Multimodal Clustering Networks for Self-supervised Learning from Unlabeled Videos
Abstract
Multimodal self-supervised learning is getting more and more attention as it allows not only to train large networks without human supervision but also to search and retrieve data across various modalities. In this context, this paper proposes a self-supervised training framework that learns a common multimodal embedding space that, in addition to sharing representations across different modalities, enforces a grouping of semantically similar instances. To this end, we extend the concept of instance-level contrastive learning with a multimodal clustering step in the training pipeline to capture semantic similarities across modalities. The resulting embedding space enables retrieval of samples across all modalities, even from unseen datasets and different domains. To evaluate our approach, we train our model on the HowTo100M dataset and evaluate its zero-shot retrieval capabilities in two challenging domains, namely text-to-video retrieval, and temporal action localization, showing state-of-the-art results on four different datasets.
Cite
@article{arxiv.2104.12671,
title = {Multimodal Clustering Networks for Self-supervised Learning from Unlabeled Videos},
author = {Brian Chen and Andrew Rouditchenko and Kevin Duarte and Hilde Kuehne and Samuel Thomas and Angie Boggust and Rameswar Panda and Brian Kingsbury and Rogerio Feris and David Harwath and James Glass and Michael Picheny and Shih-Fu Chang},
journal= {arXiv preprint arXiv:2104.12671},
year = {2021}
}
Comments
To be presented at ICCV 2021