Georgia Gkioxari proposes shifting computer vision research toward object affordances, physical-space reasoning, and efficient robotics models · Digg