NTIRE_test_code

These are all test scripts of our Grouped Residual Dense Network(GRDN) for NTIRE 2019 Real Image Denoising Challenge

"WINNER of NTIRE 2019 Real Image Denoising Challenge:sRGB track"

Paper uploaded on arXiv!

https://arxiv.org/abs/1905.11172

The challenge name

NTIRE2019 Real Image Denoising Challenge

PLEASE READ

We wanted to upload codes, models, and input mat file, but we had trouble when we try to reproduce our results because to upload files over 100mb, We need to use 'git lfs' but it may modify models and test inputs so there will be errors. So we decided to upload all files needed to reproduce our model at Google Drive and will share download link below:

Download link

https://drive.google.com/file/d/1hAPTDGYl4JSo0LUE1xdyYlIRsKzw8zYy/view?usp=sharing

Prerequisites

• pytorch 1.0
• python 3.6 or 3.7
• cv2
• numpy
• tqdm
• time
• scipy

How to Run

Track 1 : please run 'DGU-3DMlab1_track1.py'
Track 2 : please run 'DGU-3DMlab1_track2.py'

Full version for training

No plan

Realistic noise simulation(using GAN) test code

https://github.com/spnova12/NoiseMakingGan