🤗 HuggingFace for pretrained model weights hosting (#945)

This PR migrates model weight hosting from TIA servers to HuggingFace for improved distribution and caching. The changes introduce HuggingFace Hub integration while modifying default behavior to avoid automatic ImageNet weight downloads.

Key changes:
- Replace custom download logic with [HuggingFace](https://huggingface.co/TIACentre/TIAToolbox_pretrained_weights) for pretrained model weights
- Update model architecture handling for Inception and GoogleNet models
- Use `huggingface_hub` for downloading/caching pretrained weights
- Use `None` model `pretrained` argument during model initialization, to avoid redundant downloading of ImageNet weights

---------

Co-authored-by: Jiaqi Lv <[email protected]>
Co-authored-by: Shan E Ahmed Raza <[email protected]>

Author: 3 people

requirements/requirements.txt

@@ -9,6 +9,7 @@ filelock>=3.9.0
 flask>=2.2.2
 flask-cors>=4.0.0
 glymur>=0.12.7
+huggingface_hub>=0.33.3
 imagecodecs>=2022.9.26
 joblib>=1.1.1
 jupyterlab>=3.5.2

tests/test_utils.py

@@ -1618,21 +1618,24 @@ def test_from_multi_head_dat_type_dict(tmp_path: Path) -> None:
 
 def test_fetch_pretrained_weights(tmp_path: Path) -> None:
     """Test fetching pretrained weights for a model."""
-    file_path = tmp_path / "test_fetch_pretrained_weights.pth"
+    model_name = "mobilenet_v3_small-pcam"
+    file_path = tmp_path / f"{model_name}.pth"
     if file_path.exists():
         file_path.unlink()
 
-    fetch_pretrained_weights(model_name="mobilenet_v3_small-pcam", save_path=file_path)
+    _ = fetch_pretrained_weights(
+        model_name="mobilenet_v3_small-pcam", save_path=tmp_path
+    )
+
     assert file_path.exists()
     assert file_path.stat().st_size > 0
     file_path.unlink()
 
     with pytest.raises(ValueError, match="does not exist"):
-        fetch_pretrained_weights("abc", file_path)
+        fetch_pretrained_weights("abc", tmp_path)
 
     # Test save_path is str
-    file_path_str = str(file_path)
-    file_path = fetch_pretrained_weights("mobilenet_v3_small-pcam", file_path_str)
+    file_path = fetch_pretrained_weights(model_name, str(tmp_path))
     assert Path(file_path).exists()
     assert Path(file_path).stat().st_size > 0