Surface defect detection of rolled steel based on YOLOX_M model

Ziyu Wei; Fenggui Wang; Xiangdong Li; Yizhang Li

doi:10.1117/12.2652486

20 October 2022 Surface defect detection of rolled steel based on YOLOX_M model

Ziyu Wei, Fenggui Wang, Xiangdong Li, Yizhang Li

Proceedings Volume 12350, 6th International Workshop on Advanced Algorithms and Control Engineering (IWAACE 2022); 123501N (2022) https://doi.org/10.1117/12.2652486
Event: 6th International Workshop on Advanced Algorithms and Control Engineering (IWAACE 2022), 2022, Qingdao, China

Abstract

Aiming at the technical problems of intelligent recognition and accurate positioning of steel rolling surface defects, a target detection method based on machine vision and depth neural network was proposed. YOLOX_M was introduced as the model of surface defect detection using the weights trained on the COCO dataset as the initial weights. To realize the identification and location of surface defect categories of rolled steel, the YOLOX_M model was further trained using the practical dataset. The performance of YOLOX_M was compared with the other five YOLOX models. The test results show that YOLOX_M can effectively detect six different forms of surface defects, and the test accuracy (P), recall rate (R) and detection mAP can respectively reach 88.81%, 80.88% and 90.12%. The mAP of the YOLOX_M model is higher than 90% and the model size is less than 100 MB, so it can be better applied in the embedded system for real-time detection.

Citation Download Citation

Ziyu Wei, Fenggui Wang, Xiangdong Li, and Yizhang Li "Surface defect detection of rolled steel based on YOLOX_M model", Proc. SPIE 12350, 6th International Workshop on Advanced Algorithms and Control Engineering (IWAACE 2022), 123501N (20 October 2022); https://doi.org/10.1117/12.2652486

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