Adding files and setting up Git LFS for images

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+*.png filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,201 @@
+                                 Apache License
+                           Version 2.0, January 2004
+                        http://www.apache.org/licenses/
+
+   TERMS AND CONDITIONS FOR USE, REPRODUCTION, AND DISTRIBUTION
+
+   1. Definitions.
+
+      "License" shall mean the terms and conditions for use, reproduction,
+      and distribution as defined by Sections 1 through 9 of this document.
+
+      "Licensor" shall mean the copyright owner or entity authorized by
+      the copyright owner that is granting the License.
+
+      "Legal Entity" shall mean the union of the acting entity and all
+      other entities that control, are controlled by, or are under common
+      control with that entity. For the purposes of this definition,
+      "control" means (i) the power, direct or indirect, to cause the
+      direction or management of such entity, whether by contract or
+      otherwise, or (ii) ownership of fifty percent (50%) or more of the
+      outstanding shares, or (iii) beneficial ownership of such entity.
+
+      "You" (or "Your") shall mean an individual or Legal Entity
+      exercising permissions granted by this License.
+
+      "Source" form shall mean the preferred form for making modifications,
+      including but not limited to software source code, documentation
+      source, and configuration files.
+
+      "Object" form shall mean any form resulting from mechanical
+      transformation or translation of a Source form, including but
+      not limited to compiled object code, generated documentation,
+      and conversions to other media types.
+
+      "Work" shall mean the work of authorship, whether in Source or
+      Object form, made available under the License, as indicated by a
+      copyright notice that is included in or attached to the work
+      (an example is provided in the Appendix below).
+
+      "Derivative Works" shall mean any work, whether in Source or Object
+      form, that is based on (or derived from) the Work and for which the
+      editorial revisions, annotations, elaborations, or other modifications
+      represent, as a whole, an original work of authorship. For the purposes
+      of this License, Derivative Works shall not include works that remain
+      separable from, or merely link (or bind by name) to the interfaces of,
+      the Work and Derivative Works thereof.
+
+      "Contribution" shall mean any work of authorship, including
+      the original version of the Work and any modifications or additions
+      to that Work or Derivative Works thereof, that is intentionally
+      submitted to Licensor for inclusion in the Work by the copyright owner
+      or by an individual or Legal Entity authorized to submit on behalf of
+      the copyright owner. For the purposes of this definition, "submitted"
+      means any form of electronic, verbal, or written communication sent
+      to the Licensor or its representatives, including but not limited to
+      communication on electronic mailing lists, source code control systems,
+      and issue tracking systems that are managed by, or on behalf of, the
+      Licensor for the purpose of discussing and improving the Work, but
+      excluding communication that is conspicuously marked or otherwise
+      designated in writing by the copyright owner as "Not a Contribution."
+
+      "Contributor" shall mean Licensor and any individual or Legal Entity
+      on behalf of whom a Contribution has been received by Licensor and
+      subsequently incorporated within the Work.
+
+   2. Grant of Copyright License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      copyright license to reproduce, prepare Derivative Works of,
+      publicly display, publicly perform, sublicense, and distribute the
+      Work and such Derivative Works in Source or Object form.
+
+   3. Grant of Patent License. Subject to the terms and conditions of
+      this License, each Contributor hereby grants to You a perpetual,
+      worldwide, non-exclusive, no-charge, royalty-free, irrevocable
+      (except as stated in this section) patent license to make, have made,
+      use, offer to sell, sell, import, and otherwise transfer the Work,
+      where such license applies only to those patent claims licensable
+      by such Contributor that are necessarily infringed by their
+      Contribution(s) alone or by combination of their Contribution(s)
+      with the Work to which such Contribution(s) was submitted. If You
+      institute patent litigation against any entity (including a
+      cross-claim or counterclaim in a lawsuit) alleging that the Work
+      or a Contribution incorporated within the Work constitutes direct
+      or contributory patent infringement, then any patent licenses
+      granted to You under this License for that Work shall terminate
+      as of the date such litigation is filed.
