文献类型：会议论文

英文题名：Traffic Light Detection for Self-Driving Vehicles Based on Deep Learning

作者：Pan, WeiGuo[1];Chen, YingHao[2];Liu, Bo[2]

第一作者：潘卫国

通讯作者：Pan, WG[1]

机构：[1]Beijing Union Univ, Beijing Key Lab Informat Serv Engn, Coll Robot, Beijing, Peoples R China;[2]Beijing Union Univ, Coll Appl Sci & Technol, Beijing, Peoples R China

第一机构：北京联合大学北京市信息服务工程重点实验室|北京联合大学机器人学院

通讯机构：[1]corresponding author), Beijing Union Univ, Beijing Key Lab Informat Serv Engn, Coll Robot, Beijing, Peoples R China.|[1141739]北京联合大学机器人学院;[11417]北京联合大学;[11417103]北京联合大学北京市信息服务工程重点实验室;

会议论文集：15th International Conference on Computational Intelligence and Security (CIS)

会议日期：DEC 13-16, 2019

会议地点：Macao, PEOPLES R CHINA

语种：英文

外文关键词：self-driving; deep learning; traffic light

摘要：Research on traffic light detection and semantics is important in the field of intelligent vehicles. Better detection and clearer semantics can help prevent traffic accidents by self-driving vehicles at busy intersections and thus improve driving safety. However, complex traffic scenes increase the difficulty of detection and recognition algorithm. This paper presents a method to detect and recognize traffic lights based on Faster RCNN. Our work includes an annotated collection of traffic scene image data, which fills a gap in public traffic light data sets. An optimal feature extraction network is selected through experimental comparisons. Effectiveness of the method has been verified on a self-driving vehicle platform.