Video-Text Retrieval by Supervised Sparse Multi-Grained Learning

Yimu Wang, Peng Shi


Abstract
While recent progress in video-text retrieval has been advanced by the exploration of better representation learning, in this paper, we present a novel multi-grained sparse learning framework, S3MA, to learn an aligned sparse space shared between the video and the text for video-text retrieval. The shared sparse space is initialized with a finite number of sparse concepts, each of which refers to a number of words. With the text data at hand, we learn and update the shared sparse space in a supervised manner using the proposed similarity and alignment losses. Moreover, to enable multi-grained alignment, we incorporate frame representations for better modeling the video modality and calculating fine-grained and coarse-grained similarities. Benefiting from the learned shared sparse space and multi-grained similarities, extensive experiments on several video-text retrieval benchmarks demonstrate the superiority of S3MA over existing methods.
Anthology ID:
2023.findings-emnlp.46
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2023
Month:
December
Year:
2023
Address:
Singapore
Editors:
Houda Bouamor, Juan Pino, Kalika Bali
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
633–649
Language:
URL:
https://aclanthology.org/2023.findings-emnlp.46
DOI:
10.18653/v1/2023.findings-emnlp.46
Bibkey:
Cite (ACL):
Yimu Wang and Peng Shi. 2023. Video-Text Retrieval by Supervised Sparse Multi-Grained Learning. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 633–649, Singapore. Association for Computational Linguistics.
Cite (Informal):
Video-Text Retrieval by Supervised Sparse Multi-Grained Learning (Wang & Shi, Findings 2023)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-5/2023.findings-emnlp.46.pdf