def _get_adn_layer(act: Optional[Union[Tuple, str]],
dropout: Optional[Union[Tuple, str, float]],
ordering: Optional[str]) -> ADN:
if ordering:
return ADN(act=act, dropout=dropout, dropout_dim=1, ordering=ordering)
else:
return ADN(act=act, dropout=dropout, dropout_dim=1)
def test_adn_3d(self, args):
adn = ADN(**args)
print(adn)
out = adn(self.imt)
expected_shape = (1, self.input_channels, self.im_shape[1],
self.im_shape[0], self.im_shape[2])
self.assertEqual(out.shape, expected_shape)
def __init__(
self,
spatial_dims: int,
in_channels: int,
out_channels: int,
kernels: Sequence[int] = (3, 3),
dilation: Union[Sequence[int], int] = 1,
norm_type: Union[Tuple, str] = ("batch", {
"affine": True
}),
acti_type: Union[Tuple, str] = ("relu", {
"inplace": True
}),
channel_matching: Union[ChannelMatching, str] = ChannelMatching.PAD,
) -> None:
"""
Args:
spatial_dims: number of spatial dimensions of the input image.
in_channels: number of input channels.
out_channels: number of output channels.
kernels: each integer k in `kernels` corresponds to a convolution layer with kernel size k.
dilation: spacing between kernel elements.
norm_type: feature normalization type and arguments.
Defaults to ``("batch", {"affine": True})``.
acti_type: {``"relu"``, ``"prelu"``, ``"relu6"``}
Non-linear activation using ReLU or PReLU. Defaults to ``"relu"``.
channel_matching: {``"pad"``, ``"project"``}
Specifies handling residual branch and conv branch channel mismatches. Defaults to ``"pad"``.
- ``"pad"``: with zero padding.
- ``"project"``: with a trainable conv with kernel size one.
Raises:
ValueError: When ``channel_matching=pad`` and ``in_channels > out_channels``. Incompatible values.
"""
super(HighResBlock, self).__init__()
self.chn_pad = ChannelPad(spatial_dims=spatial_dims,
in_channels=in_channels,
out_channels=out_channels,
mode=channel_matching)
layers = nn.ModuleList()
_in_chns, _out_chns = in_channels, out_channels
for kernel_size in kernels:
layers.append(
ADN(ordering="NA",
in_channels=_in_chns,
act=acti_type,
norm=norm_type,
norm_dim=spatial_dims))
layers.append(
Convolution(
dimensions=spatial_dims,
in_channels=_in_chns,
out_channels=_out_chns,
kernel_size=kernel_size,
dilation=dilation,
))
_in_chns = _out_chns
self.layers = nn.Sequential(*layers)
def __init__(
self,
spatial_dims: int,
in_channels: int,
out_channels: int,
strides: Union[Sequence[int], int] = 1,
kernel_size: Union[Sequence[int], int] = 3,
adn_ordering: str = "NDA",
act: Optional[Union[Tuple, str]] = "PRELU",
norm: Optional[Union[Tuple, str]] = "INSTANCE",
dropout: Optional[Union[Tuple, str, float]] = None,
dropout_dim: Optional[int] = 1,
dilation: Union[Sequence[int], int] = 1,
groups: int = 1,
bias: bool = True,
conv_only: bool = False,
is_transposed: bool = False,
padding: Optional[Union[Sequence[int], int]] = None,
output_padding: Optional[Union[Sequence[int], int]] = None,
dimensions: Optional[int] = None,
) -> None:
super().__init__()
self.dimensions = spatial_dims if dimensions is None else dimensions
self.in_channels = in_channels
self.out_channels = out_channels
self.is_transposed = is_transposed
if padding is None:
padding = same_padding(kernel_size, dilation)
conv_type = Conv[Conv.CONVTRANS if is_transposed else Conv.CONV,
self.dimensions]
conv: nn.Module
if is_transposed:
if output_padding is None:
output_padding = stride_minus_kernel_padding(1, strides)
conv = conv_type(
in_channels,
out_channels,
kernel_size=kernel_size,
stride=strides,
padding=padding,
output_padding=output_padding,
groups=groups,
bias=bias,
dilation=dilation,
)
else:
conv = conv_type(
in_channels,
out_channels,
kernel_size=kernel_size,
stride=strides,
padding=padding,
dilation=dilation,
groups=groups,
bias=bias,
)
self.add_module("conv", conv)
if not conv_only:
self.add_module(
"adn",
ADN(
ordering=adn_ordering,
in_channels=out_channels,
act=act,
norm=norm,
norm_dim=self.dimensions,
dropout=dropout,
dropout_dim=dropout_dim,
),
)
def __init__(
self,
spatial_dims: int,
in_channels: int,
out_channels: int,
kernels: Sequence[int] = (3, 3),
dilation: Union[Sequence[int], int] = 1,
norm_type: Union[Tuple, str] = ("batch", {"affine": True}),
acti_type: Union[Tuple, str] = ("relu", {"inplace": True}),
bias: bool = False,
channel_matching: Union[ChannelMatching, str] = ChannelMatching.PAD,
) -> None:
"""
Args:
spatial_dims: number of spatial dimensions of the input image.
in_channels: number of input channels.
out_channels: number of output channels.
kernels: each integer k in `kernels` corresponds to a convolution layer with kernel size k.
dilation: spacing between kernel elements.
norm_type: feature normalization type and arguments.
Defaults to ``("batch", {"affine": True})``.
acti_type: {``"relu"``, ``"prelu"``, ``"relu6"``}
Non-linear activation using ReLU or PReLU. Defaults to ``"relu"``.
bias: whether to have a bias term in convolution blocks. Defaults to False.
According to `Performance Tuning Guide <https://pytorch.org/tutorials/recipes/recipes/tuning_guide.html>`_,
if a conv layer is directly followed by a batch norm layer, bias should be False.
channel_matching: {``"pad"``, ``"project"``}
Specifies handling residual branch and conv branch channel mismatches. Defaults to ``"pad"``.
- ``"pad"``: with zero padding.
- ``"project"``: with a trainable conv with kernel size one.
Raises:
ValueError: When ``channel_matching=pad`` and ``in_channels > out_channels``. Incompatible values.
"""
super().__init__()
self.chn_pad = ChannelPad(
spatial_dims=spatial_dims, in_channels=in_channels, out_channels=out_channels, mode=channel_matching
)
layers = nn.ModuleList()
_in_chns, _out_chns = in_channels, out_channels
for kernel_size in kernels:
layers.append(
ADN(ordering="NA", in_channels=_in_chns, act=acti_type, norm=norm_type, norm_dim=spatial_dims)
)
layers.append(
Convolution(
spatial_dims=spatial_dims,
in_channels=_in_chns,
out_channels=_out_chns,
kernel_size=kernel_size,
dilation=dilation,
bias=bias,
conv_only=True,
)
)
_in_chns = _out_chns
self.layers = nn.Sequential(*layers)
def test_no_input(self):
with self.assertRaises(ValueError):
ADN(norm="instance")