+
+   4. Redistribution. You may reproduce and distribute copies of the
+      Work or Derivative Works thereof in any medium, with or without
+      modifications, and in Source or Object form, provided that You
+      meet the following conditions:
+
+      (a) You must give any other recipients of the Work or
+          Derivative Works a copy of this License; and
+
+      (b) You must cause any modified files to carry prominent notices
+          stating that You changed the files; and
+
+      (c) You must retain, in the Source form of any Derivative Works
+          that You distribute, all copyright, patent, trademark, and
+          attribution notices from the Source form of the Work,
+          excluding those notices that do not pertain to any part of
+          the Derivative Works; and
+
+      (d) If the Work includes a "NOTICE" text file as part of its
+          distribution, then any Derivative Works that You distribute must
+          include a readable copy of the attribution notices contained
+          within such NOTICE file, excluding those notices that do not
+          pertain to any part of the Derivative Works, in at least one
+          of the following places: within a NOTICE text file distributed
+          as part of the Derivative Works; within the Source form or
+          documentation, if provided along with the Derivative Works; or,
+          within a display generated by the Derivative Works, if and
+          wherever such third-party notices normally appear. The contents
+          of the NOTICE file are for informational purposes only and
+          do not modify the License. You may add Your own attribution
+          notices within Derivative Works that You distribute, alongside
+          or as an addendum to the NOTICE text from the Work, provided
+          that such additional attribution notices cannot be construed
+          as modifying the License.
+
+      You may add Your own copyright statement to Your modifications and
+      may provide additional or different license terms and conditions
+      for use, reproduction, or distribution of Your modifications, or
+      for any such Derivative Works as a whole, provided Your use,
+      reproduction, and distribution of the Work otherwise complies with
+      the conditions stated in this License.
+
+   5. Submission of Contributions. Unless You explicitly state otherwise,
+      any Contribution intentionally submitted for inclusion in the Work
+      by You to the Licensor shall be under the terms and conditions of
+      this License, without any additional terms or conditions.
+      Notwithstanding the above, nothing herein shall supersede or modify
+      the terms of any separate license agreement you may have executed
+      with Licensor regarding such Contributions.
+
+   6. Trademarks. This License does not grant permission to use the trade
+      names, trademarks, service marks, or product names of the Licensor,
+      except as required for reasonable and customary use in describing the
+      origin of the Work and reproducing the content of the NOTICE file.
+
+   7. Disclaimer of Warranty. Unless required by applicable law or
+      agreed to in writing, Licensor provides the Work (and each
+      Contributor provides its Contributions) on an "AS IS" BASIS,
+      WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or
+      implied, including, without limitation, any warranties or conditions
+      of TITLE, NON-INFRINGEMENT, MERCHANTABILITY, or FITNESS FOR A
+      PARTICULAR PURPOSE. You are solely responsible for determining the
+      appropriateness of using or redistributing the Work and assume any
+      risks associated with Your exercise of permissions under this License.
+
+   8. Limitation of Liability. In no event and under no legal theory,
+      whether in tort (including negligence), contract, or otherwise,
+      unless required by applicable law (such as deliberate and grossly
+      negligent acts) or agreed to in writing, shall any Contributor be
+      liable to You for damages, including any direct, indirect, special,
+      incidental, or consequential damages of any character arising as a
+      result of this License or out of the use or inability to use the
+      Work (including but not limited to damages for loss of goodwill,
+      work stoppage, computer failure or malfunction, or any and all
+      other commercial damages or losses), even if such Contributor
+      has been advised of the possibility of such damages.
+
+   9. Accepting Warranty or Additional Liability. While redistributing
+      the Work or Derivative Works thereof, You may choose to offer,
+      and charge a fee for, acceptance of support, warranty, indemnity,
+      or other liability obligations and/or rights consistent with this
+      License. However, in accepting such obligations, You may act only
+      on Your own behalf and on Your sole responsibility, not on behalf
+      of any other Contributor, and only if You agree to indemnify,
+      defend, and hold each Contributor harmless for any liability
+      incurred by, or claims asserted against, such Contributor by reason
+      of your accepting any such warranty or additional liability.
