A monocular depth estimation model by Tencent that generates temporally consistent depth maps for video sequences.