一种改进YOLOv7-GCA的车型快速识别方法  

A Fast Identification Method for Vehicle Model Based on Improved YOLOv7-GCA

作　　者：斯洪云 苏盈盈 邓圆圆 阎垒 杨浩军 SI Hongyun;SU Yingying;DENG Yuanyuan;YAN Lei;YANG Haojun(School of Electrical Engineering,Chongqing University of Science and Technology,Chongqing 401331,China)

机构地区：[1]重庆科技大学电气工程学院,重庆401331

出　　处：《重庆科技学院学报（自然科学版）》2024年第3期87-92,共6页Journal of Chongqing University of Science and Technology：Natural Sciences Edition

基　　金：重庆市教育委员会科学技术研究项目“面向智能化工厂转型的通用型指针式仪表识别方法及实现”(KJQN202101510);重庆科技大学硕士研究生创新计划项目“基于YOLOv7的智能交通监测系统”(YKJCX2220419);“基于改进YOLOv7的带钢缺陷检测系统设计及其实现”(YKJCX2220408);重庆科技大学本科生创新计划项目“智能交通监测系统设计及实现”(2023010)。

摘　　要：针对道路车流量大、车型识别速度慢等问题,提出了一种改进YOLOv7-GCA的车型快速识别方法。首先,采用参数量更小、检测速度更快的轻量化卷积GhostConv替换网络中的普通卷积Conv,以提高车型识别速度;其次,为了保证模型的识别精度,在进入颈部前引入CA注意力机制模块。实验结果表明,YOLOv7-GCA模型在保证识别精度的前提下,减少了模型的参数量和复杂度,提高了车型识别速度。Aiming at the problems of high road traffic flow and slow model identification,a fast identification method for vehicle model based on improved YOLOv7-GCA is proposed.Firstly,the lightweight GhostConv with smaller parameter count and faster detection speed is used to replace the ordinary convolutional Conv in the network to improve the speed of vehicle model identification;secondly,in order to ensure the recognition accuracy of the model,the CA attention mechanism module is introduced before entering the neck.The experimental results show that the YOLOv7-GCA model reduces the number of parameters and complexity of the model and improves the vehicle model recognition speed while ensuring the identification accuracy.

关 键 词：YOLOv7 轻量化 GhostConv CA注意力机制 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]