Learning Spatial and Temporal Cues for Multi-Label Facial Action Unit Detection | IEEE Conference Publication | IEEE Xplore