GMAIL: Generative Modality Alignment for generated Image Learning

TL;DR

GMAIL introduces a novel framework that treats generated images as a separate modality from real images, aligning them in a shared latent space to improve vision-language task performance.

Contribution

The paper proposes a multi-modal learning approach that explicitly aligns generated images with real images in latent space, enhancing the utility of synthetic data for vision-language tasks.

Findings

Significant improvements in image captioning and retrieval tasks.

Effective modality alignment boosts performance across multiple vision-language benchmarks.

Positive scaling trends with increased generated data volume.

Abstract

Generative models have made it possible to synthesize highly realistic images, potentially providing an abundant data source for training machine learning models. Despite the advantages of these synthesizable data sources, the indiscriminate use of generated images as real images for training can even cause mode collapse due to modality discrepancies between real and synthetic domains. In this paper, we propose a novel framework for discriminative use of generated images, coined GMAIL, that explicitly treats generated images as a separate modality from real images. Instead of indiscriminately replacing real images with generated ones in the pixel space, our approach bridges the two distinct modalities in the same latent space through a multi-modal learning approach. To be specific, we first fine-tune a model exclusively on generated images using a cross-modality alignment loss and then employ this aligned model to further train various vision-language models with generated images. By aligning the two modalities, our approach effectively leverages the benefits of recent advances in generative models, thereby boosting the effectiveness of generated image learning across a range of vision-language tasks. Our framework can be easily incorporated with various vision-language models, and we demonstrate its efficacy throughout extensive experiments. For example, our framework significantly improves performance on image captioning, zero-shot image retrieval, zero-shot image classification, and long caption retrieval tasks. It also shows positive generated data scaling trends and notable enhancements in the captioning performance of the large multimodal model, LLaVA.