Temporally Refined Graph U-Nets for Human Shape and Pose Estimation From Monocular Videos | IEEE Journals & Magazine | IEEE Xplore