Fixed a bug where SVHN was downloaded twice

Author: Seilmast

CollaborativeCoding/dataloaders/download.py

@@ -98,11 +98,11 @@ def download_svhn(path, train: bool = True):
         train_data = parent_path / "train_32x32.mat"
         test_data = parent_path / "test_32x32.mat"
 
-        if not train_data.is_file():
+        if not train_data.exists():
             download_svhn(parent_path, train=True)
-        if not test_data.is_file():
+        if not test_data.exists():
             download_svhn(parent_path, train=False)
-        print(test_data)
+
         train_labels = loadmat(train_data)["y"]
         test_labels = loadmat(test_data)["y"]
 

CollaborativeCoding/dataloaders/svhn.py

@@ -31,7 +31,7 @@ def __init__(
         """
         super().__init__()
 
-        self.data_path = data_path
+        self.data_path = data_path / "SVHN"
         self.indexes = sample_ids
         self.split = "train" if train else "test"
 
@@ -41,7 +41,7 @@ def __init__(
         if not os.path.exists(
             os.path.join(self.data_path, f"svhn_{self.split}data.h5")
         ):
-            self._download_data(self.data_path)
+            self._create_h5py(self.data_path)
 
         assert os.path.exists(
             os.path.join(self.data_path, f"svhn_{self.split}data.h5")
@@ -53,15 +53,14 @@ def __init__(
 
         self.num_classes = len(np.unique(self.labels))
 
-    def _download_data(self, path: str):
+    def _create_h5py(self, path: str):
         """
         Downloads the SVHN dataset to the specified directory.
         Args:
             path (str): The directory where the dataset will be downloaded.
         """
         print(f"Downloading SVHN data into {path}")
 
-        SVHN(path, split=self.split, download=True)
         data = loadmat(os.path.join(path, f"{self.split}_32x32.mat"))
 
         images, labels = data["X"], data["y"]