Encoding and Controlling Global Semantics for Long-form Video Question Answering

Thong Nguyen, Zhiyuan Hu, Xiaobao Wu, Cong-Duy Nguyen, See-Kiong Ng, Anh Tuan Luu


Abstract
Seeking answers effectively for long videos is essential to build video question answering (videoQA) systems. Previous methods adaptively select frames and regions from long videos to save computations. However, this fails to reason over the whole sequence of video, leading to sub-optimal performance. To address this problem, we introduce a state space layer (SSL) into multi-modal Transformer to efficiently integrate global semantics of the video, which mitigates the video information loss caused by frame and region selection modules. Our SSL includes a gating unit to enable controllability over the flow of global semantics into visual representations. To further enhance the controllability, we introduce a cross-modal compositional congruence objective to encourage global semantics aligned with the question. To rigorously evaluate long-form videoQA capacity, we construct two new benchmarks Ego-QA and MAD-QA featuring videos of considerably long length, i.e. 17.5 minutes and 1.9 hours, respectively. Extensive experiments demonstrate the superiority of our framework on these new as well as existing datasets.
Anthology ID:
2024.emnlp-main.400
Volume:
Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing
Month:
November
Year:
2024
Address:
Miami, Florida, USA
Editors:
Yaser Al-Onaizan, Mohit Bansal, Yun-Nung Chen
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
7049–7066
Language:
URL:
https://aclanthology.org/2024.emnlp-main.400
DOI:
Bibkey:
Cite (ACL):
Thong Nguyen, Zhiyuan Hu, Xiaobao Wu, Cong-Duy Nguyen, See-Kiong Ng, and Anh Tuan Luu. 2024. Encoding and Controlling Global Semantics for Long-form Video Question Answering. In Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 7049–7066, Miami, Florida, USA. Association for Computational Linguistics.
Cite (Informal):
Encoding and Controlling Global Semantics for Long-form Video Question Answering (Nguyen et al., EMNLP 2024)
Copy Citation:
PDF:
https://aclanthology.org/2024.emnlp-main.400.pdf
Software:
 2024.emnlp-main.400.software.zip
Data:
 2024.emnlp-main.400.data.zip