Skip to content

A Hierarchical Generative Adversarial Network for Blind Denoising of Real Photographs

Notifications You must be signed in to change notification settings

RealtimeDenoising/HI-GAN

 
 

Repository files navigation

HI-GAN: A Hierarchical Generative Adversarial Network for Blind Denoising of Real Photographs

Requirements

  • python 3.6
  • pytorch 1.0
  • CUDA 8.0 or 9.0

Dataset

  1. Download train dataset from Smartphone Image Denoising Dataset Medium Dataset[1]
  2. Download test dataset from The Darmstadt Noise Dataset (DND)[2], SIDD Benchmark[1], NAM[3] ,NC12[4]

[1] Abdelrahman Abdelhamed, Lin S., Brown M. S. A High-Quality Denoising Dataset for Smartphone Cameras. CVPR, 2018.

[2] T. Plotz, and S. Roth. Benchmarking denoising algorithms with real photographs. In 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), pages 2750–2759, 2017

[3] Seonghyeon Nam*, Youngbae Hwang*, Yasuyuki Matsushita, Seon Joo Kim. A Holistic Approach to Cross-Channel Image Noise Modeling and its Application to Image Denoising. CVPR, 2016.

[4] M. Lebrun, M. Colom, and J.-M. Morel. The noise clinic: A blind image denoising algorithm. In Image Processing On Line, vol. 5, pp. 1–54, 2015.

Usage

1. Test

python test_dnd_nam.py

Optional

  • --inp : input folder
  • --out : output folder
  • --JPEG : for JPEG images such as "NAM_20_rand_pathes" testset (don't use this argument with non-JPEG images such as "DND_20_rand_patches" testset)
  • --nGpu : number of GPU

2. Demo cell

python demo_cell.py

Optional

  • --inp-dir : input image (select image on 'testsets/cell/demo' folder), default='testsets/cell/demo/avg2/Confocal_BPAE_B_4.png'
  • --out : output folder
  • --net : choices=['N2N', 'DnCNN','UNet_ND', 'UNet_D', 'HI_GAN'], default='HI_GAN'
  • --gray : Gray image, default=True (Create RGB image from 3 gray images)
  • --nGpu : number of GPU

3. Test cell (Reproduce test example in the paper)

python test_cell.py

4. Sample results

DND

NAM NC12 SIDD FMD

About

A Hierarchical Generative Adversarial Network for Blind Denoising of Real Photographs

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • Jupyter Notebook 85.8%
  • Python 14.2%