Methods for automatic pedestrian counting

Abstract

Pedestrian counting is widely utilized in various surveillance and marketing applications. However, traditional pedestrian counting devices have shortcomings such as that they can manage only one person at a time or they obstruct free movement of pedestrians. In this paper, we introduce our recent research effort for automatic pedestrian counting. Three pedestrian counting methods are introduced which use: (1) a common CCTV input, (2) arrays of IR distance sensors, and (3) an IR line laser. Experiments on real scenes showed that all three methods can count the number of pedestrians with accuracies over 95% and also with very low computational burden.