Table 6 Model’s performance of binary classification on a primary dataset.
From: Knowledge distillation-based lightweight MobileNet model for diabetic retinopathy classification
Model | Averaging | Accuracy (%) | Precision (%) | Recall (%) | F1 Score (%) |
|---|---|---|---|---|---|
Teacher | macro avg | 93.88 | 92.65 | 93.88 | 93.21 |
weighted avg | 93.88 | 94.05 | 93.88 | 93.92 | |
Student without KD | macro avg | 85.71 | 83.79 | 86.73 | 84.69 |
weighted avg | 85.71 | 87.28 | 85.71 | 86.01 | |
Student with KD (Proposed Model) | macro avg | 93.20 | 93.24 | 91.33 | 92.18 |
weighted avg | 93.20 | 93.20 | 93.20 | 93.12 |
