liuguoyou/dssn_release

Deep Spectral-Spatial Network (DSSN) for Single Image Deblurring

DSSN_release (we reuploaded the codes after errors are corrected)

This repository is a Pytorch implementation of the paper "Deep Spectral-Spatial Network for Single Image Deblurring"

Seokjae Lim, Jin Kim and Wonjun Kim
IEEE Signal Processing Letters

When using this code in your research, please cite the following paper:

Seokjae Lim, Jin Kim and Wonjun Kim, "Deep Spectral-Spatial Network for Single Image Deblurring," IEEE Signal Processing Letters vol. 27, no. 1, pp. 835-839, May 2020.

@ARTICLE{9094296,
author={S. {Lim} and J. {Kim} and W. {Kim}},
journal={IEEE Signal Processing Letters}, 
title={Deep Spectral-Spatial Network for Single Image Deblurring}, 
year={2020},
volume={27},
number={1},
pages={835-839},}

Model architecture

Experimental results with state-of-the art methods on the GOPRO dataset

Several results of single image deblurring. First row : input blurry images selected from the GOPRO dataset. Second row : deblurring results by Tao et al. Third row : deblurring result by Zhang et al. Fourth row : deblurring results by the proposed method. Note that blurred regions (red and yellow-colored rectangles) are enlarged for better view. Best views in colors.

Experimental results with state-of-the art methods on the Köhler dataset

Several results of single image deblurring. First column : input blurry images selected from the Köhler dataset. Second column : deblurring results by Nah et al. Third column : deblurring result by Zhang et al. Fourth column : deblurring results by the proposed method. Note that all experiments are conducted with parameters, which are trained based on the GOPRO dataset, without any modification.

Requirements

• Python >= 3.5
• Pytorch 0.4.0
• Ubuntu 16.04
• CUDA 8 (if CUDA available)
• cuDNN (if CUDA available)

Pretrained models

You can download pretrained DSSN model

• Trained with GOPRO

Test result

You can download test results of our DSSN Model

• Test result with GOPRO

Note

1. you should place the weights in the ./data/model/
2. Dataset is also placed in the ./data directory (i.e., ./data/GoPro_Large)
3. test results are saved in the ./data/result/

Training

• Deep Spectral-Spatial network training

python main_ULT.py n

Testing

• Deep Spectral-Spatial network testing

python main_ULT.py t