Method for estimating and early warning passenger flow retained in real time in subway station

Abstract

The invention discloses a method for estimating and early warning passenger flow retained in real time in a subway station, which comprises the following steps of: (1) reading video streams of monitoring cameras at an entrance and a platform in the station, and carrying out hard solution on the video streams into image frames by adopting a GPU (Graphics Processing Unit); (2) performing pedestrian detection on the image frame; (3) tracking a human body target by using a tracking algorithm, judging the passenger entering and exiting behaviors according to the pedestrian movement direction, counting the exit distribution number, and when the target passes through the counting datum line, increasing the counting value of the pedestrians entering or leaving the station; (4) establishing a passenger flow model according to the entrance and exit distribution number; and (5) carrying out signal linkage with the vehicles in the station, calculating the number of passengers remaining in the station, and if the number of passengers exceeds a set threshold value, sending response large passenger flow alarm information to a station service system to remind a station manager to take countermeasures. According to the invention, the monitoring camera in the subway station is used to realize the estimation of the in-station remaining passenger flow, which is helpful for the safe operation of the subway and the refined passenger flow management.

Description

technical field [0001] The invention belongs to the field of deep computer vision and intelligent monitoring, and in particular relates to a real-time passenger flow estimation and early warning method in a subway station. Background technique [0002] With the development of urban rail transit, more and more people choose to travel by subway. Due to the growth of urban population and the convergence of people's travel time, the subway often produces a large passenger flow phenomenon during operation. Because the existing AFC system cannot real-time Sensing the passenger flow in and out, the passenger flow management in the subway and the carrying situation of the passenger flow in the station often can only rely on the experience of the subway staff to judge and decide whether to restrict and guide the passenger flow. With the development of artificial intelligence and machine vision technology, the video-based passenger flow counting method can be effectively applied to th...

AI Technical Summary

Problems solved by technology

[0002] With the development of urban rail transit, more and more people choose to travel by subway. Due to the growth of urban population and the convergence of people's travel time, the subway often produces a large passenger flow phenomenon during operation. Because the existing AFC system cannot real-time Sensing the passenger flow in and out, the passenger flow management in the subway and the carrying situation of the passenger flow in the station can only rely on the experience of the subway staff to judge and decide whether to restrict and guide the passenger flow

Images