Research on the application of deep learning based surface defect detection and treatment method for hot rolled strip steel

Abstract

Current research on the detection of surface defects in hot-rolled strip steel is mainly focused on defects in the hot-rolling stage, and there is less research on surface defects in hot-rolled flattened strip steel. In addition, some of the defects detected in the hot-rolled flattening stage can be dealt with by regrinding, so it is of great practical significance to conduct research on this issue. In this paper, we use ResNet152 and other deep learning methods for defect detection in the hot rolling flattening stage based on the actual industrial background. The experimental results on the XLData-CLS dataset show that the method in this paper achieves 97.64% classification accuracy, which meets the requirements of practical detection. In addition, the method in this paper can accurately locate defects on the strip steel surface and guide the regrinding robot to perform regrinding at the specified location.