Table 2.
U-Net architecture. ‘Conv2D’ denotes a 2D convolutional layer; ‘ConvTransp2d’ denotes 2D transposed convolution; ‘BN’ denotes batch normalization; ‘f” is the number of features in the higher dimensional space, f = 64 for the saturation network, and f = 96 for the pressure buildup network.
| Stage | Layer type | Input channels | Output channels | Kernel size | Stride | Concatenation |
|---|---|---|---|---|---|---|
| Contracting path | ||||||
| 1 | Conv2d + BN + LeakyReLU | f | f | 3 | 2 | – |
| 2 | Conv2d + BN + LeakyReLU | f | f | 3 | 2 | – |
| 3 | Conv2d + BN + LeakyReLU | f | f | 3 | 1 | – |
| 4 | Conv2d + BN + LeakyReLU | f | f | 3 | 2 | – |
| 5 | Conv2d + BN + LeakyReLU | f | f | 3 | 1 | – |
| Expanding path | ||||||
| 6 | ConvTransp2d + LeakyReLU | f | f | 4 | 2 | With stage 3 |
| 7 | ConvTransp2d + LeakyReLU | 2f | f | 4 | 2 | With stage 2 |
| 8 | ConvTransp2d + LeakyReLU | 2f | f | 4 | 2 | With stage 1 |
| Output layer | ||||||
| 9 | Conv2d | 2f | f | 3 | 1 | With input |