Research On Detection Algorithm Of Pedestrian Running Red Light Based On Yolov4

Under the initiative of a civilized city,the state has always called on people to obey the traffic rules and travel in a civilized manner.However,after passing the investigation,some people still ran red lights when they passed the crosswalk.Pedestrian detection has become a hot spot in the field of target detection,and it is widely used in intelligent monitoring,traffic and vehicle-assisted driving.Among the target detection algorithms,Yolov4 algorithm is one of the most advanced algorithms in the field of pedestrian detection.Therefore,in view of this problem,it is particularly necessary to study a set of pedestrian red light detection system.This article combines the Yolov4 algorithm with deep learning and applies it to the detection of pedestrians running through red lights.The specific work is summarized as follows:(1)First,this paper studies the current domestic and foreign development of pedestrian detection.In combination with the problem of pedestrians running red lights,deep learning models are used to detect whether pedestrians are running red lights.Then it introduces the theoretical knowledge of convolutional neural network and Yolo series algorithms involved in target detection in detail,which lays the foundation for selecting Yolov4 algorithm as the research of pedestrian red light detection in this paper.(2)Secondly,this article specifically introduces the principle of each module in the Yolov4 pedestrian red light detection algorithm,which includes video capture,traffic light time detection,crosswalk detection,pedestrian red light detection and number counting.In addition,non-maximum suppression and loss function processing are used for the detection results to train a model that can detect red lights,green lights,crosswalks,and pedestrians at the same time.(3)Finally,a large number of experimental videos are used to verify the performance of the Yolov4 deep learning model for pedestrian red light detection algorithms.It mainly conducts comparative experiments before and after pedestrians without occlusion,less occlusion,and more occlusion in the detection target.Through the analysis of experimental data,the Yolov4 pedestrian red light detection model can achieve good detection results in pedestrian red light monitoring.The model can detect the video at a speed of 32 frames per second in real time.It has good generalization ability and robustness,and can meet the requirements of real-time detection of pedestrians running red lights.It can be determined that the method proposed in this paper is feasible.