DocTron-Formula: Generalized Formula Recognition in Complex and Structured Scenarios

TL;DR

DocTron-Formula introduces a unified vision-language framework for recognizing complex mathematical formulas, outperforming specialized models and enabling robust understanding of scientific documents across diverse styles and structures.

Contribution

It presents a generalized OCR approach for formulas using a unified model and introduces CSFormula, a large-scale dataset for complex scientific content recognition.

Findings

Achieves state-of-the-art accuracy on formula recognition tasks.

Outperforms specialized models in robustness and versatility.

Establishes a new paradigm for scientific document understanding.

Abstract

Optical Character Recognition (OCR) for mathematical formula is essential for the intelligent analysis of scientific literature. However, both task-specific and general vision-language models often struggle to handle the structural diversity, complexity, and real-world variability inherent in mathematical content. In this work, we present DocTron-Formula, a unified framework built upon general vision-language models, thereby eliminating the need for specialized architectures. Furthermore, we introduce CSFormula, a large-scale and challenging dataset that encompasses multidisciplinary and structurally complex formulas at the line, paragraph, and page levels. Through straightforward supervised fine-tuning, our approach achieves state-of-the-art performance across a variety of styles, scientific domains, and complex layouts. Experimental results demonstrate that our method not only surpasses specialized models in terms of accuracy and robustness, but also establishes a new paradigm for the automated understanding of complex scientific documents.