📖 [IEEE Signal Processing Letters] Adaptive Feature Selection Modulation Network for Efficient Image Super-Resolution [Paper]

Chen Wu, Ling Wang, Xin Su, and Zhuoran Zheng

---

Requirements

• Python 3.8, PyTorch >= 1.11
• BasicSR 1.4.2
• Platforms: Ubuntu 18.04, cuda-11

Installation

# Clone the repo
git clone https://github.com/DavisWANG0/AFSMNet.git
# Install dependent packages
cd AFSMNet
pip install -r requirements.txt
# Install BasicSR
python setup.py develop

You can also refer to this INSTALL.md for installation

Training

Run the following commands for training:

# train AFSMNet for x2 efficient SR
python basicsr/train.py -opt options/train/AFSMNet/train_DF2K_x2.yml
# train AFSMNet for x3 efficient SR
python basicsr/train.py -opt options/train/AFSMNet/train_DF2K_x3.yml
# train AFSMNet for x4 efficient SR
python basicsr/train.py -opt options/train/AFSMNet/train_DF2K_x4.yml

Testing

• Download the testing dataset.
• Run the following commands:

# test AFSMNet for x2 efficient SR
python basicsr/test.py -opt options/test/AFSMNet/test_benchmark_x2.yml
# test AFSMNet for x3 efficient SR
python basicsr/test.py -opt options/test/AFSMNet/test_benchmark_x3.yml
# test AFSMNet for x4 efficient SR
python basicsr/test.py -opt options/test/AFSMNet/test_benchmark_x4.yml

• The test results will be in './results'.

Results

Citation

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

@article{wu2025adaptive,
  title={Adaptive feature selection modulation network for efficient image super-resolution},
  author={Wu, Chen and Wang, Ling and Su, Xin and Zheng, Zhuoran},
  journal={IEEE Signal Processing Letters},
  year={2025},
  publisher={IEEE}
}

Acknowledgement

This code is based on BasicSR toolbox and SAFMN. Thanks for the awesome work.

Contact

If you have any questions, please feel free to reach me out at [email protected]