YeTI Dataset

ECCV 2026 · Paper (arXiv:2607.09193) · Code (GitHub) · Model (Hugging Face)

Preprocessed training / validation archives used by YeTI: You Only Need Two Noisy Images for Real-World sRGB Noise Generation (ECCV 2026). These are derived from the public SIDD (Smartphone Image Denoising Dataset) and MAI2021 (Mobile AI 2021 Real Image Denoising Challenge) benchmarks, repackaged into the patch / crop format expected by this codebase.

Files

Archive	Description	Used for
sidd_patch.tar.gz	SIDD training patches	RAE / C-DiT training
sidd_noisy.tar.gz	SIDD noisy images (full)	Denoiser (AP-BSN / MM-BSN) training
sidd_val.tar.gz	SIDD validation set	Validation
siddplus_val.tar.gz	SIDD+ validation set	Validation
mai_val.tar.gz	MAI2021 validation set	Validation

How to use

# download everything
huggingface-cli download BWLim/YeTI --repo-type dataset --local-dir data

# or a single archive
huggingface-cli download BWLim/YeTI sidd_val.tar.gz --repo-type dataset --local-dir data

Extract the archive(s) you need and point the dataroot field of the relevant config under configs/datasets/ (in the code repository) at the extracted path. See the repository README for full dataset preparation and usage instructions.

Source datasets

These archives are derived from the following public benchmarks; please also refer to their original terms of use:

• SIDD: Smartphone Image Denoising Dataset — https://www.eecs.yorku.ca/~kamel/sidd/
• MAI2021: Mobile AI 2021 Real Image Denoising Challenge

Citation

@article{ko2026yeti,
  title   = {YeTI: You Only Need Two Noisy Images for Real-World sRGB Noise Generation},
  author  = {Ko, Jaekyun and Lim, Byung Wan and Lee, Soomin and Kim, Dongjin and Kim, Tae Hyun},
  journal = {arXiv preprint arXiv:2607.09193},
  year    = {2026}
}

License

Released for academic / non-commercial research use only, in addition to the terms of the original source datasets (SIDD, MAI2021). See LICENSE for full terms.