The problem addressed by the paper is the redundancy that is inherent in most of the present video surveillance systems which rely on centralized monitoring of video streams, wherein each camera feed is treated as an isolated stream. The paper discusses a new surveillance strategy and its implementation, in which multiple cameras work in a collaborative manner in order to reduce the number of effective video streams. Also discussed is the implementation of real-time face detection and face recognition in the system. The system that is implemented is much more reliable against attacks, costs less overall, needs lesser power to operate and provides analytics, leading to lesser dependency on the operator.