ESRGAN+ nESRGAN+ Tarsier

ICASSP 2020 - ESRGAN+ : Further Improving Enhanced Super-Resolution Generative Adversarial Network

Paper arXiv

Paper IEEE Xplore

ICPR 2020 - Tarsier: Evolving Noise Injection in Super-Resolution GANs

Paper arXiv

Paper IEEE Xplore

<p align="center"> <img height="250" src="./figures/noise_per_residual_dense_block.PNG"> </p> <p align="center"> <img src="./figures/qualitative_result.PNG"> </p>

Dependencies

• Python 3 (Recommend to use Anaconda)
• PyTorch >= 1.0.0
• NVIDIA GPU + CUDA
• Python packages: pip install numpy opencv-python lmdb tensorboardX

How to test

1. Place your low-resolution images in test_image/LR folder.
2. Download pretrained models from Google Drive and place them in test_image/pretrained_models.
3. Run the command: python test_image/test.py test_image/pretrained_models/nESRGANplus.pth (or any other models).
4. The results are in test_image/results folder.

How to train

1. Prepare the datasets which can be downloaded from Google Drive.
2. Prepare the PSNR-oriented pretrained model (all pretrained models can be downloaded from Google Drive).
3. Modify the configuration file codes/options/train/train_ESRGANplus.json.
4. Run the command python train.py -opt codes/options/train/train_ESRGANplus.json.

Acknowledgement

• This code is based on BasicSR.

Citation

@INPROCEEDINGS{9054071,
    author = {N. C. {Rakotonirina} and A. {Rasoanaivo}},  
    booktitle={ICASSP 2020 - 2020 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)},   
    title={ESRGAN+ : Further Improving Enhanced Super-Resolution Generative Adversarial Network},   
    year={2020},  
    volume={},  
    number={},  
    pages={3637-3641},}

@INPROCEEDINGS{9413318,
    author={Roziere, Baptiste and Rakotonirina, Nathanaël Carraz and Hosu, Vlad and Rasoanaivo, Andry and Lin, Hanhe and Couprie, Camille and Teytaud, Olivier},
    booktitle={2020 25th International Conference on Pattern Recognition (ICPR)}, 
    title={Tarsier: Evolving Noise Injection in Super-Resolution GANs}, 
    year={2021},
    volume={},
    number={},
    pages={7028-7035},
    keywords={Training;Image quality;Gaussian noise;Superresolution;Quality assessment;Pattern recognition;Standards},
    doi={10.1109/ICPR48806.2021.9413318}}