Merge branch 'dev' into perf/panoptic-pairwise-iou-bincount

Author: aymuos15

monai/apps/utils.py

@@ -240,7 +240,7 @@ def download_url(
             if urlparse(url).netloc == "drive.google.com":
                 if not has_gdown:
                     raise RuntimeError("To download files from Google Drive, please install the gdown dependency.")
-                if "fuzzy" not in gdown_kwargs:
+                if "fuzzy" not in gdown_kwargs and not min_version(gdown, "6.0.0"):  # "fuzzy" dropped in gdown 6.0.0
                     gdown_kwargs["fuzzy"] = True  # default to true for flexible url
                 gdown.download(url, f"{tmp_name}", quiet=not progress, **gdown_kwargs)
             elif urlparse(url).netloc == "cloud-api.yandex.net":

monai/bundle/scripts.py

@@ -2005,7 +2005,6 @@ def download_large_files(bundle_path: str | None = None, large_file_name: str |
     parser.read_config(large_file_path)
     large_files_list = parser.get()["large_files"]
     for lf_data in large_files_list:
-        lf_data["fuzzy"] = True
         if "hash_val" in lf_data and lf_data.get("hash_val", "") == "":
             lf_data.pop("hash_val")
         if "hash_type" in lf_data and lf_data.get("hash_type", "") == "":

monai/networks/nets/hovernet.py

@@ -632,7 +632,7 @@ def _remap_preact_resnet_model(model_url: str):
     pattern_bna = re.compile(r"^(.+\.d\d+)\.blk_bna\.(.+)")
     # download the pretrained weights into torch hub's default dir
     weights_dir = os.path.join(torch.hub.get_dir(), "preact-resnet50.pth")
-    download_url(model_url, fuzzy=True, filepath=weights_dir, progress=False)
+    download_url(model_url, filepath=weights_dir, progress=False)
     map_location = None if torch.cuda.is_available() else torch.device("cpu")
     state_dict = torch.load(weights_dir, map_location=map_location, weights_only=True)["desc"]
 
@@ -667,7 +667,7 @@ def _remap_standard_resnet_model(model_url: str, state_dict_key: str | None = No
     pattern_downsample1 = re.compile(r"^(res_blocks.d\d+).+\.downsample\.1\.(.+)")
     # download the pretrained weights into torch hub's default dir
     weights_dir = os.path.join(torch.hub.get_dir(), "resnet50.pth")
-    download_url(model_url, fuzzy=True, filepath=weights_dir, progress=False)
+    download_url(model_url, filepath=weights_dir, progress=False)
     map_location = None if torch.cuda.is_available() else torch.device("cpu")
     state_dict = torch.load(weights_dir, map_location=map_location, weights_only=True)
     if state_dict_key is not None:

monai/transforms/utils.py

@@ -1224,15 +1224,15 @@ def get_largest_connected_component_mask(
     if num_features <= num_components:
         out = img_.astype(bool)
     else:
-        # ignore background
-        nonzeros = features[lib.nonzero(features)]
-        # get number voxels per feature (bincount). argsort[::-1] to get indices
-        # of largest components.
-        features_to_keep = lib.argsort(lib.bincount(nonzeros))[::-1]
-        # only keep the first n non-background indices
-        features_to_keep = features_to_keep[:num_components]
-        # generate labelfield. True if in list of features to keep
-        out = lib.isin(features, features_to_keep)
+        # bincount counts every label; index 0 is background, so drop it before ranking
+        counts = lib.bincount(features.reshape(-1))
+        counts[0] = 0
+        # argsort[::-1] gives labels of the largest components; keep the first n
+        features_to_keep = lib.argsort(counts)[::-1][:num_components]
+        # boolean lookup-table gather over the label field, cheaper than isin
+        keep = lib.zeros(counts.shape[0], dtype=bool)
+        keep[features_to_keep] = True
+        out = keep[features]
 
     return convert_to_dst_type(out, dst=img, dtype=out.dtype)[0]