Cross-Modal Self-Attention Network for Referring Image Segmentation | IEEE Conference Publication | IEEE Xplore