A Method of Traffic Light Status Recognition Based on Deep Learning
Xinyuan Wang, Tao Jiang, Yurui Xie · 2018
Traffic light recognition plays an important role in the field of intelligent vehicles for safe driving. Driving with intelligent vehicles has been demonstrated as a trend for the following years. However, numbers of difficulties are currently existing in traffic light recognition, such as, the appearance of traffic light, illumination, and the bad weather, etc. By showing the potential challenges in traffic light recognition, this paper introduces a real-time traffic light status recognition method based on the combination of YOLOv3 and a lightweight Convolutional Neural Network (CNN). YOLOv3 performs traffic light ROI detection, the lightweight CNN is responsible for classifying traffic light status.