Code for the Visual Question Answering (VQA) part of CVPR 2019 oral paper: "Learning to Compose Dynamic Tree Structures for Visual Contexts" - View it on GitHub