Integrating Graphics and Vision for Object Recognition