Investigating Walking Characteristics of Passengers in Subway Corridors by Video Processing based on YOLOX and Deepsort Algorithm

This paper studies the walking characteristics of passengers in the transfer corridor of a subway station. First, passenger detection and tracking are performed using YOLOX and deepsort algorithms, respectively. Then, the density, distance, speed and other characteristics of passengers are calculated. Finally, on this basis, the distribution of passenger speed, the characteristics of fundamental diagram and the relationship between speed and distance are analyzed. The following main conclusions are obtained: (1) The number of passengers with medium speed is more, while the number of passengers with higher speed or lower speed is less; (2) The passenger speed decreases with the increase of the density, but when the density reaches about 3.1 p/m², the speed will increase; (3) The passenger speed has a linear relationship with the passenger distance.