What Is Wrong with Synthetic Data for Scene Text Recognition? A Strong Synthetic Engine with Diverse Simulations and Self-Evolution

TL;DR

This paper introduces UnionST, a new synthetic data engine with diverse simulations for scene text recognition, and a self-evolution learning framework, significantly improving synthetic data effectiveness and reducing real data dependency.

Contribution

The paper presents UnionST, a novel synthetic data engine with enhanced diversity and realism, and a self-evolution learning method for efficient real data annotation in scene text recognition.

Findings

Models trained on UnionST-S outperform existing synthetic datasets.

UnionST-S can surpass real-data performance in some scenarios.

Using SEL, models achieve competitive results with only 9% of real labels.

Abstract

Large-scale and categorical-balanced text data is essential for training effective Scene Text Recognition (STR) models, which is hard to achieve when collecting real data. Synthetic data offers a cost-effective and perfectly labeled alternative. However, its performance often lags behind, revealing a significant domain gap between real and current synthetic data. In this work, we systematically analyze mainstream rendering-based synthetic datasets and identify their key limitations: insufficient diversity in corpus, font, and layout, which restricts their realism in complex scenarios. To address these issues, we introduce UnionST, a strong data engine synthesizes text covering a union of challenging samples and better aligns with the complexity observed in the wild. We then construct UnionST-S, a large-scale synthetic dataset with improved simulations in challenging scenarios. Furthermore, we develop a self-evolution learning (SEL) framework for effective real data annotation. Experiments show that models trained on UnionST-S achieve significant improvements over existing synthetic datasets. They even surpass real-data performance in certain scenarios. Moreover, when using SEL, the trained models achieve competitive performance by only seeing 9% of real data labels. Code is available at https://github.com/YesianRohn/UnionST.