Official implementation for Fast and Efficient Image Quality Enhancement via Desubpixel Convolutional Neural Networks, ECCV workshop 2018
Please cite our project if it is helpful for your research
@InProceedings{Vu_2018_ECCV_Workshops},
author = {Vu, Thang and Van Nguyen, Cao and Pham, Trung X. and Luu, Tung M. and Yoo, Chang D.},
title = {Fast and Efficient Image Quality Enhancement via Desubpixel Convolutional Neural Networks},
booktitle = {The European Conference on Computer Vision (ECCV) Workshops},
month = {September},
year = {2018}
}
Comparison of proposed FEQE with other state-of-the-art super-resolution and enhancement methods
Network architecture
Proposed desubpixel
TEAM_ALEX placed the first in overall benchmark score. Refer to PIRM 2018 for details.
Python3
tensorflow 1.10+
tensorlayer 1.9+
tensorboardX 1.4+
FEQE/
├── checkpoint
│ ├── FEQE
│ └── FEQE-P
├── data
│ ├── DIV2K_train_HR
│ ├── DIV2K_valid_HR_9
│ └── test_benchmark
├── docs
├── model
├── results
└── vgg_pretrained
└── imagenet-vgg-verydeep-19.mat
data/
directorycheckpoint/
directorypython test.py --dataset <DATASET_NAME>
results/
directorydata/
directoryvgg_pretrained/
directorypython train.py --checkpoint checkpoint/mse_s2 --alpha_vgg 0 --scale 2 --phase pretrain
python main.py --checkpoint checkpoint/mse_s4 --alpha_vgg 0 --pretrained_model checkpoint_test/mse_s2/model.ckpt
python main.py --checkpoint checkpoint/full_s4 ---pretrained_model checkpoint_test/mse_s4/model.ckpt
checkpoint/
direcorytensorboard --logdir checkpoint
YOUR_IP:6006
to your web browser.python test.py --dataset <DATASET> --model_path <FEQE-P path>
PSNR/SSIM/Perceptual-Index comparison. Red indicates the best results
Running time comparison. Red indicates the best results
Qualitative comparison