Abstract: This paper addresses the issue of surface defect detection in steel bars, proposing a one-class method based on self-supervised learning and thoroughly explores its effectiveness in online ...