Deep Modular Co-Attention Networks for Visual Question Answering | IEEE Conference Publication | IEEE Xplore