基于YOLOv5的车牌识别系统
A License Plate Recognition System Based on YOLOv5

近年来，车辆的数量不断增多，智能交通场景的不断完善，在现实生活中车牌识别技术的重要性显而易见，本文提出了一种YOLOv5网络与LPRNet网络相结合的技术，同时使用透视转化的方法对车牌进行校正，进而实现对车牌的定位与识别，该模型的模型召回率在99.7%，准确度84.4%，能够准确识别车牌号。本文先对研究的背景和内容进行了简单的介绍，接着对相关理论内容进行了概述，然后分析实验数据与结果，最后对本次研究进行总结。
In recent years, the number of vehicles has not increased significantly, and the continuous improvement of intelligent transportation sce-narios has made the importance of license plate recognition technology obvious in real life. This ar-ticle proposes a technology that combines YOLOv5 network and LPRNet network, and uses perspec-tive transformation method to correct the license plate, thereby achieving the positioning and recognition of the license plate. The model has a recall rate of 99.7%, an accuracy of 84.4%, and can accurately recognize the license plate. This article first provides a brief introduction to the back-ground and content of the research, followed by an overview of relevant theoretical content, analy-sis of experimental data and results, and finally a summary of this study.