RRCGAN：A Radiometric Resolution Compression Method for Optical Remote Sensing Images Using Contrastive Learning

Prerequisites

• Linux or macOS
• Python 3
• CPU or NVIDIA GPU + CUDA CuDNN

Update log

• 5/5/2024: Added related codes.
• 12/24/2024: Added GF2 dataset.
• 02/03/2025: Update dataset.

RRCGAN Train and Test

• Train the RRCGAN model:

python train.py --dataroot XXX --name XXX

• Test the RRCGAN model:

python test.py --dataroot XXX --name XXX

Datesets

All the data mentioned in the article, eg. JL1,GF7,GF2, has been uploaded to Baidu Cloud, link is: (https://pan.baidu.com/s/1eNE-5UvD4df62ehgllwYsA)(mhiy) .The dataset example is as follows:

Acknowledgments

Our code is developed based on contrastive-unpaired-translation and Hneg_SRC . We also thank pytorch-fid for FID computation.

Citation

@article{Zhang2025RRCGAN,
  title={RRCGAN: Unsupervised Compression of Radiometric Resolution of Remote Sensing Images Using Contrastive Learning},
  author={Tengda Zhang; Jiguang Dai; Jinsong Cheng; Hongzhou Li; Ruishan Zhao; Bing Zhang},
  journal={IEEE Transactions on Geoscience and Remote Sensing},
  year={2025}
}.