Yesianrohn/UnionST-Models

UnionST: A Strong Synthetic Engine for Scene Text Recognition

This repository contains model checkpoints for UnionST, introduced in the paper What Is Wrong with Synthetic Data for Scene Text Recognition? A Strong Synthetic Engine with Diverse Simulations and Self-Evolution.

Introduction

Scene Text Recognition (STR) relies critically on large-scale, high-quality training data. While synthetic data provides a cost-effective alternative to manually annotated real data, existing rendering-based synthetic datasets often suffer from a domain gap with real-world text due to insufficient diversity.

UnionST is a strong data engine that synthesizes text covering a union of challenging samples to better align with the complexity observed in the wild. Models trained on the resulting UnionST-S dataset achieve significant improvements over traditional synthetic datasets on challenging STR benchmarks.

Resources

• Paper: arXiv:2602.06450
• Code: GitHub - YesianRohn/UnionST
• Datasets: UnionST Dataset on Hugging Face

Training

The models (such as SVTRv2-AR) are implemented using the OpenOCR framework. Training can be initiated with:

cd OpenOCR
torchrun --nproc_per_node=8 tools/train_rec.py --c configs/rec/nrtr/svtrv2_nrtr_unionst.yml

Citation

If you find this work useful, please cite:

@inproceedings{ye2026wrong,
  title={What's Wrong with Synthetic Data for Scene Text Recognition? A Strong Synthetic Engine with Diverse Simulations and Self-Evolution},
  author={Ye, Xingsong and Du, Yongkun and Zhang, JiaXin and Li, Chen and LYU, Jing and Chen, Zhineng},
  booktitle={CVPR},
  year={2026}
}

License

This project is licensed under the MIT License.