FPN

class mmit.decoders.FPN(input_channels, input_reductions, decoder_channel=256, upsample_layer=<class 'mmit.base.upsamplers.ConvTranspose2d'>, norm_layer=<class 'torch.nn.modules.batchnorm.BatchNorm2d'>, activation_layer=<class 'torch.nn.modules.activation.ReLU'>, extra_layer=<class 'torch.nn.modules.linear.Identity'>, mismatch_layer=<class 'mmit.base.mismatch.Pad'>, merge_mode='sum')

Implementation of the FPN decoder. Paper: https://arxiv.org/abs/1612.03144

Parameters:

• input_channels (List[int]) – The channels of the input features.

• input_reductions (List[int]) – The reduction factor of the input features.

• decoder_channel (int) – The channel to use on the decoder.

• upsample_layer (Type[Module]) – Layer to use for the upsampling.

• norm_layer (Type[Module]) – Normalization layer to use.

• activation_layer (Type[Module]) – Activation function to use.

• extra_layer (Type[Module]) – Addional layer to use.

• mismatch_layer (Type[Module]) – Strategy to deal with odd resolutions.

forward(*features)

Forward pass of the decoder.

Parameters:

*features (Tensor) – Features from the encoder, the first is the input image, last one the deepest.

Return type:

Tensor

property out_classes: int

Number of output classes.