+
+   END OF TERMS AND CONDITIONS
+
+   APPENDIX: How to apply the Apache License to your work.
+
+      To apply the Apache License to your work, attach the following
+      boilerplate notice, with the fields enclosed by brackets "[]"
+      replaced with your own identifying information. (Don't include
+      the brackets!)  The text should be enclosed in the appropriate
+      comment syntax for the file format. We also recommend that a
+      file or class name and description of purpose be included on the
+      same "printed page" as the copyright notice for easier
+      identification within third-party archives.
+
+   Copyright 2025 Ole-Christian Galbo Engstrøm
+
+   Licensed under the Apache License, Version 2.0 (the "License");
+   you may not use this file except in compliance with the License.
+   You may obtain a copy of the License at
+
+       http://www.apache.org/licenses/LICENSE-2.0
+
+   Unless required by applicable law or agreed to in writing, software
+   distributed under the License is distributed on an "AS IS" BASIS,
+   WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+   See the License for the specific language governing permissions and
+   limitations under the License.

README.md

@@ -1,3 +1,70 @@
 ---
 license: apache-2.0
+tags:
+- computer-vision
+- image-segmentation
+- pytorch
+- unet
+- medical-imaging
+- semantic-segmentation
+library_name: pytorch
+pipeline_tag: image-segmentation
 ---
+
+# U-Net
+This repository contains an implementation of U-Net [[1]](#references). [unet.py](./unet.py) implements the class UNet. The implementation has been tested with PyTorch 2.7.1 and CUDA 12.6.
+![](./assets/unet_diagram.png)
+
+You can also load the U-Net from PyTorch Hub.
+```python
+import torch
+
+# These are the default parameters. They are written out for clarity. Currently no pretrained weights are available.
+model = torch.hub.load('sm00thix/unet', 'unet', pretrained=False, in_channels=3, out_channels=1, pad=True, bilinear=True, normalization=None)
+# or
+# model = torch.hub.load('sm00thix/unet', 'unet_bn', **kwargs) # Convenience function equivalent to torch.hub.load('sm00thix/unet', 'unet', normalization='bn', **kwargs)
+# or
+# model = torch.hub.load('sm00thix/unet', 'unet_ln', **kwargs) # Convenience function equivalent to torch.hub.load('sm00thix/unet', 'unet', normalization='ln', **kwargs)
+# or
+# model = torch.hub.load('sm00thix/unet', 'unet_medical', **kwargs) # Convenience function equivalent to torch.hub.load('sm00thix/unet', 'unet', in_channels=1, out_channels=1, **kwargs)
+# or
+# model = torch.hub.load('sm00thix/unet', 'unet_transconv', **kwargs) # Convenience function equivalent to torch.hub.load('sm00thix/unet', 'unet', bilinear=False, **kwargs)
+```
+
+## Options
+The UNet class provides the following options for customization.
+
+1. Number of input and output channels
+    `in_channels` is the number of channels in the input image.
+    `out_channels` is the number of channels in the output image.
+2. Upsampling
+    1. `bilinear = False`: Transposed convolution with a 2x2 kernel applied with stride 2. This is followed by a ReLU.
+    2. `bilinear = True`: Factor 2 bilinear upsampling followed by convolution with a 1x1 kernel applied with stride 1.
+3. Padding
+    1. `pad = True`: The input size is retained in the output by zero-padding convolutions and, if necessary, the results of the upsampling operations.
+    2. `pad = False`: The output is smaller than the input as in the original implementation. In this case, every 3x3 convolution layer reduces the height and width by 2 pixels each. Consequently, the right side of the U-Net has a smaller spatial size than the left size. Therefore, before concatenating, the central slice of the left tensor is cropped along the spatial dimensions to match those of the right tensor.
+4. Normalization following the ReLU which follows each convolution and transposed convolution.
+    1. `normalization = None`: Applies no normalization.
+    2. `normalization = "bn"`: Applies batch normalization [[2]](#references).
+    3. `normalization = "ln"`: Applies layer normalization [[3]](#references). A permutation of dimensions is performed before the layer to ensure normalization is applied over the channel dimension. Afterward, the dimensions are permuted back to their original order.
