SF-YOLONet metal gear end face defect detection method and system

The invention belongs to the technical field of image processing detection, and particularly relates to an SF-YOLONet metal gear end face defect detection method and system. Firstly, a metal gear end face defect detection image is obtained; secondly, carrying out image preprocessing on the original image by utilizing an improved SF algorithm; marking the defect type and the defect position in the processed image; expanding data by using a data enhancement technology, and making the data into a data set; then, improvement is carried out based on a YOLOv5s network model, and a repeated weighted bidirectional multi-scale fusion strategy and a self-adaptive convolution attention mechanism module are added; and finally, testing the model by using a test set in a metal gear end face defect data set, and the result shows that the average precision of the SF-YOLONet model on the gear end face defect test set reaches 98.01%, the F1 value is 0.99, and the FPS value is 40. The invention provides a detection method for s