Learning 3D object-centric representation through prediction

Open in new window