README.md

metadata

dataset_info:
  features:
    - name: image
      dtype: image
    - name: text
      dtype: string
    - name: guide
      dtype: image
    - name: source
      dtype: string
    - name: control_type
      dtype: string
    - name: img_id
      dtype: int64
  splits:
    - name: full
      num_bytes: 123906287
      num_examples: 500
    - name: eval
      num_bytes: 40786402
      num_examples: 150
    - name: extra
      num_bytes: 83119885
      num_examples: 350
  download_size: 245420860
  dataset_size: 247812574
configs:
  - config_name: default
    data_files:
      - split: full
        path: data/full-*
      - split: eval
        path: data/eval-*
      - split: extra
        path: data/extra-*

Dataset Card

Dataset in ImagenHub.

Citation

Please kindly cite our paper if you use our code, data, models or results:

@article{ku2023imagenhub,
  title={ImagenHub: Standardizing the evaluation of conditional image generation models},
  author={Max Ku and Tianle Li and Kai Zhang and Yujie Lu and Xingyu Fu and Wenwen Zhuang and Wenhu Chen},
  journal={arXiv preprint arXiv:2310.01596},
  year={2023}
}