Article Details

Abstract

To address the problems that traditional steel surface defect detection relies on manual inspection, suffers from low efficiency, and is susceptible to subjective factors, this paper proposes a steel surface defect classification algorithm based on an improved YOLO11. The algorithm achieves precise classification of six defect types including cracks, inclusions, patches, pitting, oxidation, and scratches. Experimental results demonstrate that the improved YOLO11 classification algorithm excels in detecting these six types of defect. After 30 epochs of training, the model converges stably, with validation loss maintained below 0.05. The overall classification accuracy reaches 99.33%, and the recognition accuracy for the five defect types other than inclusions exceeds 99%. This provides a reliable solution for the automated detection of steel surface defects in industrial scenarios.

Keywords

multi-scale feature fusion; YOLO11; metal sheet; defect classification

References

1. [1] Zhou Lu, Wang Juanping Efficient Steel Surface Defect Detection Algorithm Based on BLE-YOLO [J]. Mechanical and Electrical Engineering Technology, 2026, 55 (1): 109-118
2. [2] Du Haotian, Li Yifan, Huo Ying, etc Application of Improved YOLOv5s in Surface Defect Detection of Steel [J]. Industrial Control Computer, 2025, 38 (12): 53-54, 57
3. [3] Wang Zekai, Jiang Qingchao, Yan Xuefeng Steel surface defect detection algorithm based on GSL-YOLO [J]. Journal of East China University of Science and Technology (Natural Science Edition), 2025, 51 (5): 657-670
4. [4] Yin Fu Research on Metal Surface Defect Detection Method Based on Computer Vision [D]. Shenyang: Shenyang University, 2024
5. [5] Zhao Shilong Research and Application of Deep Learning based Surface Defect Detection Method for Metal Products [D]. Jinan: Qilu University of Technology, 2025
6. [6] Wang Y，Wang H Y，Xin Z H. Efficient Detection Model of Steel Strip Surface Defects Based on YOLO-V7[J].IEEE Access，2022，10：133936-133944.
7. [7] Guo Beitao, Ye Shenghui Metal surface defect detection based on lightweight improvement of YOLOv5 [J]. Mechanical Engineer, 2025 (9): 1-4
8. [8] Longyang Research on the Improved YOLOv8 Model for Metal Surface Defect Detection [D]. Jingzhou: Changjiang University, 2025
9. [9] Tan Jie, Li Heng, Liu Hui Lightweight MSL-YOLO Metal Surface Defect Detection Algorithm [J/OL]. Optoelectronics and Laser, 2026:1-18 https://link.cnki.net/urlid/12.1182.o4.20260129.1712.008.
10. [10] Li Jian Research on Surface Defect Detection of Strip Steel Based on Improved YOLOv5 Algorithm [D]. Guiyang: Guizhou University, 2025

About the authors

Qu Jiang (March 1983-), female, Han ethnicity, from Yangzhong, Jiangsu, lecturer, master's student, research interests include computer science and technology and artificial intelligence.