symlink -> link, move dataset_dict_loader

core/preprocess.py

@@ -66,7 +66,7 @@ def preprocess_datasets(
     train_output_dir = None
     if train is not None:
         train = clean_input_paths(train)
-        if not all(missing := path.exists() for path in train):
+        if not all((missing := path).exists() for path in train):
             raise FileNotFoundError(f"Train File/Folder not found: {missing} does not exist")
 
         train_output_dir = output_dir.joinpath("train")

datasets/dataset.py

@@ -221,7 +221,9 @@ def register_dataset(path: str | Path, name: str, ignore_label: int = 255) -> Me
     data = convert_to_paths(path, info["data"])
     classes = info["classes"]
 
-    DatasetCatalog.register(name=name, func=lambda data=data: dataset_dict_loader(data))
+    list_of_dicts = dataset_dict_loader(data)
+
+    DatasetCatalog.register(name=name, func=lambda: list_of_dicts)
 
     MetadataCatalog[name] = metadata_from_classes(classes, ignore_label)
     return MetadataCatalog.get(name)

datasets/preprocess.py

@@ -314,7 +314,7 @@ def save_image(
         image_dir.mkdir(parents=True, exist_ok=True)
 
         if self.resize_mode == "none":
-            copy_mode(image_path, out_image_path, mode="symlink")
+            copy_mode(image_path, out_image_path, mode="link")
         else:
             image = load_image_array_from_path(image_path)
 

page_xml/output_pageXML.py

@@ -116,7 +116,7 @@ def link_image(self, image_path: Path):
             raise TypeError("Output dir is None")
         image_output_path = self.output_dir.joinpath(image_path.name)
 
-        copy_mode(image_path, image_output_path, mode="symlink")
+        copy_mode(image_path, image_output_path, mode="link")
 
     def generate_single_page(
         self,