Vertical monocular passenger flow volume statistical method based on RFCN

Abstract

The invention provides a vertical monocular passenger flow volume statistical method based on RFCN. A deep learning target detection algorithm based on RFCN is adopted, pedestrians existing in a picture are detected according to real-time picture information obtained by vertically-installed monitoring equipment, the number of the pedestrians in the picture is counted, the positions where the pedestrians are located are located, and therefore passenger flow information is accurately obtained in real time. According to the RFCN-based vertical monocular passenger flow volume statistical method provided by the invention, the ResNet-18 deep convolutional neural network is used as the basis, and the RFCN-based novel deep learning target detection algorithm is adopted, so that the detection speedis high, the real-time performance is high, the detection is accurate, and the actual application requirements are met.

Description

Technical field [0001] The invention belongs to the field of video surveillance, and particularly relates to a vertical monocular passenger flow statistics method based on RFCN. Background technique [0002] With the development and wide application of video surveillance technology, the number of people can be counted based on the video images obtained by the surveillance equipment, which can conveniently, reliably and real-timely count the passenger flow in various places without feeling, and can quickly grasp the dynamics of the passenger flow. The commonly used monitoring equipment is limited by the installation location and the angle is inclined. It can still deal with the situation of small passenger flow. However, when the passenger flow is dense, the mutual blocking problem cannot be solved, resulting in inaccurate passenger flow statistics. In order to solve the occlusion problem, there is also a way to propose binocular vision, but the use of two cameras greatly increase...

AI Technical Summary

Problems solved by technology

The commonly used monitoring equipment is limited by the installation location, and the angle is inclined. It can cope with the situation of less passenger flow, but when the passenger flow is dense, it cannot solve the problem of mutual occlusion, resulting in inaccurate passenger flow statistics.

In order to solve the occlusion problem, there is also a method of binocular vision, but the use of two cameras greatly increases the cost, and the installation conditions of the equipment are also high, and it is also a waste of resources.

Images