IDN-tensorflow

Testing

Install Tensorflow 1.11, Matlab R2017a
Download Test datasets
Modify config.py (if you want to test x3 model on Set14, config.TEST.model_path = 'checkpoint_x3/model.ckpt' config.TEST.dataset = 'Set14') and test.py (scale = 3).
Run testing:

python test.py

Training

Download Training dataset
Modify config.py (if you want to train x4 model, config.TRAIN.hr_img_path = '/path/to/DIV2K_train_HR/' config.TRAIN.checkpoint_dir = 'checkpoint_x4/' config.VALID.hr_img_path = '/path/to/DIV2K_valid_HR/' config.VALID.lr_img_path = '/path/to/DIV2K_valid_LR_x4/') and train_SR.py (scale = 4)
Run training:

python train_SR.py

Note

This TensorFlow version is trained with DIV2K training dataset on RGB channels. Additionally, We modify the upsample layer to subpixel convolution (the original version is transposed convolution).

Results

Test_results

The following PSNR/SSIMs are evaluated on Matlab R2017a and the code can be referred to Evaluate_PSNR_SSIM.m.

_{Training dataset}	_Scale	_Set5	_Set14	_B100	_Urban100
₂₉₁	_×2	_{37.83 / 0.9600}	_{33.30 / 0.9148}	_{32.08 / 0.8985}	_{31.27 / 0.9196}
_DIV2K	_×2	_{37.85 / 0.9598}	_{33.58 / 0.9178}	_{32.11 / 0.8989}	_{31.95 / 0.9266}
₂₉₁	_×3	_{34.11 / 0.9253}	_{29.99 / 0.8354}	_{28.95 / 0.8013}	_{27.42 / 0.8359}
_DIV2K	_×3	_{34.24 / 0.9260}	_{30.27 / 0.8408}	_{29.03 / 0.8038}	_{27.99 / 0.8489}
₂₉₁	_×4	_{31.82 / 0.8903}	_{28.25 / 0.7730}	_{27.41 / 0.7297}	_{25.41 / 0.7632}
_DIV2K	_×4	_{31.99 / 0.8928}	_{28.52 / 0.7794}	_{27.52 / 0.7339}	_{25.92 / 0.7801}

Model Parameters

_Scale	_{Model size}
_×2	_579,276
_×3	_587,931
_×4	_600,048

Citation

If you find IDN useful in your research, please consider citing:

@inproceedings{Hui-IDN-2018,
  title={Fast and Accurate Single Image Super-Resolution via Information Distillation Network},
  author={Hui, Zheng and Wang, Xiumei and Gao, Xinbo},
  booktitle={CVPR},
  pages = {723--731},
  year={2018}
}

Zheng222 / IDN-tensorflow