Dual-domain Modulation Network for Lightweight Image Super-Resolution

Paper (ArXiv) | Paper (IEEE)

---

Contents

The contents of this repository are as follows:

1. Dependencies
2. Train
3. Test

Dataset

We used only the first 800 images of DIV2K dataset to train our model.

The test set including Set5, Set14, B100, Urban100, Manga109, which can be downloaded from here.[Password:8888]

The code and datasets need satisfy the following structures:

├── DMNet  					# Train / Test Code
├── dataset  					# all datasets for this code
|  └── DIV2K_decoded  		#  train datasets with npy format
|  |  └── DIV2K_train_HR  		
|  |  └── DIV2K_train_LR_bicubic 			
|  └── benchmark  		#  test datasets with png format 
|  |  └── Set5
|  |  └── Set14
|  |  └── B100
|  |  └── Urban100
|  |  └── Manga109
 ─────────────────

---

Results

Our SR Results can be downloaded from here.

Pretrained models can be found in experiments.

---

Dependencies

• torch 2.4.1
• basicsr 1.4.2
• cupy-cuda12x 12.3.0

# For install basicsr
pip install basicsr==1.4.2

python setup.py develop -i http://mirrors.aliyun.com/pypi/simple/

python -m pip install --upgrade pip

pip install numpy==1.24.4

pip install -v -e .

Train

# For X2
torchrun --nproc_per_node=$GPU_NUM$ basicsr/train.py -opt options/train_DMNetx2.yml --launcher pytorch

# For X3
torchrun --nproc_per_node=$GPU_NUM$ basicsr/train.py -opt options/train_DMNetx3.yml --launcher pytorch

# For X4
torchrun --nproc_per_node=$GPU_NUM$ basicsr/train.py -opt options/train_DMNetx4.yml --launcher pytorch

Test

# For X2
python basicsr/test.py -opt options/test_DMNetx2.yml

# For X3
python basicsr/test.py -opt options/test_DMNetx3.yml

# For X4
python basicsr/test.py -opt options/test_DMNetx4.yml

Parmas and FLOPs

# Calculation DMNet
python inference.py

---

Citation

If this work is helpful for your research, please consider citing the following BibTeX entry.

@article{li2025dual,
  title={Dual-domain modulation network for lightweight image super-resolution},
  author={Li, Wenjie and Guo, Heng and Hou, Yuefeng and Gao, Guangwei and Ma, Zhanyu},
  journal={IEEE Transactions on Multimedia},
  year={2025}
}

Acknowledgement

The foundation for the training process is BasicSR , which profited from the outstanding contribution of XPixelGroup .

Contact

This repo is currently maintained by lewj2408@gmail.com and is for academic research use only.