Leveraging Unpaired Data for Vision-Language Generative Models via Cycle Consistency

TL;DR

This paper introduces ITIT, a cycle consistency-based training framework that enables vision-language models to learn effectively from unpaired image and text data, reducing reliance on costly paired datasets.

Contribution

ITIT is the first framework to leverage cycle consistency for training vision-language models on unpaired data, achieving competitive performance with significantly less paired data.

Findings

ITIT performs comparably to state-of-the-art models using only 3 million paired samples.

The model effectively learns bidirectional image-text generation from unpaired datasets.

Cycle consistency enforces meaningful alignment between images and texts during training.

Abstract

Current vision-language generative models rely on expansive corpora of paired image-text data to attain optimal performance and generalization capabilities. However, automatically collecting such data (e.g. via large-scale web scraping) leads to low quality and poor image-text correlation, while human annotation is more accurate but requires significant manual effort and expense. We introduce $\textbf{ITIT}$ ($\textbf{I}$n$\textbf{T}$egrating $\textbf{I}$mage $\textbf{T}$ext): an innovative training paradigm grounded in the concept of cycle consistency which allows vision-language training on unpaired image and text data. ITIT is comprised of a joint image-text encoder with disjoint image and text decoders that enable bidirectional image-to-text and text-to-image generation in a single framework. During training, ITIT leverages a small set of paired image-text data to ensure its output matches the input reasonably well in both directions. Simultaneously, the model is also trained on much larger datasets containing only images or texts. This is achieved by enforcing cycle consistency between the original unpaired samples and the cycle-generated counterparts. For instance, it generates a caption for a given input image and then uses the caption to create an output image, and enforces similarity between the input and output images. Our experiments show that ITIT with unpaired datasets exhibits similar scaling behavior as using high-quality paired data. We demonstrate image generation and captioning performance on par with state-of-the-art text-to-image and image-to-text models with orders of magnitude fewer (only 3M) paired image-text data.