+
+In particular, setting bilinear = False, pad = False, and normalization = None will yield the U-Net as originally designed. Generally, however, bilinear = True is recommended to avoid checkerboard artifacts.
+
+As in the original implementation, all weights are initialized by sampling from a Kaiming He Normal Distribution [[4]](#references), and all biases are initialized to zero. If Batch Normalization or Layer Normalization is used, the weights of those layers are initialized to one and their biases to zero.
+
+If you use this U-Net implementation, please cite Engstrøm et al. [[5]](#references) who developed this implementation as part of their work on chemical map geenration of fat content in images of pork bellies.
+
+## Citation
+If you use the code shared in this repository, please cite this work: https://arxiv.org/abs/2504.14131. The U-Net implementation in this repository was used to generate pixel-wise fat predictions in an image of a pork belly.
+![](./assets/unet_flow.png)
+
+## References
+
+1. [O. Ronneberger, P. Fischer, and Thomas Brox (2015). U-Net: Convolutional Networks for Biomedical Image Segmentation. *MICCAI 2015*.](https://arxiv.org/abs/1505.04597)
+2. [S. Ioffe and C. Szegedy (2015). Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift. *ICML 2015*.](https://arxiv.org/abs/1502.03167)
+3. [J. L. Ba and J. R. Kiros and G. E. Hinton (2016). Layer Normalization.](https://arxiv.org/abs/1607.06450)
+4. [K. He and X. Zhang and S. Ren and J. Sun (2015). Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification.](https://openaccess.thecvf.com/content_iccv_2015/html/He_Delving_Deep_into_ICCV_2015_paper.html)
+5. [O.-C. G. Engstrøm and M. Albano-Gaglio and E. S. Dreier and Y. Bouzembrak and M. Font-i-Furnols and P. Mishra and K. S. Pedersen (2025). Transforming Hyperspectral Images Into Chemical Maps: A Novel End-to-End Deep Learning Approach.](https://arxiv.org/abs/2504.14131)
+
+## Funding
+This work has been carried out as part of an industrial Ph. D. project receiving funding from [FOSS Analytical A/S](https://www.fossanalytics.com/) and [The Innovation Fund Denmark](https://innovationsfonden.dk/en). Grant number 1044-00108B.

unet.py

@@ -0,0 +1,437 @@
+"""
+Contains an implementation of the U-Net architecture.
+U-Net paper by Ronneberger et al. (2015): https://arxiv.org/abs/1505.04597
+
+This implementation is based on the original U-Net architecture, with options for
+normalization (batch normalization or layer normalization), bilinear upsampling,
+and padding in the convolution layers.
+
+Author: Ole-Christian Galbo Engstrøm
+E-mail: [email protected]
+"""
+
+from typing import Iterable
+
+import torch
+from torch import nn
+from torch.nn import functional as F
+
+
+def conv3x3(in_channels: int, out_channels: int, bias: bool, pad: bool) -> nn.Conv2d:
+    """
+    Applies a convolution with a 3x3 kernel.
+    """
+    if pad:
+        padding = 1
+    else:
+        padding = "valid"
+    layer = nn.Conv2d(
+        in_channels,
+        out_channels,
+        kernel_size=3,
+        padding=padding,
+        bias=bias,
+    )
+    return layer
+
+
+def conv_block(
+    in_channels: int,
+    out_channels: int,
+    non_linearity: nn.Module,
+    normalization: None | str,
+    bias: bool,
+    pad: bool,
+) -> nn.Sequential:
+    """
+    A block of two convolutional layers, each followed by a non-linearity
+    and optionally a normalization layer.
+
+    In the U-Net architecture illustration in the U-Net paper,
+    this corresponds to two blue arrows.
