deepdraw.models.backbones.mobilenetv2

Functions

mobilenet_v2_for_segmentation([pretrained, ...])

Classes

MobileNetV24Segmentation(*args, **kwargs)

Adaptation of base MobileNetV2 functionality to U-Net style segmentation.

class deepdraw.models.backbones.mobilenetv2.MobileNetV24Segmentation(*args, **kwargs)

Bases: MobileNetV2

Adaptation of base MobileNetV2 functionality to U-Net style segmentation.

This version of MobileNetV2 is slightly modified so it can be used through torchvision’s API. It outputs intermediate features which are normally not output by the base MobileNetV2 implementation, but are required for segmentation operations.

Parameters:

return_features (list, Optional) – A list of integers indicating the feature layers to be returned from the original module.

forward(x)

Defines the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

deepdraw.models.backbones.mobilenetv2.mobilenet_v2_for_segmentation(pretrained=False, progress=True, **kwargs)