Can we apply both offset field and vector field in similar situations such as finding keypoints in objects shapes in images? If yes then how and also which one is better?