RS-CLIP: Zero Shot Remote Sensing Scene Classification via Contrastive Vision-Language Supervision

Introduction

• In this paper, we introduced a CLIP-based vision-language model for zero-shot (few-shot) remote sensing scene classification.
• We introduced a pseudo-labeling technique that can automatically generate pseudo-labels from unlabeled data. Moreover, a curriculum learning strategy is developed to boost the performance of zero-shot remote sensing scene classification.
• We conducted experiments on four benchmark datasets and showed significantly better performance than previous state-of-the-art methods on both zero-shot and few-shot remote sensing scene classification.

Getting Started

Installation

Following CLIP to install required packages

Reproduce

refer to run_ucm.sh for experiments on the UCM dataest.

Acknowledgement

• The code is partially borrowed from CLIP. Don't forget to check this great work if you don't know it before!

If you're using RS-CLIP in your research or applications, please cite using this BibTeX:

@article{li2023rs,
  title={RS-CLIP: Zero shot remote sensing scene classification via contrastive vision-language supervision},
  author={Li, Xiang and Wen, Congcong and Hu, Yuan and Zhou, Nan},
  journal={International Journal of Applied Earth Observation and Geoinformation},
  volume={124},
  pages={103497},
  year={2023},
  publisher={Elsevier}
}

License

This repository is under BSD 3-Clause License. BSD 3-Clause License here.