TRINS: Towards Multimodal Language Models that Can Read

TL;DR

TRINS is a new dataset designed to improve multimodal language models' ability to read and understand text within images, addressing a key limitation of existing models by providing longer, more complex annotations.

Contribution

The paper introduces TRINS, a large text-rich image dataset, and a novel architecture LaRA that significantly enhances multimodal models' reading capabilities.

Findings

LaRA outperforms existing models on TRINS and classical benchmarks.

TRINS dataset contains over 39,000 text-rich images with long annotations.

Models trained on TRINS show improved understanding of textual content in images.

Abstract

Large multimodal language models have shown remarkable proficiency in understanding and editing images. However, a majority of these visually-tuned models struggle to comprehend the textual content embedded in images, primarily due to the limitation of training data. In this work, we introduce TRINS: a Text-Rich image INStruction dataset, with the objective of enhancing the reading ability of the multimodal large language model. TRINS is built upon LAION using hybrid data annotation strategies that include machine-assisted and human-assisted annotation processes. It contains 39,153 text-rich images, captions, and 102,437 questions. Specifically, we show that the number of words per annotation in TRINS is significantly longer than that of related datasets, providing new challenges. Furthermore, we introduce a simple and effective architecture, called a Language-vision Reading Assistant (LaRA), which is good at understanding textual content within images. LaRA outperforms existing state-of-the-art multimodal large language models on the TRINS dataset, as well as other classical benchmarks. Lastly, we conducted a comprehensive evaluation with TRINS on various text-rich image understanding and generation tasks, demonstrating its effectiveness.