4 Si3N4 ceramic bearing inner
ring surface defect identification and classification model
The experimental platform of Si3N4ceramic bearing inner ring surface defect image detection is built by
ourselves, and the surface defect images are obtained and made into
dataset. As shown in Figure 4, the completion of surface defect
recognition and classification is divided into four processes, which are
preprocessing and Labelling, data augmentation, feature extraction and
defects classification recognition.