fixing rle decode for segmentation

vision_agent/tools/tools.py

@@ -668,7 +668,7 @@ def generic_segmentation(image: np.ndarray) -> List[Dict[str, Any]]:
  "score": round(answer["scores"][i], 2),
  "label": answer["labels"][i],
  "mask": rle_decode(
- mask_rle=answer["masks"][i], shape=answer["mask_shape"]
+ mask_rle=answer["masks"][i], shape=answer["mask_shape"][0]
  ),
  }
  )