Table 4 Hyperparameters for pretrained models.
Hyperparameter | Value |
|---|---|
Include top | False |
Input shape | (389, 389, 3) |
Weights | imagenet |
Pooling | Max |
Optimizer | Adam |
Learning rate | 0.001 |
Loss function | sparse_categorical_crossentropy |
Epochs | 25 |
Batch size | 32 |
Class mode | Binary |
Monitor (Early stopping) | Accuracy |
Min LR (Learning rate decay) | 1e-6 |
Patience (Early stopping) | 5 |
Patience (Reduce LR) | 3 |
Additional dense layer units | 1024 (ResNet50, EfficientNetB0, Xception, InceptionV3) |
Classifier activation | Softmax |
