Heads3D#

class vision_architectures.blocks.heads_3d.ClassificationHead3D(in_channels, classes, pooling='avg', dropout=0.2, activation=None)[source]#

Bases: Sequential

A general purpose classification head for 3D inputs.

# Inspiration: qubvel-org/segmentation_models.pytorch

__init__(in_channels, classes, pooling='avg', dropout=0.2, activation=None)[source]#

Initializes the head.

Parameters:

  • in_channels (int) – Number of input channels.

  • classes (int) – Number of output classes

  • pooling (str) – Should be one of “avg” or “max”. Defaults to “avg”.

  • dropout (float) – Amount of dropout to apply. Defaults to 0.2.

  • activation (Optional[str]) – Type of activation to perform. Defaults to None.

Raises:

ValueError – Incorrect pooling type.

class vision_architectures.blocks.heads_3d.SegmentationHead3D(in_channels, out_channels, kernel_size=3, activation=None, upsampling=1)[source]#

Bases: Sequential

A general purpose segmentation head for 3D inputs.”

Inspiration: qubvel-org/segmentation_models.pytorch

__init__(in_channels, out_channels, kernel_size=3, activation=None, upsampling=1)[source]#

Initializes the segmentation head

Parameters:

  • in_channels (int) – Number of input channels.

  • out_channels (int) – Number of output channels.

  • kernel_size (float) – Size of the kernel. Defaults to 3.

  • activation (Optional[str]) – Type of activation to perform. Defaults to None.

  • upsampling (float) – Scale factor. Defaults to 1.