---
license: mit
pipeline_tag: image-to-text
tags:
- ocr
- scene-text-recognition
- synthetic-data
---

# 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](https://huggingface.co/papers/2602.06450).

## 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](https://arxiv.org/abs/2602.06450)
- **Code:** [GitHub - YesianRohn/UnionST](https://github.com/YesianRohn/UnionST)
- **Datasets:** [UnionST Dataset on Hugging Face](https://huggingface.co/datasets/Yesianrohn/UnionST)

## Training
The models (such as SVTRv2-AR) are implemented using the [OpenOCR](https://github.com/Topdu/OpenOCR) framework. Training can be initiated with:

```bash
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:
```bibtex
@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.