Industrial defect target detection based on YoloV5 and attention mechanism

Chunzhao Ye

doi:10.1117/12.2653431

18 November 2022 Industrial defect target detection based on YoloV5 and attention mechanism

Chunzhao Ye

Proceedings Volume 12473, Second International Conference on Optics and Communication Technology (ICOCT 2022); 124730X (2022) https://doi.org/10.1117/12.2653431
Event: Second International Conference on Optics and Communication Technology (ICOCT 2022), 2022, Hefei, China

Abstract

Crack detection of industrial defect target detection is one of the most critical aspects of industrial product quality control, and to address the problems of false detection, missed detection and insufficient feature extraction for fine cracks in target detection, this paper introduces a hybrid attention mechanism based on the original YOLOv5, which improves the accuracy of the backbone feature extraction network for fine crack detection. The experimental results show that the target loss of the validation set of the improved YOLOv5s model converges significantly, the model training results are accurate, there is no overfitting or underfitting phenomenon, and the average accuracy mean value is improved by 3.8% compared with the original YOLOv5s model. The improved YOLOv5s model can identify and detect fine cracks under both illumination or dim conditions, and the model generalization ability is good enough to meet the relevant requirements in industrial production processes.

Citation Download Citation

Chunzhao Ye "Industrial defect target detection based on YoloV5 and attention mechanism", Proc. SPIE 12473, Second International Conference on Optics and Communication Technology (ICOCT 2022), 124730X (18 November 2022); https://doi.org/10.1117/12.2653431

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