Railway Foreign Object Intrusion Detection based on Deep Learning
Xuewen Ding, Xinnan Cai, Ziyi Zhang, Wenyan Liu, Wenwen Song · 2022 International Conference on Computer Engineering and Artificial Intelligence (ICCEAI) · 2022
To detect foreign objects in the intrusion orbit and prevent railway safety accidents caused by foreign matter intrusion, the railway foreign objects intrusion detection algorithms based on neural network were studied in the paper. First, according to the common the railway foreign body intrusion condition, the foreign body intrusion image data were collected, cleaned and labeled. The railway foreign body intrusion data set was constructed. Then the YOLOv5 detection model of deep learning was built and trained on the established data set. Finally the performance of the detection model was verified by the test data set and compared with the popular network model. The experiment results show a good mean average accuracy score is obtained, which basically meets the safety requirements of train operation. The algorithm has important academic value and reference significance for the application of railway track foreign object detection.