Xintian Mao, Yiming Liu, Fengze Liu, Qingli Li, Wei Shen and Yan Wang
Paper: xxx
code: https://github.com/INVOKERer/DeepRFT/tree/AAAI2023
Xintian Mao, Yiming Liu, Wei Shen, Qingli Li and Yan Wang
Paper: https://arxiv.org/abs/2111.11745
Overall Framework of DeepRFT |
The model is built in PyTorch 1.8.0 and tested on Ubuntu 18.04 environment (Python3.8, CUDA11.1).
For installing, follow these intructions
conda create -n pytorch python=3.8
conda activate pytorch
conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cudatoolkit=11.1 -c pytorch -c conda-forge
pip install matplotlib scikit-image opencv-python yacs joblib natsort h5py tqdm kornia tensorboard ptflops
Install warmup scheduler
cd pytorch-gradual-warmup-lr; python setup.py install; cd ..
To test the pre-trained models of Deblur and Defocus Google Drive or 百度网盘(提取码:phws) on your own images, run
python test.py --weights ckpt_path_here --input_dir path_to_images --result_dir save_images_here --win_size 256 --num_res 8 [4:small, 20:plus]# deblur
python test.py --weights ckpt_path_here --input_dir path_to_images --result_dir save_images_here --win_size 512 --num_res 8 # defocus
Here is an example to train:
python train.py
Experiment for image deblurring.
Deblurring on GoPro Datasets. |
- https://github.com/yangyanli/DO-Conv
- https://github.com/swz30/MPRNet
- https://github.com/chosj95/MIMO-UNet
- https://github.com/codeslake/IFAN
If you use DeepRFT, please consider citing:
@inproceedings{xint2023freqsel,
title = {Intriguing Findings of Frequency Selection for Image Deblurring},
author = {Xintian Mao, Yiming Liu, Fengze Liu, Qingli Li, Wei Shen and Yan Wang},
booktitle = {Proceedings of the 37th AAAI Conference on Artificial Intelligence},
year = {2023}
}
or
@inproceedings{,
title={Deep Residual Fourier Transformation for Single Image Deblurring},
author={Xintian Mao, Yiming Liu, Wei Shen, Qingli Li, Yan Wang},
booktitle={arXiv:2111.11745},
year={2021}
}
If you have any question, please contact [email protected]