Pedestrian Counting Without Tracking for the Security Application of IoT Framework

Currently, the security of the Internet of Things (IoT) has aroused great concern. Pedestrian counting under video surveillance has become a key problem affecting social security. In this paper we describe a novel and real-time pedestrian counting framework without using any tracking algorithms. Current research under wide overhead cameras are mainly focus on the tracking-based algorithms, however, effective tracking is difficult in most cases. Therefore, we design a line sampling process, based on this strategy, we can achieve a temporal slice image that contains useful head feature information, which can be used for pedestrian counting without the necessity for visual tracking. As is expected, our algorithm is more stable and accurate than existing approaches. In addition, we also design a two-stage detection algorithm, which is used to locate head position. Experimental results indicate that our constructed algorithm can obtain better performance.