An Modified YOLOv5 Algorithm to Improved Image Identification for Autonomous Driving

In this paper we propose a solution to improve image identification on the early expressway engineering warning signs for autonomous driving. This scheme uses the modified lightweight YOLOv5 algorithm model to train feature categories. After manual labeling and data augmentation, the dataset is sent to the advance deep neural network for training. The practical experimental results show that the modified algorithm model can effectively identify engineering warning signs on the expressway. It allows drivers and construction units on the expressway to use the road more safely.