From 0cbd7fcc4646de95e207f153d75edc92cdb8395b Mon Sep 17 00:00:00 2001
From: Mayeul Aubin <mayeul.aubin@iap.fr>
Date: Mon, 30 Jun 2025 14:30:10 +0200
Subject: [PATCH] better UNets with different batch norms and custom depth

---
 sCOCA_ML/models/UNet_models.py | 41 ++++++++++++++++++++++++++++------
 1 file changed, 34 insertions(+), 7 deletions(-)

diff --git a/sCOCA_ML/models/UNet_models.py b/sCOCA_ML/models/UNet_models.py
index 8c99007..9138189 100644
--- a/sCOCA_ML/models/UNet_models.py
+++ b/sCOCA_ML/models/UNet_models.py
@@ -9,17 +9,24 @@ from .base_class_models import BaseModel
 from .FiLM import FiLM
 
 class UNetBlock(nn.Module):
-    def __init__(self, in_channels, out_channels, style_dim=None):
+    def __init__(self, in_channels, out_channels, style_dim=None, batch_norm=True):
         super(UNetBlock, self).__init__()
         self.conv1 = nn.Conv3d(in_channels, out_channels, kernel_size=3, padding=1)
         self.conv2 = nn.Conv3d(out_channels, out_channels, kernel_size=3, padding=1)
-        self.norm = nn.BatchNorm3d(out_channels)
+        
+        if batch_norm:
+            self.norm1 = nn.BatchNorm3d(out_channels)
+            self.norm2 = nn.BatchNorm3d(out_channels)
+        else:
+            self.norm1 = nn.Identity()
+            self.norm2 = nn.Identity()
+            
         self.relu = nn.ReLU(inplace=True)
         self.film = FiLM(out_channels, style_dim) if style_dim else None
 
     def forward(self, x, style=None):
-        x = self.relu(self.norm(self.conv1(x)))
-        x = self.relu(self.norm(self.conv2(x)))
+        x = self.relu(self.norm1(self.conv1(x)))
+        x = self.relu(self.norm2(self.conv2(x)))
         if self.film:
             x = self.film(x, style)
         return x
@@ -51,6 +58,7 @@ class UNet3D(BaseModel):
                  in_channels: int = 2,
                  out_channels: int = 1,
                  style_dim: int = 2,
+                 depth: int = None,
                  device: torch.device = torch.device('cpu'),
                  first_layer_channel_exponent: int = 3,
                  ):
@@ -61,7 +69,9 @@ class UNet3D(BaseModel):
         - in_channels: Number of input channels (default is 2).
         - out_channels: Number of output channels (default is 1).
         - style_dim: Dimension of the style vector (default is 2).
+        - depth: Depth of the U-Net (default is None, which will be computed based on N).
         - device: Device to load the model onto (default is CPU).
+        - first_layer_channel_exponent: Exponent for the number of channels in the first layer (default is 3).
         This model implements a 3D U-Net architecture with downsampling and upsampling blocks.
         The model uses convolutional layers with ReLU activations and batch normalization.
         The FiLM layers are used to condition the feature maps on style parameters.
@@ -74,7 +84,12 @@ class UNet3D(BaseModel):
                      device=device)
         import numpy as np
         
-        self.depth = np.floor(np.log2(N)).astype(int) - 1 # Depth of the U-Net based on input size N
+        if depth is not None:
+            self.depth = depth
+            if np.floor(np.log2(N)).astype(int) - 1 < depth:
+                raise ValueError(f"Depth {depth} is too large for input size {N}. Maximum depth is {np.floor(np.log2(N)).astype(int) - 1}.")
+        else:
+            self.depth = np.floor(np.log2(N)).astype(int) - 1 # Depth of the U-Net based on input size N
         self.first_layer_channel_exponent = first_layer_channel_exponent
 
         self.enc=[]
@@ -100,7 +115,12 @@ class UNet3D(BaseModel):
         self.dec = nn.ModuleList(self.dec)
     
 
-        self.final = nn.Conv3d(2**(self.first_layer_channel_exponent), out_channels, kernel_size=1)
+        # self.final = nn.Conv3d(2**(self.first_layer_channel_exponent), out_channels, kernel_size=3, padding=1)
+        self.final = nn.Sequential(
+            nn.Conv3d(2**(self.first_layer_channel_exponent), 2**(self.first_layer_channel_exponent), kernel_size=3, padding=1),
+            nn.ReLU(inplace=True),
+            nn.Conv3d(2**(self.first_layer_channel_exponent), out_channels, kernel_size=1)
+        )
 
         
 
@@ -127,6 +147,7 @@ class UNet3D_Shrink(BaseModel):
                  in_channels: int = 2,
                  out_channels: int = 1,
                  style_dim: int = 2,
+                 depth: int = None,
                  device: torch.device = torch.device('cpu'),
                  first_layer_channel_exponent: int = 3,
                  shrink_factor_exponent: int = 1,
@@ -142,7 +163,13 @@ class UNet3D_Shrink(BaseModel):
                          device=device)
         import numpy as np
         
-        self.depth_enc = np.floor(np.log2(N)).astype(int) - 1 # Depth of the U-Net based on input size N
+        if depth is not None:
+            self.depth_enc = depth
+            if np.floor(np.log2(N)).astype(int) - 1 < depth:
+                raise ValueError(f"Depth {depth} is too large for input size {N}. Maximum depth is {np.floor(np.log2(N)).astype(int) - 1}.")
+        else:
+            self.depth_enc = np.floor(np.log2(N)).astype(int) - 1 # Depth of the U-Net based on input size N
+            
         self.depth_dec = self.depth_enc - shrink_factor_exponent # Depth of the U-Net based on input size N and shrink factor
         self.first_layer_channel_exponent = first_layer_channel_exponent
         self.shrink_factor_exponent = shrink_factor_exponent