arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.16784cs.CV
收藏

基于可微弱时间对齐的视频-文本表示学习

Video-Text Representation Learning via Differentiable Weak Temporal Alignment

Korea University(高丽大学);Kakao Brain;Hanyang University(汉阳大学)

Dohwan Ko, Joonmyung Choi, Juyeon Ko, Shinyeong Noh, Kyoung-Woon On, Eun-Sol Kim, Hyunwoo J. Kim

摘要:

通过监督方法学习视频和文本的通用联合表示需要大量手动注释的视频数据集。作为一种实用的替代方案,最近推出了一个大规模但未经策划和叙述的视频数据集HowTo 100 M。但是,由于其模糊性和非顺序对齐性,以自监督的方式学习视频和文本的联合嵌入仍然具有挑战性。在本文中,我们提出了一种新的多模态自监督框架视频文本时间弱对齐为基础的对比学习(VT-THERM),以捕捉显着的信息,从噪声和弱相关的数据使用动态时间规整(DTW)的变体。我们观察到,标准DTW固有地不能处理弱相关的数据,只考虑全局最优的对齐路径。为了解决这些问题,我们开发了一个可微DTW,它也反映了弱时间对齐的局部信息。此外,我们提出的模型应用对比学习计划学习弱相关数据的特征表示。我们广泛的实验表明,VT-Tcloud在多模态表示学习方面取得了显着的改进,并优于各种具有挑战性的下游任务。代码可在https://github.com/mlvlab/VT-TWINS上获得。

英文摘要:

Learning generic joint representations for video and text by a supervised method requires a prohibitively substantial amount of manually annotated video datasets. As a practical alternative, a large-scale but uncurated and narrated video dataset, HowTo100M, has recently been introduced. But it is still challenging to learn joint embeddings of video and text in a self-supervised manner, due to its ambiguity and non-sequential alignment. In this paper, we propose a novel multi-modal self-supervised framework Video-Text Temporally Weak Alignment-based Contrastive Learning (VT-TWINS) to capture significant information from noisy and weakly correlated data using a variant of Dynamic Time Warping (DTW). We observe that the standard DTW inherently cannot handle weakly correlated data and only considers the globally optimal alignment path. To address these problems, we develop a differentiable DTW which also reflects local information with weak temporal alignment. Moreover, our proposed model applies a contrastive learning scheme to learn feature representations on weakly correlated data. Our extensive experiments demonstrate that VT-TWINS attains significant improvements in multi-modal representation learning and outperforms various challenging downstream tasks. Code is available at https://github.com/mlvlab/VT-TWINS.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