Vision detection system for vehicles not yielding to pedestrians at zebra crossing

Abstract

The invention provides a vision detection system for vehicles not yielding to pedestrians at zebra crossing, comprising a monitoring camera covering the whole zebra crossing area, a HD (high definition) snapshot camera for taking snapshots of vehicles, a traffic cloud server, and a vision detection subsystem. The vision detection subsystem comprises a zebra crossing area detection module, a not-yielding-to-pedestrians behavior detection module, a not-yielding-to-pedestrians behavior vehicle image snapshot module, a vehicle type identification module, a license plate positioning and identifyingmodule, and a violation detection report generating module. The traffic cloud server is used for receiving video data of the whole zebra crossing area obtained by the monitoring camera, and submitting the video data to the vision detection subsystem to conduct not-yielding-to-pedestrians behavior detection and personal identity verification of violation vehicles. A violation detection report canbe automatically generated and stored with the detection and identification result, three snapshots of not-yielding-to-pedestrians vehicle images, and the process video images. The invention is quitehigh in detection precision and robustness.

Description

technical field [0001] The invention relates to the application of artificial intelligence, convolutional neural network and computer vision in the detection of vehicles' impolite pedestrian behavior on a zebra crossing, and belongs to the field of intelligent transportation. Background technique [0002] Article 47 of the "Road Traffic Safety Law" stipulates that when a motor vehicle passes a crosswalk, it shall slow down; when encountering a pedestrian passing a crosswalk, it shall stop and give way. When a motor vehicle passes a road without traffic signals, it shall give way to pedestrians crossing the road. However, in real life, some drivers do not abide by the regulations, and some even speed up and whistle when they see pedestrians passing the crosswalk, which brings hidden dangers to traffic safety. Every year, there are more than 1,000 fatal accidents caused by drivers not giving way to pedestrians on zebra crossings. Even in Hangzhou, the city with the best cour...

AI Technical Summary

Problems solved by technology

[0011] In order to overcome the shortcomings of low detection accuracy and low detection robustness of the existing visual detection methods for vehicles on zebra crossings that do not give way to pedestrians, the present invention provides a multi-task depth-based detection method with high detection accuracy and high robustness. Convolutional Neural Network Vision Detection System for Vehicles on Zebra Crossings and Pedestrians

Images