License plate recognition based on YOLOv5-LPRNet

In recent years, the number of domestic vehicles has been increasing, the fine management of vehicles has become more difficult, and the importance of license plate recognition technology has become increasingly prominent. The traditional license plate recognition algorithm can be effectively applied to ordinary life scenes, but it is difficult to show strong robustness in the face of complex scenes such as image distortion and blurring, and often fails to recognize the phenomenon. This paper uses YOLOv5 and LPRNet deep learning models to recognize license plates in complex scenes in real-time. The main task of the former is to locate the license plate position within the image and crop the detection frame, while the main task of the latter is to recognize the license plate characters in the detection frame. Compared with traditional license plate recognition algorithms, this method using deep learning improves the accuracy of license plate recognition. In contrast, the method has the advantages of a small model, high precision, and embeddability.