License Plate Recognition Using Raspberry Pi

In this modern world, the population is increasing rapidly. As the population increases, the number of motor vehicles also increases. Today, the main concern is the authenticity of the driver and the vehicle. Over the past decade, various vehicle license plate recognition methods have been proposed for various transportation applications. At the same time, automation is also important in this busy world. In this paper, we provide a detailed review of the literature related to license plate recognition and reading system. Despite the numerous methods of vehicle number plate detection technology, current ANPR systems have a difficult time accurately identifying number plates because they vary in size, orientation, and language depending on where they are used in the world. In this research, we have used the state-of-the-art YOLOv5 object detection model for both number plate detection as well as for recognizing the characters in the number plate. The model is trained twice on a dataset of license plates with only English alphabets and digits. The dataset is annotated, the images are fed to YOLOv5 for training, the labels are added, and the weight files are generated. The weight files are loaded to PyTorch. Based on the weight files generated from training, the proposed system detects the number plate and the characters in the number plate. This system can accurately read and interpret number plates from different angles and distances. The implementation of number plate recognition using YOLOv5 has significant potential to improve traffic safety, reduce congestion, and enhance security by automating the identification and tracking of vehicles.