LAVT: Language-Aware Vision Transformer for Referring Image Segmentation | IEEE Conference Publication | IEEE Xplore