Table 9 Model’s performance of ternary classification on a primary dataset.
From: Knowledge distillation-based lightweight MobileNet model for diabetic retinopathy classification
Model | Averaging | Accuracy (%) | Precision (%) | Recall(%) | F1 Score(%) |
|---|---|---|---|---|---|
Teacher Model | macro avg | 86.39 | 85.70 | 84.49 | 85.04 |
weighted avg | 86.39 | 86.33 | 86.39 | 86.33 | |
Student without KD | macro avg | 76.19 | 73.78 | 78.07 | 75.04 |
weighted avg | 76.19 | 78.01 | 76.19 | 76.36 | |
Student with KD (Proposed Model) | macro avg | 85.71 | 83.58 | 83.38 | 83.46 |
weighted avg | 85.71 | 85.62 | 85.71 | 85.65 |
