Table 5.
Performance comparison of different models in DPC-SF.
| Name | Accuracy (%) | Recall (%) | Precision (%) | F1-Score | AUC |
|---|---|---|---|---|---|
| VGGNet (42) | 38.9 | 33.4 | 40.4 | 0.365 | 0.478 |
| ResNet18 (10) | 87.5 | 91.3 | 90.8 | 0.910 | 0.967 |
| ResNet34 (10) | 74.5 | 77.6 | 78.3 | 0.779 | 0.902 |
| ResNet50 (10) | 76.5 | 80.0 | 79.6 | 0.797 | 0.912 |
| ResNet101 (10) | 79.8 | 83.2 | 83.3 | 0.832 | 0.931 |
| DenseNet (32) | 87.5 | 90.9 | 91.0 | 0.909 | 0.971 |
| SENet_18 (36) | 78.5 | 81.4 | 82.2 | 0.817 | 0.912 |
| ShuffleNet (34) | 71.7 | 75.1 | 74.5 | 0.747 | 0.889 |
| MobileNet V2 (33) | 89.7 | 93.1 | 93.2 | 0.931 | 0.978 |
| ECANet_18 (37) | 86.9 | 90.3 | 90.5 | 0.903 | 0.967 |
| HRNet (39) | 86.8 | 90.0 | 90.2 | 0.901 | 0.968 |
| GhostNet (35) | 86.6 | 89.9 | 90.1 | 0.899 | 0.971 |
| RepVGG (30) | 89.8 | 81.6 | 94.8 | 0.877 | 0.973 |
| Vision transformer (11) | 83.5 | 79.7 | 77.5 | 0.785 | 0.806 |
| CONVADD (15) | 82.9 | 85.5 | 85.6 | 0.855 | 0.946 |
| Efficient B2 (41) | 87.3 | 91.1 | 90.6 | 0.908 | 0.990 |
| Our proposed | 93.7 | 97.3 | 97.3 | 0.973 | 0.998 |