Table 2 The configuration parameters of DRSN-CW model.
Number | Layer type | Convolution kernel | Number | Activation function | Dropout | Output size |
|---|---|---|---|---|---|---|
1 | Input | – | – | – | – | 180 × 180 × 1 |
2 | Convolution layer | 3 × 3 | 32 | Relu | – | 180 × 180 × 32 |
3 | Residual shrinkage layer | – | – | – | – | 90 × 90 × 32 |
4 | Maximum pooling layer | 2 × 2 | 32 | – | – | 45 × 45 × 32 |
5 | Convolution layer | 3 × 3 | 64 | Relu | – | 45 × 45 × 64 |
6 | Maximum pooling layer | 2 × 2 | 64 | – | – | 22 × 22 × 64 |
7 | Convolution layer | 3 × 3 | 64 | Relu | – | 22 × 22 × 64 |
8 | Maximum pooling layer | 2 × 2 | 64 | – | – | 11 × 11 × 64 |
9 | Convolution layer | 3 × 3 | 128 | Relu | – | 11 × 11 × 128 |
10 | Maximum pooling layer | 2 × 2 | 128 | – | – | 11 × 11 × 128 |
11 | Global average pooling | – | – | – | – | 128 |
12 | Fully connected layer | – | – | Relu | 0.5 | 2048 |
13 | Fully connected layer | – | – | Relu | 0.5 | 1024 |
14 | Output | – | – | Softmax | – | 8 |
