Subway Station Platform Crowd Density of Off-peak hours smart Monitoring on YOLOv3

The whole process of monitoring the crowd density of subway station based on YOLOv3 is proposed and implemented with some parameters adjusted as well as a new feature to identify crowd density added, which laids a foundation for intelligent dispersion and guidance of subway station pedestrain flow. Hundreds of picture samples are taken to train the network model. The results of cross validation shows that the rate of missed detection can be less than 10%, which preliminarily satisfies the precision requirements of operation.