Update src/score_calculation/score.py

src/score_calculation/score.py

@@ -117,6 +117,8 @@ def resample_to_match_length(
     trace_1: np.ndarray, trace_2: np.ndarray
 ) -> Tuple[np.ndarray, np.ndarray]:
 
+    if len(trace_1) == 0 or len(trace_2) == 0:
+        raise ValueError("One of the trace is empty")
     if len(trace_1) == len(trace_2):
         return trace_1, trace_2
     elif len(trace_1) > len(trace_2):
@@ -308,11 +310,6 @@ def score_predictions(results_df, dataset):
     scores = []
     for _, row in tqdm(results_df.iterrows(), total=len(results_df), desc="Scoring predictions"):
 
-        # Skip invalid predictions
-        if len(row["prediction"]) == 0:
-            scores.append(np.nan)
-            continue
-            
         # Get the corresponding ground truth sample using the lookup
         sample_id = row["sample_id"]
         sample = dataset[id_to_index[sample_id]]
@@ -326,6 +323,11 @@ def score_predictions(results_df, dataset):
         if ground_truths is None:
             raise ValueError(f"The sample {sample} has hidden ground-truths")
 
+        # Skip invalid predictions
+        if len(prediction) == 0:
+            scores.append(np.nan)
+            continue
+        
         # Calculate the score and append it to the list
         s = score(prediction, ground_truths, segmentation_mask, embodiment)
         scores.append(s)