Table 3.
Details of the proposed encoder-decoder convolutional network.
| Module | Block | Layer | Filter Size/Channels | Stride | Padding | |
|---|---|---|---|---|---|---|
| Encoder | GPM | GPMleft | Conv | 7 × 1/128 | 1 | Yes |
| Conv | 1 × 7/256 | 1 | Yes | |||
| GPMright | Conv | 1 × 7/128 | 1 | Yes | ||
| Conv | 7 × 1/256 | 1 | Yes | |||
| LPM | Conv | 3 × 3/256 | 1 | Yes | ||
| ReLU | ||||||
| Conv | 3 × 3/256 | 1 | Yes | |||
| Conv | Conv | 3 × 3/128 | 1 | Yes | ||
| ReLU | ReLU | |||||
| LPM | Conv | 3 × 3/128 | 1 | Yes | ||
| ReLU | ||||||
| Conv | 3 × 3/128 | 1 | Yes | |||
| Contrast Layer | Contrast Layer | avg_pool | 3 × 3/128 | 1 | No | |
| Decoder | DeConv | DeConv | 3 × 3/384 | 2 | Yes | |
| ReLU | ReLU | |||||
| LPM | Conv | 3 × 3/384 | 1 | Yes | ||
| ReLU | ||||||
| SConv | 3 × 3/384 | 1 | Yes | |||
| ReLU | ReLU | |||||