For access to this article, please select a purchase option:
IET members benefit from discounts to all IET publications and free access to E&T Magazine. If you are an IET member, log in to your account and the discounts will automatically be applied.
Your recommendation has been sent to your librarian.
High-speed rails are important tools for traveling across the country. Because wheel treads, which are the running parts of high-speed trains, are vulnerable to damage, their condition is vital to the safety of high-speed trains. Stains similar to defects can be found on the surfaces of wheels and rails, and images containing stains are easier to obtain than images containing defects. Therefore, a wheel tread defect detection method based on Faster R-CNN is proposed in this paper to detect defects and stains simultaneously in the case of imbalanced samples, thus reducing the difficulty of defect identification. Experimental results show that, when the ratio of total defects to stains on the image is 5:1 and 2.7:1, the detection accuracy of wheel tread defects can be respectively improved by 23.62% and 19.73% when the proposed algorithm is adopted.
Inspec keywords: railway engineering; learning (artificial intelligence); automatic optical inspection; rails; railway safety; wheels
Subjects: Optical, image and video signal processing; Inspection and quality control; Machine learning (artificial intelligence); Computer vision and image processing techniques; Computing in other engineering fields