A YOLOV8-based approach for steel plate surface defect detection

Abstract

Hot-rolled steel strips are a commonly used product in both production and daily life. However, the manufacturing process inevitably leads to the occurrence of surface defects. To solve this problem, Our method uses YOLOV8 and squeeze-and-excitation (SE) attention mechanism to detect surface defects in hot-rolled steel strips. Our method balances accuracy and real-time performance, while detecting four common surface defects. The method has an average accuracy of 90,9 % and a maximum accuracy of 98,5 % for detecting a single category of surface defects. Experimental results confirm good performance of our proposed method in classifying and localizing surface defects in hot-rolled steel strips, and has the potential for broad application and promotion.

Keywords

• steel strip
• hot-rolled
• surface defect
• object detection
• YOLOV8
• SE attention mechanism