CenterBlock¶
tiatoolbox.models.architecture.kongnet.CenterBlock
- class CenterBlock(in_channels)[source]¶
Center block that applies attention mechanism at the bottleneck.
This block is placed at the center of the U-Net architecture (deepest level) to enhance feature representation using attention mechanisms.
- Parameters:
in_channels (int) – Number of input channels
Initialize CenterBlock with attention.
- Parameters:
in_channels (int) – Number of input channels.
Methods
Forward pass through center block.
Attributes
training- forward(x)[source]¶
Forward pass through center block.
- Parameters:
x (torch.Tensor) – Input tensor.
- Returns:
Output tensor with attention applied.
- Return type: