RaP: Redundancy-aware Video-language Pre-training for Text-Video Retrieval

Xing Wu, Chaochen Gao, Zijia Lin, Zhongyuan Wang, Jizhong Han, Songlin Hu


Abstract
Video language pre-training methods have mainly adopted sparse sampling techniques to alleviate the temporal redundancy of videos. Though effective, sparse sampling still suffers inter-modal redundancy: visual redundancy and textual redundancy. Compared with highly generalized text, sparsely sampled frames usually contain text-independent portions, called visual redundancy. Sparse sampling is also likely to miss important frames corresponding to some text portions, resulting in textual redundancy. Inter-modal redundancy leads to a mismatch of video and text information, hindering the model from better learning the shared semantics across modalities. To alleviate it, we propose Redundancy-aware Video-language Pre-training. We design a redundancy measurement of video patches and text tokens by calculating the cross-modal minimum dis-similarity. Then, we penalize the high-redundant video patches and text tokens through a proposed redundancy-aware contrastive learning. We evaluate our method on four benchmark datasets, MSRVTT, MSVD, DiDeMo, and LSMDC, achieving a significant improvement over the previous state-of-the-art results.
Anthology ID:
2022.findings-emnlp.221
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2022
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates
Editors:
Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
3036–3047
Language:
URL:
https://aclanthology.org/2022.findings-emnlp.221
DOI:
10.18653/v1/2022.findings-emnlp.221
Bibkey:
Cite (ACL):
Xing Wu, Chaochen Gao, Zijia Lin, Zhongyuan Wang, Jizhong Han, and Songlin Hu. 2022. RaP: Redundancy-aware Video-language Pre-training for Text-Video Retrieval. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 3036–3047, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
RaP: Redundancy-aware Video-language Pre-training for Text-Video Retrieval (Wu et al., Findings 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/naacl24-info/2022.findings-emnlp.221.pdf