Mind the Modality Gap: Towards a Remote Sensing Vision-Language Model via Cross-modal Alignment

Pretrained models

The weights of both aligned and patched models can be accessed using the following links:

• Patched
• Aligned

Citation

If you use this work please cite:

@misc{zavras2024mindmodalitygapremote,
      title={Mind the Modality Gap: Towards a Remote Sensing Vision-Language Model via Cross-modal Alignment}, 
      author={Angelos Zavras and Dimitrios Michail and Begüm Demir and Ioannis Papoutsis},
      year={2024},
      eprint={2402.09816},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2402.09816}, 
}