Original scientific paper

A method for detecting surface defects in hot-rolled strip steel based on deep learning

H. Ren ; School of Computer Science and Software Engineering, University of Science and Technology Liaoning, China
Y. J. Zhang ; School of Computer Science and Software Engineering, University of Science and Technology Liaoning, China *
J. T. Chen ; School of Computer Science and Software Engineering, University of Science and Technology Liaoning, China
X. N. Wei ; School of Computer Science and Software Engineering, University of Science and Technology Liaoning, China
H. K. Chen ; School of Computer Science and Software Engineering, University of Science and Technology Liaoning, China
P. Liu ; School of Computer Science and Software Engineering, University of Science and Technology Liaoning, China

* Corresponding author.

Abstract

Hot-rolled strip steel is a material widely used in production activities and daily life. However, the appearance of surface defects during its production process is inevitable. To address this issue, we introduce a new detection method using Gold-Yolo to detect surface defects on hot-rolled strip steel. Our method effectively balances accuracy and real-time performance while detecting four common types of surface defects, achieving an average accuracy rate of 82,2 % for detecting individual types of surface defects. Experimental data prove that our method excels in classifying and locating surface defects on hot-rolled steel strip, demonstrating broad application prospects and promotional value.

Keywords

steel strip; hot-rolled; surface defect; object detection; Gold-Yolo

Hrčak ID:

315692

URI

https://hrcak.srce.hr/315692

Publication date:

1.7.2024.

Visits: 1.204 *