KnowIT VQA: Answering Knowledge-Based Questions about Videos

TL;DR

This paper introduces KnowIT VQA, a new dataset and model for answering knowledge-based questions about videos, combining visual, textual, and knowledge reasoning to advance video understanding.

Contribution

It presents a novel dataset with knowledge-based questions about videos and a model that fuses visual, textual, and knowledge information for improved video question answering.

Findings

Knowledge incorporation significantly improves VQA performance.

Current models still lag behind human accuracy.

The dataset enables studying limitations of current video models.

Abstract

We propose a novel video understanding task by fusing knowledge-based and video question answering. First, we introduce KnowIT VQA, a video dataset with 24,282 human-generated question-answer pairs about a popular sitcom. The dataset combines visual, textual and temporal coherence reasoning together with knowledge-based questions, which need of the experience obtained from the viewing of the series to be answered. Second, we propose a video understanding model by combining the visual and textual video content with specific knowledge about the show. Our main findings are: (i) the incorporation of knowledge produces outstanding improvements for VQA in video, and (ii) the performance on KnowIT VQA still lags well behind human accuracy, indicating its usefulness for studying current video modelling limitations.