Image Recognition by YoloV4 for Vehicle Type Distinction in Side of Road

At present, most of the domestic and foreign research on vehicle recognition focuses on real-time camera recognition in front of the vehicle while driving and real-time camera recognition over the road for overhead camera angles. For roads in mountainous areas or special terrain, due to the small hinterland on the left and right side of the road, it is not easy to set up overhead camera systems for the road. Therefore, this thesis proposes “Yolo_v4-based road vehicle recognition” to solve the above problem. Unlike the overhead camera system, this side camera detachable temporary road vehicle recognition system only needs a very small road hinterland and can reduce the cost of space construction. The Yolo_V4 deep learning algorithm can accurately identify different vehicle types to ensure road safety and traffic monitoring. In this study, the accuracy of the vehicle recognition model for the side camera angle is 96%, which is comparable to the accuracy of the vehicle recognition model for the top camera angle.