+    """
+    layers = []
+    for _ in range(2):
+        layers.append(
+            conv3x3(
+                in_channels=in_channels, out_channels=out_channels, bias=bias, pad=pad
+            )
+        )
+        layers.append(non_linearity)
+        layers.append(
+            get_norm_layer(normalization=normalization, in_channels=out_channels)
+        )
+        in_channels = out_channels
+    return nn.Sequential(*layers)
+
+
+def batch_norm(in_channels: int) -> nn.Sequential:
+    """
+    Apply Batch Normalization over the channel dimension.
+    Batch Normalization paper by Ioffe and Szegedy (2015): https://arxiv.org/abs/1502.03167
+    """
+    return nn.BatchNorm2d(in_channels, momentum=0.01)
+
+
+class Permute(nn.Module):
+    """
+    Permute the dimensions of a tensor.
+    """
+
+    def __init__(self, dims: Iterable[int]):
+        super().__init__()
+        self.dims = dims
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return x.permute(self.dims)
+
+    def __repr__(self):
+        return f'{self.__class__.__name__}({", ".join(map(str, self.dims))})'
+
+
+def layer_norm(in_channels: int) -> nn.Sequential:
+    """
+    Apply Layer Normalization over the channel dimension.
+    Layer Normalization paper by Ba et al. (2016): https://arxiv.org/abs/1607.06450
+    """
+    layers = [
+        # (B, C, H, W) -> (B, H, W, C)
+        Permute((0, 2, 3, 1)),
+        # LayerNorm expects the last dimension to be the feature dimension
+        # (we want the normalized shape to be (C,))
+        nn.LayerNorm(in_channels),
+        # (B, H, W, C) -> (B, C, H, W)
+        Permute((0, 3, 1, 2)),
+    ]
+    return nn.Sequential(*layers)
+
+
+def get_norm_layer(normalization: None | str, in_channels: int) -> nn.Module:
+    """
+    Get the normalization layer based on the specified type.
+    Either 'bn' for batch normalization, 'ln' for layer normalization,
+    or None for no normalization layer.
+    """
+    if normalization == "bn":
+        return batch_norm(in_channels)
+    if normalization == "ln":
+        return layer_norm(in_channels)
+    return nn.Identity()
+
+
+def copy_and_crop(large: torch.Tensor, small: torch.Tensor) -> torch.Tensor:
+    """
+    Implementation of a copy-and-crop block in the U-Net architecture.
+    Copy the large image and crop it to the size of the small image.
+    The large image is cropped in the middle, and then the two images are
+    concatenated along the channel dimension.
+
+    In the U-Net architecture illustration in the U-Net paper,
+    this corresponds to a gray arrow.
+    """
+    large_height, large_width = large.shape[-2:]
+    small_height, small_width = small.shape[-2:]
+    start_x = (large_height - small_height) // 2
+    start_y = (large_width - small_width) // 2
+    cropped_large = large[
+        ..., start_x : start_x + small_height, start_y : start_y + small_width
+    ]
+    return torch.cat([cropped_large, small], dim=-3)
+
+
+class ContractionBlock(nn.Module):
+    """
+    Implementation of a contraction block in the U-Net architecture.
+    This block consists of a max pooling layer followed by a convolution block.
+
+    In the U-Net architecture illustration in the U-Net paper, this corresponds to
+    one red arrow followed by the subsequent two blue arrows.
+    """
+
+    def __init__(
+        self,
+        in_channels: int,
+        out_channels: int,
+        non_linearity: nn.Module,
+        nonormalization: None | str,
+        bias: bool,
+        pad: bool,
+    ):
+        super().__init__()
+        self.max_pool = self._max_pool()
+        self.conv_block = conv_block(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            non_linearity=non_linearity,
+            normalization=nonormalization,
+            bias=bias,
+            pad=pad,
+        )
+
+    def _max_pool(self) -> nn.MaxPool2d:
+        layer = nn.MaxPool2d(kernel_size=2, stride=2)
+        return layer
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x = self.max_pool(x)
+        x = self.conv_block(x)
+        return x
+
+
+class Upsample(nn.Module):
+    """
+    Implementation of an upsampling block in the U-Net architecture.
+    This block consists of either a transposed convolution or bilinear upsampling,
+    followed by a convolution block.
+
+    In the U-Net architecture illustration in the U-Net paper, this corresponds to
+    one green arrow.
