Table 10 Comparison of Student with knowledge distillation model cross-validation on APTOS 2019, and Primary Dataset.
From: Knowledge distillation-based lightweight MobileNet model for diabetic retinopathy classification
Dataset | Fold (k) | Accuracy(%) | Precision(%) | Recall (%) | F1 Score (%) |
|---|---|---|---|---|---|
APTOS 2019 | 1 | 87.64 | 89.28 | 87.64 | 88.23 |
2 | 92.73 | 92.74 | 92.73 | 92.73 | |
3 | 90.55 | 90.26 | 90.55 | 90.38 | |
4 | 87.09 | 88.68 | 87.09 | 87.26 | |
5 | 93.09 | 93.06 | 93.09 | 93.07 | |
Mean | 90.22 | 90.80 | 90.22 | 90.33 | |
Std | \(\pm 2.63\) | \(\pm 1.90\) | \(\pm 2.63\) | \(\pm 2.38\) | |
Primary Dataset | 1 | 83.67 | 84.02 | 83.67 | 83.58 |
2 | 85.03 | 85.77 | 85.03 | 84.63 | |
3 | 85.03 | 84.85 | 84.03 | 84.65 | |
4 | 85.71 | 85.62 | 85.71 | 85.65 | |
5 | 83.67 | 84.02 | 83.67 | 83.58 | |
Mean | 84.62 | 84.86 | 84.42 | 84.42 | |
Std | \(\pm 0.88\) | \(\pm 0.83\) | \(\pm 0.89\) | \(\pm 0.86\) |
