Learning Deep Spatio-Temporal Dependence for Semantic Video Segmentation | IEEE Journals & Magazine | IEEE Xplore