README.md

Datasets

Deraindrop

• DeRainDrop(train & test):
  • train & test

Dehaze

• I-Haze(train):
  • train
• O-Haze(train):
  • train
• Dense-haze(test):
  • test
• NH-haze(test):
  • test

Deblur

• GoPro(train & test):
  • train
  • test
• HIDE(test):
  • test

Preprocess

You have to crop the high-resolution training images to fixed size training patches by generate_patches.py.
Our experiment setting can refer Details of Hyperparameters of train section in main README.

Tree

datasets
  ├── deraindrop  
  |    ├── test
  |    |     ├── input
  |    |     └── target    
  |    └── train
  |          ├── input
  |          └── target    
  ├── dehaze
  |    ├── test
  |    |     ├── input
  |    |     └── target    
  |    └── train
  |          ├── input
  |          └── target    
  └── deblur
       ├── test
       |     ├── input
       |     └── target    
       └── train
             ├── input
             └── target