Metal shaft surface defect identification method and system based on deep learning

The invention relates to the technical field of data processing, in particular to a metal shaft surface defect identification method and system based on deep learning. The method comprises the following steps: carrying out asynchronous light source surface irradiation image data acquisition and gray level image data conversion on the metal uranium device by utilizing light source equipment with light source wavelength difference, and generating abnormal gray level image block data; performing physical modeling on the metal uranium surface image data of the metal uranium device by using a three-dimensional modeling technology, and performing abnormal region extraction by using abnormal gray level image block data to generate abnormal surface region image data; performing static metal uranium defect grade calculation on the abnormal surface area image data to generate a static metal uranium defect grade; and performing metal uranium surface dynamic defect prediction on the abnormal surface area image data by us