Generating Natural Questions from Images for Multimodal Assistants | IEEE Conference Publication | IEEE Xplore