direct.nn.mwcnn package

Submodules

direct.nn.mwcnn.mwcnn module

class direct.nn.mwcnn.mwcnn.ConvBlock(in_channels, out_channels, kernel_size, bias=True, batchnorm=False, activation=ReLU(inplace=True), scale=1.0)

Bases: Module

Convolution Block for MWCNN as implemented in [1].

References

[1]

Liu, Pengju, et al. “Multi-Level Wavelet-CNN for Image Restoration.” ArXiv:1805.07071 [Cs], May 2018. arXiv.org, http://arxiv.org/abs/1805.07071.

forward(x)

Performs forward pass of ConvBlock.

Parameters:

x: torch.Tensor

Input with shape (N, C, H, W).

Returns:

output: torch.Tensor

Output with shape (N, C’, H’, W’).

Return type:

Tensor

class direct.nn.mwcnn.mwcnn.DWT

Bases: Module

2D Discrete Wavelet Transform as implemented in [1].

References

[1]

Liu, Pengju, et al. “Multi-Level Wavelet-CNN for Image Restoration.” ArXiv:1805.07071 [Cs], May 2018. arXiv.org, http://arxiv.org/abs/1805.07071.

forward(x)

Computes DWT(x) given tensor x.

Parameters:

x: torch.Tensor

Input tensor.

Returns:

out: torch.Tensor

DWT of x.

Return type:

Tensor

class direct.nn.mwcnn.mwcnn.DilatedConvBlock(in_channels, dilations, kernel_size, out_channels=None, bias=True, batchnorm=False, activation=ReLU(inplace=True), scale=1.0)

Bases: Module

Double dilated Convolution Block fpr MWCNN as implemented in [1].

References

[1]

Liu, Pengju, et al. “Multi-Level Wavelet-CNN for Image Restoration.” ArXiv:1805.07071 [Cs], May 2018. arXiv.org, http://arxiv.org/abs/1805.07071.

forward(x)

Performs forward pass of DilatedConvBlock.

Parameters:

x: torch.Tensor

Input with shape (N, C, H, W).

Returns:

output: torch.Tensor

Output with shape (N, C’, H’, W’).

Return type:

Tensor

class direct.nn.mwcnn.mwcnn.IWT

Bases: Module

2D Inverse Wavelet Transform as implemented in [1].

References

[1]

Liu, Pengju, et al. “Multi-Level Wavelet-CNN for Image Restoration.” ArXiv:1805.07071 [Cs], May 2018. arXiv.org, http://arxiv.org/abs/1805.07071.

forward(x)

Computes IWT(x) given tensor x.

Parameters:

x: torch.Tensor

Input tensor.

Returns:

h: torch.Tensor

IWT of x.

Return type:

Tensor

class direct.nn.mwcnn.mwcnn.MWCNN(input_channels, first_conv_hidden_channels, num_scales=4, bias=True, batchnorm=False, activation=ReLU(inplace=True))

Bases: Module

Multi-level Wavelet CNN (MWCNN) implementation as implemented in [1].

References

[1]

Liu, Pengju, et al. “Multi-Level Wavelet-CNN for Image Restoration.” ArXiv:1805.07071 [Cs], May 2018. arXiv.org, http://arxiv.org/abs/1805.07071.

static crop_to_shape(x, shape)

Return type:

Tensor

forward(input_tensor, res=False)

Computes forward pass of MWCNN.

Parameters:

input_tensor: torch.Tensor

Input tensor.

res: bool

If True, residual connection is applied to the output. Default: False.

Returns:

x: torch.Tensor

Output tensor.

Return type:

Tensor

static pad(x)

Return type:

Tensor

Module contents