Table 3 Abnormal detection and localization results on magnetic tiles.
From: A novel dual-student reverse knowledge distillation method for magnetic tile defect detection
Model | Image Auroc | Pixel Auroc | Pixel Aupro | Image F1 | Image ACC | Flops(G) | Parameters(M) |
|---|---|---|---|---|---|---|---|
FOD | 0.848 | 0.779 | 0.683 | 0.806 | 0.885 | 37.49 | 42.61 |
CKAAD | 0.854 | 0.697 | 0.742 | 0.889 | 0.816 | 26.87 | 2.97 |
SimpleNet | 0.768 | 0.706 | 0.712 | 0.873 | 0.786 | 150.72 | 72.82 |
PBAS | 0.871 | 0.692 | 0.723 | 0.897 | 0.832 | 108.13 | 27.66 |
FastFlow | 0.884 | 0.732 | 0.709 | 0.901 | 0.832 | 9.12 | 5.57 |
RD4AD | 0.947 | 0.717 | 0.703 | 0.947 | 0.918 | 162.22 | 89.58 |
IKD | 0.965 | 0.815 | 0.751 | 0.929 | 0.954 | 164.92 | 59.09 |
BSDRD | 0.974 | 0.802 | 0.793 | 0.972 | 0.957 | 133.23 | 101.08 |
