Generate diverse human hand grasps for objects from single RGB-D images using models trained on real-world data.
