Abstract: Aiming at the problems of strong scene dependence and poor real-time performance in traditional vehicle detection methods, a front vehicle recognition system based on YOLOv7 deep learning model is proposed. Firstly, the limitations of traditional vehicle detection methods are analyzed, including two types of methods based on image processing technology and feature extraction. Then, it introduces the theoretical basis of deep learning, focusing on the system framework of YOLOv7 model. Finally, it carried out experiments on COCO data set. The results show that the average accuracy of YOLOv7 model in vehicle recognition task reaches 69.7%, which significantly improves the detection accuracy and real-time performance compared with traditional methods. This study provides an effective technical solution for vehicle detection in intelligent transportation systems.

Key words: vehicle recognition; image processing; deep learning; YOLOv7

摘要： 针对传统车辆检测方法存在的场景依赖性强、实时性差等问题，提出基于 YOLOv7 深 度学习模型的前方车辆识别系统。首先分析传统车辆检测方法的局限性，包括基于图像处理 技术和特征提取的两类方法；然后介绍深度学习理论基础，重点研究 YOLOv7 模型的系统框 架；最后在 COCO 数据集进行实验验证。结果表明，YOLOv7 模型在车辆识别任务中平均精 度达到 69.7%，相比传统方法显著提高了检测准确率和实时性。研究为智能交通系统中的车 辆检测提供了有效的技术方案。

关键词: 车辆识别；图像处理；深度学习；YOLOv7