This is the official code for the paper titled "Supervised Conversion from Landsat-8 Images to Sentinel-Images with Deep Learning."
The dataset can be downloaded here. Put the dataset in the directory named "dataset" at the same level as the code directory.
To train a model, run this command:
python main.py -m [MODEL NAME]
The model name options are: "sub", "submax", "trans", and "transmax".
To calculate the performance of each model, put the model_path.pth file generated by the training code the directory named as the corresponding model name in the save directory. Afterward, run this command:
python calculate_performance.py
Samples of predicted images and the corresponding ground truth can be downloaded here.
The distribution plot figures can be downloaded here.