Traffic Light Recognition and Detection Based on Machine Learning

Abstract

In this thesis, the traffic light was chosen as the main research object, the YOLOv5 framework For detection and identification. By improving contextual feature extraction and Highlight the small target detection function and improve the accuracy of traffic lights. And it can effectively guarantee the detection time.About detecting traffic lights based on machine learning. The current use and significance of traffic lights is examined. Create your own data set and use it with the official data set. Looking for a lighter model.