feat: ⚡️ Add get_class_weights to dataset and subdataset.

pyproject.toml

@@ -30,8 +30,9 @@ dependencies = [
     "torchvision",
     "numpy",
     "matplotlib",
-    "fibsem-tools",
-    "cellmap_schemas",
+    "fibsem-tools==6.3.2",
+    # "fibsem-tools",
+    "cellmap-schemas",
     "tensorstore",
     "xarray-tensorstore",
     "cellpose",

src/cellmap_data/dataset.py

@@ -425,6 +425,24 @@ def set_target_value_transforms(self, transforms: Callable):
                 if isinstance(source, CellMapImage):
                     source.value_transform = transforms
 
+    def get_class_weights(self):
+        """
+        Returns the class weights for the multi-dataset based on the number of samples in each class.
+        """
+        if len(self.classes) > 1:
+            class_counts = {c: 0 for c in self.classes}
+            class_count_sum = 0
+            for c in self.classes:
+                class_counts[c] += self.class_counts["totals"][c]
+                class_count_sum += self.class_counts["totals"][c]
+
+            class_weights = {
+                c: 1 - (class_counts[c] / class_count_sum) for c in self.classes
+            }
+        else:
+            class_weights = {self.classes[0]: 0.1}  # less than 1 to avoid overflow
+        return class_weights
+
 
 # Example input arrays:
 # {'0_input': {'shape': (90, 90, 90), 'scale': (32, 32, 32)},

src/cellmap_data/subdataset.py

@@ -1,3 +1,4 @@
+from typing import Callable
 from torch.utils.data import Dataset
 
 
@@ -16,3 +17,31 @@ def __len__(self):
     def to(self, device):
         self.dataset.to(device)
         return self
+
+    def set_raw_value_transforms(self, transforms: Callable):
+        """Sets the raw value transforms for the subset dataset."""
+        self.dataset.set_raw_value_transforms(transforms)
+
+    def set_target_value_transforms(self, transforms: Callable):
+        """Sets the target value transforms for the subset dataset."""
+        self.dataset.set_target_value_transforms(transforms)
+
+    def get_class_weights(self):
+        """
+        Returns the class weights for the multi-dataset based on the number of samples in each class.
+        """
+        if len(self.dataset.classes) > 1:
+            class_counts = {c: 0 for c in self.dataset.classes}
+            class_count_sum = 0
+            for c in self.dataset.classes:
+                class_counts[c] += self.dataset.class_counts["totals"][c]
+                class_count_sum += self.dataset.class_counts["totals"][c]
+
+            class_weights = {
+                c: 1 - (class_counts[c] / class_count_sum) for c in self.dataset.classes
+            }
+        else:
+            class_weights = {
+                self.dataset.classes[0]: 0.1
+            }  # less than 1 to avoid overflow
+        return class_weights