Update README.md

Author: mathpluscode

README.md

@@ -192,6 +192,15 @@ make pip
 
 ## Build Data Sets
 
+:warning: For using your own data set, the following steps are not needed. Please check the
+[existing datasets](https://github.com/mathpluscode/ImgX-DiffSeg/blob/main/imgx_datasets/) for
+examples of using TFDS. Particularly,
+[BraTS 2021](https://github.com/mathpluscode/ImgX-DiffSeg/blob/main/imgx_datasets/brats2021_mr/brats2021_mr_dataset_builder.py)
+does not require downloading.
+
+:muscle: We are working on a toy example for using custom data sets without TFDS, thanks for your
+understanding.
+
 Use the following commands to (re)build all data sets. Check the [README](imgx_datasets/README.md)
 of imgx_datasets for details. Especially, manual downloading is required for the BraTS 2021 dataset.
 
@@ -200,6 +209,15 @@ make build_dataset
 make rebuild_dataset
 ```
 
+Or build the selected data set by running one of the following commands.
+
+```bash
+tfds build imgx_datasets/male_pelvic_mr
+tfds build imgx_datasets/amos_ct
+tfds build imgx_datasets/muscle_us
+tfds build imgx_datasets/brats2021_mr  # requires downloading data manually
+```
+
 ## Experiment
 
 ### Training and Testing

imgx_datasets/brats2021_mr/brats2021_mr_dataset_builder.py

@@ -302,7 +302,7 @@ def _split_generators(
         Returns:
             dict mapping split to generators.
         """
-        # Download data from zenodo
+        # Manually downloaded
         data_dir = dl_manager.manual_dir / BRATS2021_MR_TFDS_FOLD / "BraTS2021_Training_Data"
         preprocessed_dir = dl_manager.manual_dir / BRATS2021_MR_TFDS_FOLD / "preprocessed"
         preprocessed_dir.mkdir(parents=True, exist_ok=True)