OK-VQA: A Visual Question Answering Benchmark Requiring External Knowledge
This dataset enables research in knowledge-based VQA, addressing a gap in scene understanding for AI, though it is incremental as it builds on existing VQA tasks.
The authors introduced OK-VQA, a visual question answering benchmark requiring external knowledge, with over 14,000 questions where image content alone is insufficient, and showed that state-of-the-art VQA models perform poorly on it.
Visual Question Answering (VQA) in its ideal form lets us study reasoning in the joint space of vision and language and serves as a proxy for the AI task of scene understanding. However, most VQA benchmarks to date are focused on questions such as simple counting, visual attributes, and object detection that do not require reasoning or knowledge beyond what is in the image. In this paper, we address the task of knowledge-based visual question answering and provide a benchmark, called OK-VQA, where the image content is not sufficient to answer the questions, encouraging methods that rely on external knowledge resources. Our new dataset includes more than 14,000 questions that require external knowledge to answer. We show that the performance of the state-of-the-art VQA models degrades drastically in this new setting. Our analysis shows that our knowledge-based VQA task is diverse, difficult, and large compared to previous knowledge-based VQA datasets. We hope that this dataset enables researchers to open up new avenues for research in this domain. See http://okvqa.allenai.org to download and browse the dataset.