Modality-Specific Multimodal Global Enhanced Network for Text-Based Visual Question Answering | IEEE Conference Publication | IEEE Xplore