Rail surface anomaly detection based on deep learning

Lei Shi; Junjie Wu; Yongkui Sun

doi:10.1117/12.2689593

7 September 2023 Rail surface anomaly detection based on deep learning

Lei Shi, Junjie Wu, Yongkui Sun

Author Affiliations +

Proceedings Volume 12790, Eighth International Conference on Electromechanical Control Technology and Transportation (ICECTT 2023); 1279031 (2023) https://doi.org/10.1117/12.2689593
Event: 8th International Conference on Electromechanical Control Technology and Transportation (ICECTT 2023), 2023, Hangzhou, China

Abstract

The role of the rail is to directly bear the weight of the train and its load transmitted by the wheels, and to guide the direction of the train. Prolonged train travel can cause unavoidable defect to the rails; however, China's railways are developing at a high speed, and rail damage will also increase. In order to ensure the safety of railway operation, the detection of rail damage must be fast and efficient. So, a method is needed to quickly and accurately detect the type of rail defect. Due to the superiority of machine vision detection, a rail surface anomaly detection method based on deep learning is proposed. Firstly, the image data features are extracted through the image annotation tool labeling. After that, the YOLO v5 deep learning network is trained by the training set. Finally, the accuracy of the trained network is checked, and the relevant parameters are modified to adjust the accuracy to the best. The overall rail defect detection accuracy reaches 97%, indicating the feasibility and effectiveness of the proposed method.

Citation Download Citation

Lei Shi, Junjie Wu, and Yongkui Sun "Rail surface anomaly detection based on deep learning", Proc. SPIE 12790, Eighth International Conference on Electromechanical Control Technology and Transportation (ICECTT 2023), 1279031 (7 September 2023); https://doi.org/10.1117/12.2689593

ACCESS THE FULL ARTICLE

INSTITUTIONAL
Select your institution to access the SPIE Digital Library.

SELECT YOUR INSTITUTION

PERSONAL
Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.

PERSONAL SIGN IN

No SPIE Account? Create one

;

PURCHASE THIS CONTENT

SUBSCRIBE TO DIGITAL LIBRARY

50 downloads per 1-year subscription

Members: $195

Non-members: $335 ADD TO CART

25 downloads per 1 - year subscription

Members: $145

Non-members: $250 ADD TO CART

PURCHASE SINGLE ARTICLE