This paper describes an application that uses a neural network to classify vehicular traffic on video. With the increase in the number of vehicles, the need to regulate traffic on the roads to solve the problems of traffic congestion and high accident rate has arisen. Collecting data from video of vehicles on the road will help to create statistics, the use of which can be used to effectively think about the regulation of traffic on the roads. An advanced real-time object detection system YOLOv5 was used to solve the problem of vehicle classification on video. To train the neural network the data set was used consisting of 750 images from the outdoor surveillance cameras. As a result of testing the system, the recognition accuracy was 89%. The data set can be used by other researchers in their research.
Vehicle Classification Application on Video Using Yolov5 Architecture
2021-09-05
3012768 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
STD-Yolov5: a ship-type detection model based on improved Yolov5
Taylor & Francis Verlag | 2024
|