In this work, we train an adversarial multi-view images generator from original RGB image to provide semi supervision and estimate 3D human pose.
確定! 回上一頁