Review of Steel Surface Defect Detection and Lightweighting Methods

DOI: https://doi.org/10.62517/jiem.202503412

Author(s)

Miao Chen

Affiliation(s)

Jilin International University - School of Artificial Intelligence, Changchun, China *Corresponding Author

Abstract

Surface defects of steel materials affect product quality and industrial safety. High-precision and high-efficiency detection of surface defects in steel is a necessary condition for the high-quality development of the steel industry. Therefore, this paper reviews the research work related to steel surface defect detection and lightweighting. Starting from the data sets, it then elaborates on the defect development of traditional detection technologies, followed by a summary of detection based on machine learning and deep learning. The optimization approaches and lightweighting technologies represented by the YOLO series, such as structural optimization, model compression, and auxiliary strategies, are emphasized and elaborated on as the key points. Moreover, YOLO series model cases are used for illustration, and the effects after different improvement methods in the three cases are analyzed and compared. The YOLOv8-CSG has a better balance: the computational cost is reduced by 37% and the parameter quantity is reduced by 35.2%; for the improvement of the YOLOv8 model, the optimal compression is mentioned: the model size is reduced by 44%, and the lightweighting methods have improved the detection accuracy, detection efficiency, and real-time performance. Finally, the relevant technologies for detecting surface defects of steel were briefly summarized, and the future development of these technologies was also prospected.

Keywords

Steel Surface; Defect Detection; Lightweighting; Deep Learning

References

[1] Song Yubin, Kong Weibin, Chen Xi, et al. Review on Research on Detection of Surface Defects of Steel [J]. Software Review, 2024, 23(03): 203-211. [2] Li Yanshun. Research on Steel Surface Defect Detection Based on Deep Learning [D]. Shandong Jiaotong University, 2024. [3] Zhao Yufeng. Steel Surface Defect Detection Based on Deep Learning [D]. Shenyang University of Technology, 2024. [4] Li Yuandong, He Zhiming. An Improved YOLOv8 Algorithm for Steel Surface Defect Detection [J]. Software Review, 2024, 23(11): 200-205. [5] You Qinghua. Research on Steel Surface Defect Detection Method Based on Deep Learning [D]. Wuhan Textile University, 2021. [6] Hu Mingqi, Chen Huiming, Xu Wei, et al. Lightweight YOLOv8 Steel Surface Defect Detection Incorporating MobileNetv3 [J]. Science & Technology & Engineering, 2025, 25(16): 6831-6840. [7] Ma Rui, Fu Xingkui, Zheng Guijun, et al. Steel Surface Defect Detection Based on Improved YOLOv5 Algorithm [J]. Internet of Things Technology, 2025, 15(03): 39-43. [8] Jiang Bo, Wan Yi, Xie Xianzhong. A Lightweight Steel Surface Defect Detection Model Based on Improved YOLOv5s [J]. Computer Science, 2023, 50(S2): 271-277. [9] Ma Yanting. Research on Steel Surface Defect Detection Technology Based on Deep Learning [D]. Hebei University of Technology, 2023. [10] Huang Aoguo, Luo Xiaoling, Pan Xin. STC-YOLOv8: A Detection Method for Steel Surface Defects in Complex Backgrounds [J]. Journal of Inner Mongolia Agricultural University (Natural Science Edition), 2024, 45(05): 67-76. [11] Zhao Baiting, Zhang Min, Jia Xiaofen. YOLOv8n-CSG: A Lightweight Algorithm for Detecting Surface Defects of Steel Materials [J/OL]. Journal of Electronic Measurement and Instrumentation, 1-13 [2025-10-11] [12] Chang Le, Liu Guiru, Wang Lu Lin, et al. A Lightweight YOLOv8 Algorithm for Detecting Surface Defects of Steel Materials [J/OL]. Journal of Tianjin University of Technology, 1-10 [2025-10-11]