This repository contains official implementation of our ACM MM 2025 paper "DSDNet: Raw Domain Demoiréing via Dual Color-Space Synergy", by Qirui Yang, Fangpu Zhang, Yeying Jin, Qihua Cheng, Peng-Tao Jiang, Huanjing Yue, Jingyu Yang.

We propose a single-stage raw domain demoiréing framework, Dual-Stream Demoiréing Network (DSDNet), which leverages the synergy of raw and YCbCr images to remove moiré while preserving luminance and color fidelity.

Please download the TMM22 and NIPS23 datasets from TMM22 and NIPS23.

Please download the results from GoogleDrive or BaiduNetdisk:pq4s.

Please download the pre-train models from GoogleDrive or BaiduNetdisk:ewnv, place them as follows:

DSDNet
└── weights
    ├── nips23.pth
    └── tmm22.pth

For quick test , run the scipts:

# TMM22 dataset
python test.py -opt ./options/test/test_tmm22.yml

# Nips23 dataset
python test.py -opt ./options/test/test_nips23.yml

# TMM22 dataset
python train.py -opt ./options/train/train_tmm22.yml

# Nips23 dataset
python train.py -opt ./options/train/train_nips23.yml

If you find this work useful for your research, please cite:

@inproceedings{yang2025dsdnet,
  title={Dsdnet: Raw domain demoir{\'e}ing via dual color-space synergy},
  author={Yang, Qirui and Zhang, Fangpu and Jin, Yeying and Cheng, Qihua and Jiang, Pengtao and Yue, Huanjing and Yang, Jingyu},
  booktitle={Proceedings of the 33rd ACM International Conference on Multimedia},
  pages={7230--7238},
  year={2025}
}

We thank the authors of RRID, RDNet, VD_raw and for sharing their codes