Spatial-Semantic Attention for Grounded Image Captioning | IEEE Conference Publication | IEEE Xplore