Regular bottlenecks are the main building block of ENet. Main branch: 1. Shortcut connection. Extension branch: 1. 1x1 convolution whi

Iterate two sets of indices An 'epoch' is one iteration through the primary indices. During the epoch, the secondary indices are iterated thr

Upssampling followed by ConvBlock

BraTS2019 Dataset

Crop randomly the image in a sample Args: output_size (int): Desired output size

Crop randomly flip the dataset in a sample Args: output_size (int): Desired output size

Convert ndarrays in sample to Tensors.

Iterate two sets of indices An 'epoch' is one iteration through the primary indices. During the epoch, the secondary indices are iterated thr

Downsampling followed by ConvBlock

two convolution layers with batch norm and leaky relu

Downsampling bottlenecks further downsample the feature map size. Main branch: 1. max pooling with stride 2; indices are saved to be used for

r""" Swin Transformer Block. Args: dim (int): Number of input channels. input_resolution (tuple[int]): Input resulotion.

Upssampling followed by ConvBlock

The upsampling bottlenecks upsample the feature map resolution using max pooling indices stored from the corresponding downsampling bottleneck.

A basic Swin Transformer layer for one stage. Args: dim (int): Number of input channels. input_resolution (tuple[int]): Input re

A basic Swin Transformer layer for one stage. Args: dim (int): Number of input channels. input_resolution (tuple[int]): Input re

two convolution layers with batch norm and leaky relu

Generate the ENet model. Keyword arguments: - num_classes (int): the number of classes to segment. - encoder_relu (bool, optional): When `

Unet_ is a fully convolution neural network for image semantic segmentation Args: encoder_name: name of classification model (without las

The initial block is composed of two branches: 1. a main branch which performs a regular convolution with stride 2; 2. an extension branch whi

r""" Image to Patch Embedding Args: img_size (int): Image size. Default: 224. patch_size (int): Patch token size. Default: 4.

r""" Swin Transformer A PyTorch impl of : `Swin Transformer: Hierarchical Vision Transformer using Shifted Windows` - https://arxiv

returns weakly and strongly augmented images Args: object (tuple): output size of network

r""" Window based multi-head self attention (W-MSA) module with relative position bias. It supports both of shifted and non-shifted window. A

Computes and stores the average and current value

fixes a bug in ConvDropoutNormNonlin where lrelu was used regardless of nonlin. Bad.

Add encoder functionality such as: - output channels specification of feature tensors (produced by encoder) - patching first convoluti

Add encoder functionality such as: - output channels specification of feature tensors (produced by encoder) - patching first convoluti