An algorithm using YOLOv4 and DeepSORT for tracking vehicle speed on the highway

Abstract

Currently, expressways are increasingly developed and expanded. Several highways of Vietnam allow vehicles to travel up to 120 kilometers per hour helping to transport goods quickly and bringing a lot of socio-economic benefits. Vehicle monitoring plays an important role in reducing traffic accidents helping to handle violations. The standard for vehicle speed estimation is radar or lidar speed signs which can be costly to buy and maintain. The paper proposes a model to identify and monitor car speed on highways. The proposed method uses YOLOv4 combined with DeepSORT for vehicle identification and tracking. We then calculate the speed of the car based on video recording and sending it back from the highway. The execution context is a highway where vehicles move very fast. The results show that the system improves 46% average accuracy compared with [27] and [28] and execution times for up to 70 frames per second that is suitable for real systems.