Autonomous monitoring system for a public residential street

Video surveillance has become a widely used monitoring medium. Most of the recorded data is not useful so storing it all is not practical. For these systems to be useful it requires a human operator to consistently monitor the footage. This paper proposes the use of an autonomous monitoring system for public residential streets. Only relevant footage is recorded and stored greatly reducing the storage requirements. At the core of this system, is a human detection and tracking algorithm. For the desired system, an improved HOG human detector was used to extract human features in video frames while a Kalman filter was used to track detected human subjects. The overall system accuracy was determined as 86.39% demonstrated the effectiveness and robustness of the proposed system.