Partitioned intelligent identification method for particle components on surface of weir plug body

The invention discloses a barrier body surface particle component partition intelligent identification method, which comprises the following steps of S1, acquiring a barrier body surface image to obtain a barrier body surface image training set which comprises a first training subset, a second training subset and a third training subset; s2, inputting the first training subset and the second training subset into a YOLO5 network to obtain barrier body surface image parameters; and S3, setting a YOLO5 network according to the parameters of the surface image of the barrier body, and performing particle contour recognition on the input third training subset through the YOLO5 network to obtain a grading curve of the surface image of the barrier body so as to complete barrier body surface particle component partition recognition. According to the method, when the YOLO5 network is established, the feature pyramid module and the boundary optimization module are designed by using multi-scale and global context informa