Table 3.
The detailed information of the CNN model.
| # | Layer | Filter Size | Stride | Output Dimension | Activation Function |
|---|---|---|---|---|---|
| 1 | Input | — | — | (128, 128, 3) | — |
| 2 | Conv1 | 3 × 3 | 1 | (128, 128, 64) | ReLU |
| 3 | Maxpool1 | 2 × 2 | 2 | (64, 64, 64) | — |
| 4 | Conv2 | 3 × 3 | 1 | (64, 64, 32) | ReLU |
| 5 | Maxpool2 | 2 × 2 | 2 | (32, 32, 32) | — |
| 6 | Conv3 | 3 × 3 | 1 | (32, 32, 16) | ReLU |
| 7 | Maxpool3 | 2 × 2 | 2 | (16, 16, 16) | — |
| 8 | FC1 | — | — | (128, 1) | — |
| 9 | FC2 | — | — | (100, 1) | dropout = 0.5 |
| 10 | Output | — | — | (0, 1) | softmax |
Conv = convolutional layer; Maxpool = max-pooling layer; FC = fully connected layer.