This is a video question-answering dataset with subsets: cartoon, documentary, live, selfmedia, sport, and tv.
