Explicit Knowledge-based Reasoning for Visual Question Answering

TL;DR

This paper introduces a knowledge-based reasoning method for visual question answering that can handle complex questions, provide explanations, and outperforms LSTM-based approaches, supported by a new dataset and evaluation protocol.

Contribution

It presents a novel explicit knowledge reasoning approach for VQA, enabling complex question answering with explanations and establishing a new evaluation framework.

Findings

Outperforms LSTM-based methods significantly

Capable of answering more complex questions

Provides explanations for its answers

Abstract

We describe a method for visual question answering which is capable of reasoning about contents of an image on the basis of information extracted from a large-scale knowledge base. The method not only answers natural language questions using concepts not contained in the image, but can provide an explanation of the reasoning by which it developed its answer. The method is capable of answering far more complex questions than the predominant long short-term memory-based approach, and outperforms it significantly in the testing. We also provide a dataset and a protocol by which to evaluate such methods, thus addressing one of the key issues in general visual ques- tion answering.