Steel surface defect detection algorithm combined with attention mechanism

Siying Wang; Fu Su; Xie Zhang; Kainan Chen

doi:10.1117/12.2634420

4 May 2022 Steel surface defect detection algorithm combined with attention mechanism

Siying Wang, Fu Su, Xie Zhang, Kainan Chen

Proceedings Volume 12172, International Conference on Electronic Information Engineering and Computer Communication (EIECC 2021); 1217218 (2022) https://doi.org/10.1117/12.2634420
Event: International Conference on Electronic Information Engineering and Computer Communication (EIECC 2021), 2021, Nanchang, China

Abstract

To solve the problem of low detection accuracy of steel surface defects due to background interference and various target shapes, a steel surface defect detection algorithm with attention mechanism is proposed to improve detection accuracy. In view of the small proportion of the target defect area in the overall image and background interference, a two-way attention module (TWA-Block) is proposed to establish the long-distance dependence of the spatial domain and channel domain features. It enhances the contour and texture features of defect area in shallow features, and suppresses the background to a certain extent. The experimental results show that the average accuracy (MAP) of the YOLOv3 model fused with the attention mechanism on the NEU-DET dataset reaches 79.5%, which is 14.4% higher than the YOLOv3 algorithm. Compared with the standard steel surface defect detection methods, the algorithm effectively improves the detection accuracy.

Citation Download Citation

Siying Wang, Fu Su, Xie Zhang, and Kainan Chen "Steel surface defect detection algorithm combined with attention mechanism", Proc. SPIE 12172, International Conference on Electronic Information Engineering and Computer Communication (EIECC 2021), 1217218 (4 May 2022); https://doi.org/10.1117/12.2634420

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

Includes PDF, HTML & Video, when available