METHOD FOR IDENTIFYING AND DETECTING EDGES AND CORNERS OF SEED CRYSTAL SINGLE CRYSTAL WIRE IN CRYSTAL GROWTH PROCESS

The present application relates to the technical field of identification during crystal growth, and specifically relates to a method for identifying and detecting the edges and corners of a seed crystal single crystal wire in a crystal growth process, comprising four steps, i.e., constructing a high-quality data set, constructing an artificial intelligence algorithm model, performing model training, and performing model evaluation and verification. The present invention overcomes the defects of manual determination of the starting time of a seeding procedure, and provides a method for automatically determining the seeding time, so that a control system automatically controls the operation of starting the seeding procedure after temperature adjustment is ended. Specifically, in the present invention, a large batch of image data of related scenarios is used for training by means of a deep learning algorithm to obtain an optimal model, so that the form of the edges and corners of a seed crystal single crystal wi