Research on traffic light target detection method based on improved YOLOv5 algorithm

Aiming at the real-time requirement of traffic light target detection in assisted driving, a traffic light target detection model based on improved YOLOv5 algorithm is established. The improved model is improved from two aspects of the original YOLOv5 model: C3 module and feature extraction network. The feature fusion mode of C3 module is changed from concat operation to add operation, and the Focus module in the feature extraction network is eliminated. The parameter configuration is optimized to avoid frequent slicing operations, and the parameters of the model are greatly reduced. The improved YOLOv5-lite model improves the processing speed while maintaining the detection accuracy, and is easier to deploy on the vehicle platform.