Aiming at the phenomenon that the red light often brings serious traffic accidents,a statistical solution based on YOLOv3 for the number of red light lines is proposed. The program is divided into two parts: front- end image processing and back-end data statistics,including real-time video capture, traffic light time detection,pedestrian detection and data statistics. The system improved the YOLOv3 algorithm,and used the VOC dataset to train the model that can only detect the characters and detect the video stream,which greatly improved the real- time performance of the operation,and increased the statistics and box coordinates of the number of selected characters. The real-time return of information realizes the statistics of the number of pedestrians red light. Background data stores key information through file operations. In practice,the system has good real-time and accuracy,and can provide effective data support for urban managers to plan traffic,which is conducive to urban security and smart city construction.
Electronic Design Engineering