Efficient Vision-Language Pre-training by Cluster Masking

TL;DR

This paper introduces a cluster masking strategy for vision-language pre-training that enhances representation quality and training efficiency by masking visually similar image patches based on pixel intensities, providing additional learning signals.

Contribution

The proposed cluster masking method is a simple yet effective approach that improves visual-language pre-training by masking clusters of similar patches, outperforming existing masking strategies.

Findings

Outperforms FLIP in representation quality

Speeds up training by reducing data per image

Provides extra learning signal through masked visual structures

Abstract

We propose a simple strategy for masking image patches during visual-language contrastive learning that improves the quality of the learned representations and the training speed. During each iteration of training, we randomly mask clusters of visually similar image patches, as measured by their raw pixel intensities. This provides an extra learning signal, beyond the contrastive training itself, since it forces a model to predict words for masked visual structures solely from context. It also speeds up training by reducing the amount of data used in each image. We evaluate the effectiveness of our model by pre-training on a number of benchmarks, finding that it outperforms other masking strategies, such as FLIP, on the quality of the learned representation.