Table 6.
Comparisons results of the various TL-based DL models.
| State-of-the-Art | Classes | Augment | Epochs | 1 Time (S) | 1 ACC | F1-Score | 1 MCC |
|---|---|---|---|---|---|---|---|
| Light-Dermo Model | 7 | Yes | 40 | 2.4 | 98.1% | 98.1% | 98.1% |
| 9CNN models [6] | 7 | Yes | 40 | 12 | 80.5% | 80.5% | 82.5% |
| AlexNet [8] | 7 | Yes | 40 | 17 | 81.9% | 81.9% | 81.9% |
| MobileNet-LSTM [10] | 7 | Yes | 40 | 13 | 82.3% | 82.3% | 80.3% |
| FixCaps [11] | 7 | Yes | 40 | 15 | 84.8% | 84.8% | 83.8% |
| EfficientNet [12] | 7 | Yes | 40 | 18 | 75.4% | 75.4% | 74.4% |
| CNN-Leaky [13] | 7 | Yes | 40 | 20 | 76.5% | 76.5% | 75.5% |
| DCNN [15] | 7 | Yes | 40 | 22 | 77.9% | 77.9% | 76.9% |
1 ACC: Accuracy, MCC: Matthew’s correlation coefficient, S: Seconds.