def unet_v1(features, mode):
""" U-Net: Convolutional Networks for Biomedical Image Segmentation
Source:
https://arxiv.org/pdf/1505.04597
"""
skip_connections = []
out, skip = input_block(features, filters=64)
skip_connections.append(skip)
for idx, filters in enumerate([128, 256, 512]):
out, skip = downsample_block(out, filters=filters, idx=idx)
skip_connections.append(skip)
out = bottleneck(out, filters=1024, mode=mode)
for idx, filters in enumerate([512, 256, 128]):
out = upsample_block(out,
residual_input=skip_connections.pop(),
filters=filters,
idx=idx)
return output_block(out,
residual_input=skip_connections.pop(),
filters=64,
n_classes=2)
def __call__(self, features):
skip_128 = input_block(x=features,
out_channels=32,
normalization=self._normalization,
mode=self._mode)
skip_64 = downsample_block(x=skip_128,
out_channels=64,
normalization=self._normalization,
mode=self._mode)
skip_32 = downsample_block(x=skip_64,
out_channels=128,
normalization=self._normalization,
mode=self._mode)
skip_16 = downsample_block(x=skip_32,
out_channels=256,
normalization=self._normalization,
mode=self._mode)
skip_8 = downsample_block(x=skip_16,
out_channels=320,
normalization=self._normalization,
mode=self._mode)
x = downsample_block(x=skip_8,
out_channels=320,
normalization=self._normalization,
mode=self._mode)
x = upsample_block(x, skip_8,
out_channels=320,
normalization=self._normalization,
mode=self._mode)
x = upsample_block(x, skip_16,
out_channels=256,
normalization=self._normalization,
mode=self._mode)
x = upsample_block(x, skip_32,
out_channels=128,
normalization=self._normalization,
mode=self._mode)
x = upsample_block(x, skip_64,
out_channels=64,
normalization=self._normalization,
mode=self._mode)
x = upsample_block(x, skip_128,
out_channels=32,
normalization=self._normalization,
mode=self._mode)
return output_layer(x=x,
out_channels=self._n_classes,
activation='softmax')
def __call__(self, features):
x = input_block(inputs=features,
filters=16,
kernel_size=self._kernel_size,
normalization=self._normalization,
activation=self._activation,
mode=self._mode)
skip_connections = [x]
for depth in self._downscale_blocks:
x = downsample_block(inputs=x,
depth=depth,
kernel_size=self._kernel_size,
pooling=self._pooling,
normalization=self._normalization,
activation=self._activation,
mode=self._mode)
skip_connections.append(x)
del skip_connections[-1]
for depth in self._upscale_blocks:
x = upsample_block(inputs=x,
residual_inputs=skip_connections.pop(),
depth=depth,
upsampling=self._upsampling,
kernel_size=self._kernel_size,
normalization=self._normalization,
activation=self._activation,
mode=self._mode)
return output_block(inputs=x,
residual_inputs=skip_connections.pop(),
kernel_size=self._kernel_size,
n_classes=self._n_classes,
upsampling=self._upsampling,
normalization=self._normalization,
activation=self._activation,
mode=self._mode)
def __call__(self, features):
""" Build UNet3D
:param features: Input features
:return: Output of the graph
"""
skip_128 = input_block(inputs=features,
out_channels=32,
normalization=self._normalization,
mode=self._mode)
skip_64 = downsample_block(inputs=skip_128,
out_channels=64,
normalization=self._normalization,
mode=self._mode)
skip_32 = downsample_block(inputs=skip_64,
out_channels=128,
normalization=self._normalization,
mode=self._mode)
skip_16 = downsample_block(inputs=skip_32,
out_channels=256,
normalization=self._normalization,
mode=self._mode)
skip_8 = downsample_block(inputs=skip_16,
out_channels=320,
normalization=self._normalization,
mode=self._mode)
out = downsample_block(inputs=skip_8,
out_channels=320,
normalization=self._normalization,
mode=self._mode)
out = upsample_block(out,
skip_8,
out_channels=320,
normalization=self._normalization,
mode=self._mode)
out = upsample_block(out,
skip_16,
out_channels=256,
normalization=self._normalization,
mode=self._mode)
out = upsample_block(out,
skip_32,
out_channels=128,
normalization=self._normalization,
mode=self._mode)
out = upsample_block(out,
skip_64,
out_channels=64,
normalization=self._normalization,
mode=self._mode)
out = upsample_block(out,
skip_128,
out_channels=32,
normalization=self._normalization,
mode=self._mode)
return output_layer(out,
out_channels=self._n_classes,
activation='softmax')