration of social signal and video search to enhance the traditional surveillance system. We shows the performance of face-based identity inference under surveillance. as well as video indexing and search results. Finally, we discuss the related future research direction in this area.