Knowledge-Based Visual Question Answering in Videos
This addresses the challenge of knowledge-based visual question answering in videos for AI researchers, but it is incremental as it builds on existing VQA tasks by adding a knowledge component.
The authors tackled the problem of video understanding by introducing a new dataset, KnowIT VQA, with 24,282 question-answer pairs from a sitcom, and a model that fuses visual, textual, and knowledge-based elements, showing that knowledge incorporation significantly improves video question answering performance, though it still lags behind human accuracy.
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.