metadata

dataset_info:
  features:
    - name: id
      dtype: string
    - name: hq_img
      dtype: image
    - name: lq_img
      dtype: image
    - name: text
      sequence: string
    - name: bbox
      sequence:
        array2_d:
          shape:
            - 2
            - 2
          dtype: int32
    - name: poly
      sequence:
        array2_d:
          shape:
            - 16
            - 2
          dtype: int32
  splits:
    - name: test
      num_bytes: 55089874
      num_examples: 847
  download_size: 54622145
  dataset_size: 55089874
configs:
  - config_name: default
    data_files:
      - split: test
        path: data/test-*
language:
  - en
size_categories:
  - 10M<n<100M
task_categories:
  - image-to-image
tags:
  - image-restoration
  - diffusion-models
  - text-recognition

Real-Text

Text-Aware Image Restoration with Diffusion Models (arXiv:2506.09993)
Real-world evaluation dataset for the TAIR task.

📄 Paper: https://arxiv.org/abs/2506.09993
🌐 Project Page: https://cvlab-kaist.github.io/TAIR/
💻 GitHub: https://github.com/cvlab-kaist/TAIR
🛠 Dataset Pipeline: https://github.com/paulcho98/text_restoration_dataset

Dataset Description

Real-Text is an evaluation dataset constructed from RealSR and DrealSR using the same pipeline as SA-Text. It reflects real-world degradation and distortion, making it suitable for robust benchmarking.

Notes

This dataset is designed for testing oour model, TeReDiff, under realistic settings.
Check SA-text for training dataset.
Please refer to our dataset pipeline.

Citation

Please cite the following paper if you use this dataset:

{
  @article{min2024textaware,
  title={Text-Aware Image Restoration with Diffusion Models},
  author={Min, Jaewon and Kim, Jin Hyeon and Cho, Paul Hyunbin and Lee, Jaeeun and Park, Jihye and Park, Minkyu and Kim, Sangpil and Park, Hyunhee and Kim, Seungryong},
  journal={arXiv preprint arXiv:2506.09993},
  year={2025}
}