Direct Unsupervised Super-Resolution using Generative Adversarial Network (DUS-GAN) for Real-World Data

The repository contains the official code for the work "Direct Unsupervised Super-Resolution using Generative Adversarial Network (DUS-GAN) for Real-World Data" published in an IEEE Transaction in Image Processing (TIP).

- Pre-Trained models

The pre-trained model for is shared with the repository..

- Training the model

Training code has been released. To train the network, run the following command.

python train.py -opt path_for_training_json_file

Note the following changes are needed to run the code.

• Need to provide pre-train QA network path at line number 307 for model/DS_Model.py file.
• Change the root folder and training dataset path into train_ntireEx.json file located at options/train folder.

- Testng the model

To test your/our pre-trained model, you need to set root directory and dataset directory into options/test/test_ntire_ex.json file. Then run the following command to start the training.

python test.py -opt PATH-to-json_file

- Requirement of packages

The list of packages required to run the code is given in chasnet.yml file.

We are thankful to Xinntao for their ESRGAN code using which this work has been implemented. For any problem, you may contact at [email protected].