Fetching the paper…

TAG: Boosting Text-VQA via Text-aware Visual Question-answer Generation · Around