+    """
+
+    def __init__(
+        self,
+        in_channels: int,
+        out_channels: int,
+        non_linearity,
+        normalization: None | str,
+        bias: bool,
+        bilinear: bool,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.non_linearity = non_linearity
+        self.normalization = normalization
+        self.bias = bias
+        self.bilinear = bilinear
+        self.up = self._upsample(in_channels, out_channels)
+
+    def _upsample(self, in_channels: int, out_channels: int) -> nn.Sequential:
+        if self.bilinear:
+            up = self._up_bilinear(in_channels, out_channels)
+        else:
+            up = self._up_trans_conv2x2(in_channels, out_channels)
+        return up
+
+    def _up_trans_conv2x2(self, in_channels: int, out_channels: int) -> nn.Sequential:
+        layers = [
+            nn.ConvTranspose2d(
+                in_channels, out_channels, kernel_size=2, stride=2, bias=self.bias
+            ),
+            self.non_linearity,
+        ]
+        layers.append(get_norm_layer(self.normalization, out_channels))
+        return nn.Sequential(*layers)
+
+    def _up_bilinear(self, in_channels: int, out_channels: int) -> nn.Sequential:
+        layers = [
+            nn.Upsample(mode="bilinear", scale_factor=2, align_corners=True),
+            nn.Conv2d(
+                in_channels=in_channels, out_channels=out_channels, kernel_size=1
+            ),
+            self.non_linearity,
+        ]
+        layers.append(get_norm_layer(self.normalization, out_channels))
+        return nn.Sequential(*layers)
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        return self.up(x)
+
+
+class ExpansionBlock(nn.Module):
+    """
+    Implementation of an expansion block in the U-Net architecture.
+    This block consists of an upsampling block followed by a copy-and-crop block and
+    a convolution block.
+
+    In the U-Net architecture illustration in the U-Net paper, this corresponds to
+    one green arrow followed by a gray arrow and then two blue arrows.
+    """
+
+    def __init__(
+        self,
+        in_channels: int,
+        out_channels: int,
+        non_linearity: nn.Module,
+        normalization: None | str,
+        bias: bool,
+        bilinear: bool,
+        pad: bool,
+    ):
+        super().__init__()
+        self.pad = pad
+        self.upsample = Upsample(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            non_linearity=non_linearity,
+            normalization=normalization,
+            bias=bias,
+            bilinear=bilinear,
+        )
+        self.conv_block = self.conv_block = conv_block(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            non_linearity=non_linearity,
+            normalization=normalization,
+            bias=bias,
+            pad=pad,
+        )
+
+    def forward(self, large: torch.Tensor, small: torch.Tensor) -> torch.Tensor:
+        x = self.upsample(small)
+        if self.pad:
+            diff_h = large.shape[-2] - x.shape[-2]
+            diff_w = large.shape[-1] - x.shape[-1]
+            pad_left = diff_w // 2
+            pad_right = diff_w - pad_left
+            pad_top = diff_h // 2
+            pad_bottom = diff_h - pad_top
+            x = F.pad(
+                x,
+                (pad_left, pad_right, pad_top, pad_bottom),
+                mode="constant",
+                value=0.0,
+            )
+        x = copy_and_crop(large, x)
+        x = self.conv_block(x)
+        return x
+
+
+class UNet(nn.Module):
+    """
+    in_channels : int\\
+        Number of input channels.
+
+    out_channels : int\\
+        Number of output channels
+
+    pad : bool, default=True\\
+        If True use padding in the convolution layers, preserving the input size.
+        If False, the output size will be reduced compared to the input size.
+
+    bilinear : bool, default=True\\
+        If True use bilinear upsampling.
+        If False use transposed convolution.
+
+    normalization: None | str, default=None\\
+        If None use no normalization.
+        If 'bn' use batch normalization.
+        If 'ln' use layer normalization.
