Refactor type aliases in types.py for improved clarity and consistency

ellipse_rcnn/utils/types.py

@@ -1,5 +1,4 @@
-from typing import TypedDict, NamedTuple
-
+from typing import TypedDict, NamedTuple, TypeAlias
 import torch
 
 
@@ -29,13 +28,13 @@ class PredictionDict(TypedDict):
     ellipse_matrices: torch.Tensor
 
 
-type ImageTargetTuple = tuple[torch.Tensor, TargetDict]  # Tensor shape: (C, H, W)
-type CollatedBatchType = tuple[
+ImageTargetTuple: TypeAlias = tuple[torch.Tensor, TargetDict]  # Tensor shape: (C, H, W)
+CollatedBatchType: TypeAlias = tuple[
     tuple[torch.Tensor, ...], tuple[TargetDict, ...]
 ]  # Tensor shape: (C, H, W)
-type UncollatedBatchType = list[ImageTargetTuple]
+UncollatedBatchType: TypeAlias = list[ImageTargetTuple]
 
-type EllipseType = torch.Tensor
+EllipseType: TypeAlias = torch.Tensor
 
 
 class EllipseTuple(NamedTuple):