摘要
交通信号灯检测和识别是无人驾驶和辅助驾驶领域的重要研究内容,能够避免在通过路口时由于交通信号灯判断失误导致的交通事故,提升驾驶的安全性。客观的复杂交通场景增加了检测识别算法难度。实现了基于Faster-RCNN的交通信号的检测识别,采集了交通场景数据进行标注,填充了国内交通信号灯公开数据集的空白。通过实验对比,选择最优的特征提取网络,并在智能车实验平台上验证了方法的有效性。
Traffic signal detection and recognition is an important research content in the field of manless-driving and driving-assist system.And the system can avoid traffic accident in intersection,improve the security of driving.However,exist various complicated natural scene factors increase the difficulty of recognition algorithm.The proposed alogrithm realizes traffic light detection and identification based on Faster-RCNN.Acquire traffic scene data and annotate,fill the blank of traffic lights public datasets.By experimental comparision,the optimal feature extraction network is chosed,the effectiveness of the method is verified on experimental platform of intelligent vehicle.
作者
潘卫国
陈英昊
刘博
石洪丽
PAN Weiguo;CHEN Yinghao;LIU Bo;SHI Hongli(Beijing Key Laboratory of Information Service Engineering,Beijing Union University,Beijing100101,China;College of Robotics,Beijing Union University,Beijing100101,China;College of Applied Science and Technology,Beijing Union University,Beijing100101,China)
出处
《传感器与微系统》
CSCD
2019年第9期147-149,160,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金青年科学基金资助项目(61802019)
北京市教育委员会科技计划资助项目(KM201711417005)
关键词
交通信号灯
深度学习
目标检测
traffic lights
deep learning
object detection