Abstract: In order to effectively resolve the problems of high safety risks, high labor costs, and low efficiency in the manual inspection process of the appearance of coal transportation vehicles, the YOLOv5 algorithm-based coal transportation vehicle appearance detection system adopted by Mengjiahao Station of Shaanxi Yuheng Railway Co.,Ltd.,can not only accurately identify abnormal information on the vehicle, such as abnormal gaps at the doors, damaged bodies, missing key components of the body, abnormal conditions of key parts of the body, abnormal conditions at the top, and other abnormalities; it can also identify text information on the vehicle, such that of the vehicle type, vehicle number, the time of the last maintenance, the time of the next maintenance, the time of the last plant maintenance, the time of the next plant maintenance, the payload, the self-weight, the volume, and the length variation.This system has performed excellently at Mengjiahao Station of Shaanxi Yuheng Railway Company, achieving an accuracy rate of over 95% while maintaining a good recall rate.