VISTA: Visualized Text Embedding For Universal Multi-Modal Retrieval

TL;DR

VISTA is a novel multi-modal retrieval model that extends text encoders with visual understanding, using innovative data generation and training strategies to achieve superior zero-shot and supervised retrieval performance.

Contribution

The paper introduces a flexible architecture, data generation strategies, and a multi-stage training algorithm for universal multi-modal retrieval.

Findings

VISTA outperforms existing models in multi-modal retrieval tasks.

Effective zero-shot and supervised performance across datasets.

High-quality composed image-text data enhances training.

Abstract

Multi-modal retrieval becomes increasingly popular in practice. However, the existing retrievers are mostly text-oriented, which lack the capability to process visual information. Despite the presence of vision-language models like CLIP, the current methods are severely limited in representing the text-only and image-only data. In this work, we present a new embedding model VISTA for universal multi-modal retrieval. Our work brings forth threefold technical contributions. Firstly, we introduce a flexible architecture which extends a powerful text encoder with the image understanding capability by introducing visual token embeddings. Secondly, we develop two data generation strategies, which bring high-quality composed image-text to facilitate the training of the embedding model. Thirdly, we introduce a multi-stage training algorithm, which first aligns the visual token embedding with the text encoder using massive weakly labeled data, and then develops multi-modal representation capability using the generated composed image-text data. In our experiments, VISTA achieves superior performances across a variety of multi-modal retrieval tasks in both zero-shot and supervised settings. Our model, data, and source code are available at https://github.com/FlagOpen/FlagEmbedding.