Design of a Subway Train Fault Detection System based on Yolov7 and EfficientNet Networks

Due to the current issues of dim lighting, fatigue, and high false detection rates in manual visual inspection of subway trains, a train bottom fault detection system has been designed to address these issues. The robot can automatically operate under the train, capture images of target inspection points through the robot vision system, locate bolts using an improved Yolov7 algorithm, and then use the EfficientNet+B3 model to identify bolt faults. The experimental results show that the accuracy of the designed method for locating bolts is 0.9942 , and the accuracy of identifying bolt faults is 0.979, demonstrating excellent fault recognition performance.