+    """
+
+    def __init__(
+        self,
+        in_channels: int,
+        out_channels: int,
+        pad: bool = True,
+        bilinear: bool = True,
+        normalization: None | str = None,
+    ):
+        super().__init__()
+        self.in_channels = in_channels
+        self.out_channels = out_channels
+        self.pad = pad
+        self.bilinear = bilinear
+        self.normalization = normalization
+        if self.normalization not in [None, "bn", "ln"]:
+            raise ValueError(
+                "Normalization must be None, 'bn' for batch normalization,"
+                "or 'ln' for layer normalization"
+            )
+        # Whether to use bias in the convolution layers
+        # If normalization is used, bias is already included in the normalization layer
+        self.bias_conv = normalization is None
+        self.non_linearity = nn.ReLU(inplace=True)
+        self.intermediate_channels = [64 * 2**i for i in range(5)]
+        self.first_convs = conv_block(
+            in_channels=in_channels,
+            out_channels=self.intermediate_channels[0],
+            non_linearity=self.non_linearity,
+            normalization=self.normalization,
+            bias=self.bias_conv,
+            pad=self.pad,
+        )
+        self.last_conv = nn.Conv2d(
+            self.intermediate_channels[0], out_channels, kernel_size=1
+        )
+
+        self.contraction1 = self._get_contraction_block(
+            in_channels=self.intermediate_channels[0],
+            out_channels=self.intermediate_channels[1],
+        )
+        self.contraction2 = self._get_contraction_block(
+            in_channels=self.intermediate_channels[1],
+            out_channels=self.intermediate_channels[2],
+        )
+        self.contraction3 = self._get_contraction_block(
+            in_channels=self.intermediate_channels[2],
+            out_channels=self.intermediate_channels[3],
+        )
+        self.contraction4 = self._get_contraction_block(
+            in_channels=self.intermediate_channels[3],
+            out_channels=self.intermediate_channels[4],
+        )
+        self.expansion4 = self._get_expansion_block(
+            in_channels=self.intermediate_channels[4],
+            out_channels=self.intermediate_channels[3],
+        )
+        self.expansion3 = self._get_expansion_block(
+            in_channels=self.intermediate_channels[3],
+            out_channels=self.intermediate_channels[2],
+        )
+        self.expansion2 = self._get_expansion_block(
+            in_channels=self.intermediate_channels[2],
+            out_channels=self.intermediate_channels[1],
+        )
+        self.expansion1 = self._get_expansion_block(
+            in_channels=self.intermediate_channels[1],
+            out_channels=self.intermediate_channels[0],
+        )
+
+        # Init weights
+        for m in self.modules():
+            if isinstance(m, (nn.Conv2d, nn.ConvTranspose2d)):
+                nn.init.kaiming_normal_(m.weight)
+                if m.bias is not None:
+                    nn.init.constant_(m.bias, 0)
+            elif isinstance(m, (nn.BatchNorm2d, nn.LayerNorm)):
+                nn.init.constant_(m.weight, 1)
+                nn.init.constant_(m.bias, 0)
+
+    def _get_contraction_block(
+        self, in_channels: int, out_channels: int
+    ) -> ContractionBlock:
+        return ContractionBlock(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            non_linearity=self.non_linearity,
+            nonormalization=self.normalization,
+            bias=self.bias_conv,
+            pad=self.pad,
+        )
+
+    def _get_expansion_block(
+        self, in_channels: int, out_channels: int
+    ) -> ExpansionBlock:
+        return ExpansionBlock(
+            in_channels=in_channels,
+            out_channels=out_channels,
+            non_linearity=self.non_linearity,
+            normalization=self.normalization,
+            bias=self.bias_conv,
+            bilinear=self.bilinear,
+            pad=self.pad,
+        )
+
+    def forward(self, x: torch.Tensor) -> torch.Tensor:
+        x1 = self.first_convs(x)
+        x2 = self.contraction1(x1)
+        x3 = self.contraction2(x2)
+        x4 = self.contraction3(x3)
+        x5 = self.contraction4(x4)
+        x = self.expansion4(x4, x5)
+        x = self.expansion3(x3, x)
+        x = self.expansion2(x2, x)
+        x = self.expansion1(x1, x)
+        x = self.last_conv(x)
+        return x