Table 4 Parameters of fully convolutional network
Layer | Type | Maps | Size | Kernel | Stride | Padding | Activation |
|---|---|---|---|---|---|---|---|
In | Input | 3 (RGB) | 576 × 150 | – | – | – | – |
C1 | Conv | 128 | 576 × 150 | 3 × 3 | 1 | 1 | BatchNorm, ReLU |
C2 | Conv | 64 | 576 × 150 | 1 × 1 | 1 | 0 | BatchNorm |
C3 | Conv | 32 | 576 × 150 | 3 × 3 | 1 | 1 | BatchNorm, ReLU |
C4 | Conv | 16 | 576 × 150 | 1 × 1 | 1 | 0 | BatchNorm |
C5 | Conv | 8 | 576 × 150 | 3 × 3 | 1 | 1 | BatchNorm, ReLU |
C6 | Conv | 4 | 576 × 150 | 1 × 1 | 1 | 0 | BatchNorm |
Out | – | – | 576 × 150 | – | – | – | Softmax |