ColorFM Stage 1 Color-Transfer Dataset

This dataset contains image triplets generated for training the ColorFM-L model: a content image, a style image, and the corresponding ColorFM-O color-transfer result.

Dataset Details

Dataset Description

The dataset was prepared for Stage 2 offline training of ColorFM-L. Stage 1 uses ColorFM-O to generate a color-transfer target for a content/style image pair.

• Repository: heyh97791/ColorFM
• Paper: ColorFM on arXiv
• Generated target resolution: 512 × 512

Dataset Sources

The source photographs were obtained from Unsplash. The generated target images were produced by applying ColorFM-O to content/style pairs.

Dataset Structure

The repository is organized by image category. Each category has an original-image folder and a corresponding styled_<category> folder. For example:

animal/
  001.jpg
  002.jpg
styled_animal/
  001/
    001_002.png
    001_003.png
  002/
    002_001.png

A generated filename contentID_styleID.png identifies the content image and style image used to create that target. The content and style images are stored once in the original-image folder and may be reused across multiple pairs.

The available categories are animal, city_street_building, indoor, nature, portraits, random, and still_life.

The release contains the available generated pairs after curation; it should not be assumed to contain every possible pair.

Source Data and Rights

The original images in dataset ColorFM were sourced from Unsplash and are used solely for non-commercial academic research. Copyright remains with the respective photographers; the images are not relicensed under this project's license. For rights-related inquiries or removal requests, please contact us.

Citation

If you use ColorFM, please cite:

@inproceedings{he2026colorfm,
  title={ColorFM: An Optimization-to-Learning Framework for Color Transfer via Flow Matching},
  author={He, Yuhang and Zhang, Kai and Li, Xiaoming and Chen, Du and Yang, Jian},
  booktitle={European Conference on Computer Vision},
  year={2026}
}