360∘ videos convey holistic views for the surroundings of a scene. It provides audio-visual cues beyond pre-determined normal field of views and displays distinctive spatial relations on a sphere. However, previous benchmark tasks for panoramic videos are still limited to evaluate the semantic understanding of audio-visual relationships or spherical spatial property in surroundings. We propose a novel benchmark named Pano-AVQA as a large-scale grounded audio-visual question answering dataset on panoramic videos. Using 5.4K 360∘ video clips harvested online, we collect two types of novel question-answer pairs with bounding-box grounding: spherical spatial relation QAs and audio-visual relation QAs. We train several transformer-based models from Pano-AVQA, where the results suggest that our proposed spherical spatial embeddings and multimodal training objectives fairly contribute to a better semantic understanding of the panoramic surroundings on the dataset.
@article{arxiv.2110.05122,
title = {Pano-AVQA: Grounded Audio-Visual Question Answering on 360$^\circ$ Videos},
author = {Heeseung Yun and Youngjae Yu and Wonsuk Yang and Kangil Lee and Gunhee Kim},
journal= {arXiv preprint arXiv:2110.05122},
year = {2021}
}