HIIF (CVPR2025) [Paper]

<p align="center"> <img src="asset/HIIF_archi.png" style="border-radius: 15px"> </p>

conda create -n hiif python=3.10.14
conda init
conda activate hiif

pip install torch==2.1.1 torchvision==0.16.1 torchaudio==2.1.1 --index-url https://download.pytorch.org/whl/cu118
pip install tqdm
pip install einops
pip install typing_extensions
pip install imageio
pip install timm
pip install pytorch_wavelets
pip install opencv-python
pip install matplotlib
pip install tensorboardX

<a name="cite"></a> Citation

Please cite us if our work is useful for your research.

@article{jiang2024hiif,
  title={HIIF: Hierarchical Encoding based Implicit Image Function for Continuous Super-resolution},
  author={Jiang, Yuxuan and Kwan, Ho Man and Peng, Tianhao and Gao, Ge and Zhang, Fan and Zhu, Xiaoqing and Sole, Joel and Bull, David},
  journal={arXiv preprint arXiv:2412.03748},
  year={2024}
}

📑 Train & Test

EDSR-HIIF

python train.py --config configs/train-div2k/train_edsr-baseline-hiif.yaml
bash ./scripts/test-div2k-fast.sh ./save/edsr_hiif.pth 0

---

📊 Model Summary

DIV2K pre-trained model Model|Download :-:|:-: edsr_hiif|Google Drive rdf_hiif|Google Drive swinir_hiif|Google Drive

🔍 Qualitative Results

<p align="center"> <img width="800" src="asset/vs1.png"> </p> <p align="center"> <img width="800" src="asset/vs2.png"> </p> <p align="center"> <img width="800" src="asset/vs3.png"> </p> <p align="center"> <img width="800" src="asset/vs4.png"> </p> <p align="center"> <img width="800" src="asset/vs5.png"> </p>

🔍 Arbitrary-Scale demo

python demo.py --input_folder [LR path] --output_folder [Output path] --model ./save/edsr_hiif.pth --scale 12.34

<p align="center"> <img width="800" src="asset/vs6.png"> </p>

Acknowledgements

This code is built on LIIF and LTE. We thank the authors for sharing their codes.

---