Update rsna-atd.py

rsna-atd.py

@@ -33,27 +33,27 @@ class RSNAATD(datasets.GeneratorBasedBuilder):
             description=_DESCRIPTION,
             features=datasets.Features(
                 {
-                    "test": datasets.Value("string"),
-                    # "patient_id": datasets.Value("int64"),
-                    # "series_id": datasets.Value("int64"),
-                    # "image": datasets.Array3D(shape=(None, 512, 512), dtype="uint8"),
-                    # "mask": datasets.Array3D(shape=(None, 512, 512), dtype="uint8"),
-                    # "aortic_hu": datasets.Value("float64"),
-                    # "incomplete_organ": datasets.Value("int64"),
-                    # "bowel_healthy": datasets.Value("int64"),
-                    # "bowel_injury": datasets.Value("int64"),
-                    # "extravasation_healthy": datasets.Value("int64"),
-                    # "extravasation_injury": datasets.Value("int64"),
-                    # "kidney_healthy": datasets.Value("int64"),
-                    # "kidney_low": datasets.Value("int64"),
-                    # "kidney_high": datasets.Value("int64"),
-                    # "liver_healthy": datasets.Value("int64"),
-                    # "liver_low": datasets.Value("int64"),
-                    # "liver_high": datasets.Value("int64"),
-                    # "spleen_healthy": datasets.Value("int64"),
-                    # "spleen_low": datasets.Value("int64"),
-                    # "spleen_high": datasets.Value("int64"),
-                    # "any_injury": datasets.Value("int64"),
+                    # "test": datasets.Value("string"),
+                    "patient_id": datasets.Value("int64"),
+                    "series_id": datasets.Value("int64"),
+                    "image": datasets.Array3D(shape=(None, 512, 512), dtype="uint8"),
+                    "mask": datasets.Array3D(shape=(None, 512, 512), dtype="uint8"),
+                    "aortic_hu": datasets.Value("float64"),
+                    "incomplete_organ": datasets.Value("int64"),
+                    "bowel_healthy": datasets.Value("int64"),
+                    "bowel_injury": datasets.Value("int64"),
+                    "extravasation_healthy": datasets.Value("int64"),
+                    "extravasation_injury": datasets.Value("int64"),
+                    "kidney_healthy": datasets.Value("int64"),
+                    "kidney_low": datasets.Value("int64"),
+                    "kidney_high": datasets.Value("int64"),
+                    "liver_healthy": datasets.Value("int64"),
+                    "liver_low": datasets.Value("int64"),
+                    "liver_high": datasets.Value("int64"),
+                    "spleen_healthy": datasets.Value("int64"),
+                    "spleen_low": datasets.Value("int64"),
+                    "spleen_high": datasets.Value("int64"),
+                    "any_injury": datasets.Value("int64"),
                 }
             ),
             supervised_keys=None,
@@ -62,8 +62,8 @@ class RSNAATD(datasets.GeneratorBasedBuilder):
         )
 
     def _split_generators(self, dl_manager):
-        train_images = dl_manager.download(f"{_DATA}images.zip")
-        train_masks = dl_manager.download(f"{_DATA}masks.zip")
+        train_images = dl_manager.download_and_extract(f"{_DATA}images.zip")
+        train_masks = dl_manager.download_and_extract(f"{_DATA}masks.zip")
 
         metadata = dl_manager.download(f"{_DATA}metadata.csv")
         train_images = dl_manager.iter_files(train_images)
@@ -86,13 +86,13 @@ class RSNAATD(datasets.GeneratorBasedBuilder):
         #         "test": images
         #     }
         # return
-        for i, image in enumerate(images):
-            # image_path = Path(image).name
-            # test = hkl.load(image_path)
-            yield i, {
-                "test": image
-            }
-        return
+        # for i, image in enumerate(images):
+        #     # image_path = Path(image).name
+        #     # test = hkl.load(image_path)
+        #     yield i, {
+        #         "test": image
+        #     }
+        # return
         for idx, (data, image_path, mask_path) in enumerate(zip(df.to_numpy(), images, masks)):
             image_path = Path(image_path) / data[0]
             mask_path = Path(mask_path) / data[0]