Table 2.
Details of the implemented context aggregation network.
| Layer | Activations | Learnable Variable | Descriptions |
|---|---|---|---|
| Image input | 4 × 250 × 1 | – | 4 × 500 × 1 images |
| Convolutional | 4 × 250 × 32 | Weights 2 × 2 × 1 × 32, Bias 1 × 1 × 32 | 1 padding |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 2 padding, 1 Stride, 2 dilation | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 4 padding, 1 Stride, 4 dilation | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 8 padding, 1 Stride, 8 dilation | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 16 padding, 1 Stride, 16 dilation | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | 4 × 250 × 32 | – | – |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 32 padding, 1 Stride, 32 dilation | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 64 padding, 1 Stride, 64 dilation | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 128 padding, 1 Stride, 128 dilation | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | – | Scale 0.2 | |
| Convolutional | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 1 padding, 1 Stride | |
| Batch normalization | Offset 1 × 1 × 32, Scale 1 × 1 × 32 | – | |
| Adaptive normalization | – | – | |
| Leaky ReLU | 4 × 250 × 32 | – | Scale 0.2 |
| Convolutional | 4 × 250 × 3 | Weights 3 × 3 × 32 × 32, Bias 1 × 1 × 32 | 0 padding, 1 Stride |
| Regression output | 1 × 1 × 250 | – | Mean square error |