Fused bead electron microscope image crystal grain detection method and system based on deep learning

The invention discloses a molten bead electron microscope image crystal grain detection method and system based on deep learning, and belongs to the technical field of molten bead electron microscope image crystal grain detection, and the method comprises the steps: marking a crystal grain in a collected molten bead electron microscope image through a polygon, carrying out the normalization processing of the x and y coordinates of each vertex of the polygon, and obtaining the polygon information of the crystal grain; based on crystal grain polygon information, training a YOLOv8 neural network model by obtaining pixel information of crystal grains, and constructing an instance segmentation model; collecting a to-be-detected molten bead electron microscope image, and performing crystal grain detection on the to-be-detected molten bead electron microscope image through the instance segmentation model; the method provides an important index for judging the type of the molten bead, provides a scientific and object