基于深度学习PyTorch框架下YOLOv3的交通信号灯检测  

YOLOv3 Traffic Signal Detection Based on Deep Learning PyTorch Framework

作　　者：李蒋[1] LI Jiang(Suzhou Institute of Construction&Communications,Suzhou 215100,China)

机构地区：[1]苏州建设交通高等职业技术学校,江苏苏州215100

出　　处：《汽车电器》2022年第6期4-7,共4页Auto Electric Parts

摘　　要：随着《智能汽车创新发展战略》的发布,国家大力发展智能网联汽车势在必行。在自动驾驶技术中,复杂环境中目标物体的识别和判别是一项高难度的挑战,也是需要解决的重点任务之一。本文主要针对汽车智能化设计领域占有重要基础地位的图像识别功能,利用深度学习框架PyTorch和YOLOv3算法对于交通信号灯的判定进行学习训练,最终对获取的交通路况实景图进行识别,取得一定的实际检测效果,以期推动汽车智能化设计领域的技术创新和突破。With the release of the“Smart Vehicle Innovation Development Strategy”,it is imperative for the country to vigorously develop intelligent networked vehicles.In autonomous driving technology,the recognition and discrimination of target objects in complex environments is a difficult challenge and one of the key tasks that need to be solved.This article mainly focuses on the image recognition function that occupies an important basic position in the field of automotive intelligent design.The deep learning framework PyTorch and YOLOv3 algorithms learn and train on the judgment of traffic lights,and finally recognize the acquired real-life maps of traffic conditions,and obtain certain actual detection results,in order to promote technological innovation and breakthroughs in the field of intelligent vehicle design.

关 键 词：PyTorch YOLOv3 交通信号灯 

分 类 号：U463.6[机械工程—车辆工程]