models package

Subpackages

• models.backbones package
  • Submodules
  • models.backbones.VGG module
  • models.backbones.adaptive_unet module
  • models.backbones.baby_cnn module
  • models.backbones.backboned_unet module
  • models.backbones.deeplabv3 module
  • models.backbones.deeplabv3_aspp module
  • models.backbones.deeplabv3_resnet module
  • models.backbones.doc_ufcn module
  • models.backbones.resnet module
  • models.backbones.resnetdd module
  • models.backbones.segnet module
  • models.backbones.unet module
  • Module contents
• models.headers package
  • Submodules
  • models.headers.fully_connected module
  • models.headers.fully_convolution module
  • models.headers.unet module
  • Module contents

Submodules

models.backbone_header_model module

class BackboneHeaderModel(backbone: Union[LightningModule, Module], header: Union[LightningModule, Module], backbone_output_layer: Optional[str] = None)

Bases: LightningModule

A generic model class to provide the possibility to create different backbone/header combinations. The backbone and header compatibility can be tested with the callback CheckBackboneHeaderCompatibility during runtime. The loading of the different parts is done in execute.

Parameters:

• backbone (Union[pl.LightningModule, torch.nn.Module]) – The backbone model

• header (Union[pl.LightningModule, torch.nn.Module]) – The header model

• backbone_output_layer (Optional[str]) – The name of the output layer of the backbone. If None, the last layer of the backbone is used.

forward(x)

Same as torch.nn.Module.forward().

Parameters:

• *args – Whatever you decide to pass into the forward method.

• **kwargs – Keyword arguments are also possible.

Returns:

Your model’s output

training: bool

Module contents