Depth and Video Segmentation Based Visual Attention for Embodied Question Answering | IEEE Journals & Magazine | IEEE Xplore