Fix no dataset split error handling

tracklab/datastruct/tracking_dataset.py

@@ -5,6 +5,15 @@
 import pandas as pd
 
 
+class SetsDict(dict):
+    def __getitem__(self, key):
+        if key not in self:
+            raise KeyError(f"Trying to access a '{key}' split of the dataset that is not available. "
+                           f"Available splits are {list(self.keys())}. "
+                           f"Make sur this split name is correct or is available in the dataset folder.")
+        return super().__getitem__(key)
+
+
 @dataclass
 class TrackingSet:
     video_metadatas: pd.DataFrame
@@ -25,12 +34,8 @@ def __init__(
         **kwargs
     ):
         self.dataset_path = Path(dataset_path)
-        self.sets = sets
-        self.train_set = None
-        self.val_set = None
-        self.test_set = None
-
-        sub_sampled_sets = {}
+        self.sets = SetsDict(sets)
+        sub_sampled_sets = SetsDict()
         for set_name, split in self.sets.items():
             vid_list = vids_dict[set_name] if vids_dict is not None and set_name in vids_dict else None
             sub_sampled_sets[set_name] = self._subsample(split, nvid, nframes, vid_list)

tracklab/wrappers/datasets/soccernet/soccernet_game_state.py

@@ -28,17 +28,12 @@ def __init__(self,
             download_dataset(self.dataset_path)
         assert self.dataset_path.exists(), f"'{self.dataset_path}' directory does not exist. Please check the path or download the dataset following the instructions here: https://github.com/SoccerNet/sn-gamestate"
 
-        train_set = load_set(self.dataset_path / "train", nvid, vids_dict.get("train", [])) if os.path.exists(self.dataset_path / "train") else None
-        val_set = load_set(self.dataset_path / "valid", nvid, vids_dict.get("valid", [])) if os.path.exists(self.dataset_path / "valid") else None
-        test_set = load_set(self.dataset_path / "test", nvid, vids_dict.get("test", [])) if (self.dataset_path / "test").exists() else None
-        challenge = load_set(self.dataset_path / "challenge", nvid, vids_dict.get("challenge", [])) if os.path.exists(self.dataset_path / "challenge") else None
-
-        sets = {
-            "train": train_set,
-            "valid": val_set,
-            "test": test_set,
-            "challenge": challenge
-        }
+        sets = {}
+        for split in ["train", "valid", "test", "challenge"]:
+            if os.path.exists(self.dataset_path / split):
+                sets[split] = load_set(self.dataset_path / split, nvid, vids_dict.get(split, []))
+            else:
+                log.warning(f"Warning: The '{split}' set does not exist in the SoccerNetGS dataset at '{self.dataset_path}'")
 
         # We pass 'nvid=-1', 'vids_dict=None' because video subsampling is already done in the load_set function
         super().__init__(dataset_path, sets, nvid=-1, vids_dict=None, *args, **kwargs)
@@ -202,7 +197,6 @@ def load_set(dataset_path, nvid=-1, vids_filter_set=None):
     annotations_pitch_camera_list = []
     detections_list = []
     categories_list = []
-    image_gt_challenge = []
     split = os.path.basename(dataset_path)  # Get the split name from the dataset path
     video_list = os.listdir(dataset_path)
     video_list.sort()