基于深度学习的计算机视觉技术在交通场景中的应用  

Application of computer vision technology based on deep learning in traffic scene

作　　者：潘超[1] 王雪涵 高俊平 王赢庆 尹栋程 李佳 肖巍[1] PAN Chao;WANG Xuehan;GAO Junping;WANG Yingqing;YIN Dongcheng;LI Jia;XIAO Wei(School of Computer Science&Engineering,Changchun University of Technology,Changchun 130102,China;Beihu School of Changchun No.11 High School,Changchun 130102,China)

机构地区：[1]长春工业大学计算机科学与工程学院,吉林长春130102 [2]长春市十一高中北湖学校,吉林长春130102

出　　处：《长春工业大学学报》2022年第3期251-257,共7页Journal of Changchun University of Technology

基　　金：吉林省教育厅科学研究项目(JJKH20220691KJ)。

摘　　要：基于深度学习的计算机视觉技术实现了采用YOLO V3神经网络模型对视频中交通目标物体进行检测、识别与分析。实现车牌识别、路口饱和度计算、机动车违章判断等功能。结果表明,检测结果准确度高、检测速度较快,达到了交通自动化监管的目的。Based on the computer vision technology of deep learning,this paper realizes the system of detecting,recognizing and analyzing the target object in video by using Yolo V3 neural network model.This page realizes the functions of license plate recognition,intersection saturation calculation,motor vehicle violation judgment and so on.It is proved by the experiment that the design is practicable,accurate and fast,and that it can contribute to automizing the management of traffic.

关 键 词：计算机视觉 交通 YOLO V3 深度学习 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]