/progressive-dehaze

Progressive Update Guided Interdependent Networks for Single Image Dehazing

Primary LanguagePythonMIT LicenseMIT

Progressive Update Guided Interdependent Networks for Single Image Dehazing

This is the PyTorch implementation for our paper:

**Aupendu Kar, Sobhan Kanti Dhara, Debashis Sen, Prabir Kumar Biswas. Progressive Update Guided Interdependent Networks for Single Image Dehazing. [Project Website]

Data Preparation (NR-Haze Dataset)

Training data

1.1 Download the Training Set from Google Drive

1.2 Put the dataset in your_data_path as follows:

your_data_path
└── train
        ├── NR-Indoor
                    ├── Clean
                            ├── 1.png
                            ├── .....
                            └── 1349.png
                    └── Depth
                            ├── 1.mat
                            ├── .....
                            └── 1349.mat
        └── NR-Outdoor
                    ├── Clean
                            ├── 0002.png
                            ├── .....
                            └── 8961.png
                    └── Depth
                            ├── 0002.mat
                            ├── .....
                            └── 8961.mat

Validation and Test data

1.1 Download the Validation Set from Google Drive

1.2 Download the 7 Test Sets from Google Drive

Training Codes are in TrainCode folder

Test Codes are in TestCode folder

Contact

Aupendu Kar: mailtoaupendu[at]gmail[dot]com