Update base.py

ultralytics/data/base.py

@@ -102,13 +102,13 @@ class BaseDataset(Dataset):
         counts = np.bincount(cls.astype(int), minlength=len(self.data["names"]))
         class_weights = counts.sum() / counts
         # weights = np.zeros(len(self.labels))
-        weights = np.ones(len(self.labels))
+        im_weights = np.ones(len(self.labels))
         for i, label in enumerate(self.labels):
             cls = label["cls"].reshape(-1).astype(np.int32)
             if len(cls) == 0:
                 continue
-            weights[i] = np.sum(class_weights[cls])
-        return weights / weights.sum()
+            im_weights[i] = np.sum(class_weights[cls])
+        return im_weights / im_weights.sum()
 
         # weights[i] = np.mean(counts[cls])
         # set mean value of weights for background images