A SURFACE DEFECT DETECTION METHOD OF THE MAGNESIUM ALLOY SHEET BASED ON DEFORMABLE CONVOLUTION NEURAL NETWORK

In the rolling process of the magnesium alloy sheet, due to improper control parameters or inaccurate production equipment and other reasons, the surface of the magnesium alloy sheet is prone to appearance of edge crack, fold, inclusion, ripple, scratch and other defects. In order to improve the surface quality of the magnesium alloy sheet, a surface defect detection method based on deformable convolution neural network is proposed in the paper, which presents a higher detection accuracy than those traditional methods on the convolutional neural network (CNN), support vector machine (SVM) and Bayes. The experiment result shows the final detecting accuracy is greater than 95 %.