基于改进 YOLOv5 的行车目标检测与分割研究

巫 少方, 陈 朝大, 陈 浩霖
广州航海学院 船舶与海洋工程学院 广东广州 510725

摘要

行车道路目标检测与分割是计算机视觉领域的重要研究课题。为更准确且快速的检测汽车、车道线等目标，本文提出了一种基于改进 YOLOv5 算法的目标检测模型进行道路目标检测与语义分割的方法。在 YOLOv5 模型中利用全卷积神经网络（Fully Convolutional Networks, FCN）进行道路目标的检测及语义分割。实验结果显示对道路的目标特征的预测精度达 88%，召回率达 99%，实现了快速准确的道路目标的检测与分割，有效提高了行车目标检测效果。

关键词

行车道路;目标检测;语义分割;YOLOv5

参考

[1]赵麒博,王梦梅,潘海英,李晨欢,吴祥龙,郭晨阳. 基于改进YOLOv5 和 DeepSort 的 车 道 线 检 测 方 法 [J]. 信 息 技 术 与 信 息化,2023,(07):217-220.ZHAO Qibo, WANG Mengmei, Pan Haiying, Li Chenhuan, WU Xianglong, Guo Chenyang. Lane detection Method based on improved YOLOv5 and DeepSort [J]. Information Technology and Informatization,2023,(07):217-220.[2]邓亚平,李迎江. YOLO 算法及其在自动驾驶场景中目标检测研究综述[J]. 计算机应用:1-12.Deng Yaping, Li Yingjiang. YOLO Algorithm and its Application to target Detection in Autonomous Driving Scenarios [J]. Computer Applications :1-12.[3]张凯祥,朱明. 基于 YOLOv5 的多任务自动驾驶环境感知算法[J]. 计算机系统应用,2022,31(09):226-232.Zhang Kaixiang, Zhu Ming. Environment Awareness Algorithm for Multi-Task Automatic Driving Based on YOLOv5 [J]. Computer System Applications, 2002,31(09):226-232. [4]杨奎河,张宇. 交通场景下基于 YOLOv5 的目标检测与分割算法[J]. 长江信息通信,2023,36(04):48-50.Yang Kuihe, ZHANG Yu. Object Detection and Segmentation Algorithm Based on YOLOv5 in Traffic Scenario [J]. Changjiang Information and Communication, 2019,36(04):48-50.[5]王军,葛宝康,程勇. 基于改进 YOLOv5s 算法的交通信号灯检测[J]. 计算机系统应用:1-10.Wang Jun, GE Bao-Kang, CHENG Yong. Traffic light Detection Based on improved YOLOv5s algorithm [J]. Computer System Applications :1-10.[6]彭瑾,桑正霄,李木易. 一种基于 YOLOv5s 的交通标志检测算法[J]. 自动化技术与应用,2023,42(09):53-57.Peng Jin, SANG Zhengxiao, LI Muyi. A Traffic sign detection Algorithm based on YOLOv5s [J]. Automation Technology and Application, 2019,42(09):53-57.[7]L. Xiaomeng, F. Jun and C. Peng. Vehicle Detection in Traffic Monitoring Scenes Based on Improved YOLOV5s. 2022 International Conference on Computer Engineering and Artificial Intelligence (ICCEAI), Shijiazhuang, China, 2022, pp. 467-471.[8]X. Zhang, H. Song, F. Wan and X. Yang. A Pedestrian Detection Method Based on Improved YOLOv5s. 2022 International Conference on Cloud Computing, Big Data Applications and Software Engineering (CBASE), Suzhou, China, 2022, pp. 197-201.[9]B. Xiao, J. Guo and Z. He. Real-Time Object Detection Algorithm of Autonomous Vehicles Based on Improved YOLOv5s. 2021 5th CAA International Conference on Vehicular Control and Intelligence (CVCI), Tianjin, China, 2021, pp. 1-6.[10]宁俊彦. 自动驾驶复杂场景下目标检测研究[D].西安工业大学,2023.Ning Junyan. Research on Object Detection in Complex Autonomous Driving Scenarios [D]. Xi 'an Technological University,2023.[11]刘昌华. 复杂交通场景下自动驾驶道路目标检测[D].大连理工大学,2022.Liu Changhua. Automatic Driving Road Object Detection in Complex Traffic Scenarios [D]. Dalian University of Technology,2022.[12]Z. Lu, L. Ding, Z. Wang, L. Dong and Z. Guo. Road Condition Detection Based on Deep Learning YOLOv5 Network. 2023 IEEE 3rd International Conference on Electronic Technology, Communication and Information (ICETCI), Changchun, China, 2023, pp. 497-501.