DTV: Detection, Tracking and Validation Framework for Unique People Count

Counting the unique number of people in a video (i.e., counting a person only once while the person is within the field of view), is required in many significant video analytical applications, such as transit passenger and pedestrian volume count in railway stations, shopping malls and road intersections and many others. In this paper, a novel framework is proposed for counting passengers, mainly in a railway station. The framework has three components: people detection, tracking and validation. In the detection step, a person is detected when he or she enters the field of view. Then, the person is tracked by optical flow based tracking until the person leaves the field of view. Finally, the trajectory generated by the tracker is validated through a spatiotemporal validation technique. The number of valid trajectories denotes the number of people. The novelty of the framework is the inclusion of the validation step, which is overlooked by the existing methods. Extensive experiments have been conducted on the datasets having both top views and whole body views of the passengers. Experimental results demonstrate that the proposed framework generates more than 90% accuracy on both types of views. It also detects and tracks persons having different hair colors and wearing hoodies, caps, long winter jackets, carrying bags and more. The proposed algorithm shows promising results also for people moving in different directions.