LAVA: Language Audio Vision Alignment for Contrastive Video Pre-Training

TL;DR

LAVA is a self-supervised contrastive learning method that jointly learns language, audio, and video representations, enabling effective video understanding without extensive annotations.

Contribution

It introduces a novel self-supervised contrastive approach for joint language, audio, and video representation learning using transformer encoders.

Findings

Competitive performance on UCF-101 and HMDB-51 datasets.

Uses less unlabeled data than existing methods.

Effective for video action recognition.

Abstract

Generating representations of video data is of key importance in advancing the field of machine perception. Most current techniques rely on hand-annotated data, which can be difficult to work with, expensive to generate, and hard to scale. In this work, we propose a novel learning approach based on contrastive learning, LAVA, which is capable of learning joint language, audio, and video representations in a self-supervised manner. We pre-train LAVA on the Kinetics 700 dataset using transformer encoders to learn representations for each modality. We then demonstrate that LAVA performs competitively with the current state-of-the-art self-supervised and weakly-supervised pretraining techniques on UCF-101 and HMDB-51 video action recognition while using a fraction of the unlabeled data.