Official Code for: Jinyuan Liu, Guanyao Wu, Zhu Liu, Long Ma, Risheng Liu, Xin Fan*, "Where Elegance Meets Precision: Towards a Compact, Automatic, and Flexible Framework for Multi-modality Image Fusion and Applications", in Proceedings of the 33rd International Joint Conference on Artificial Intelligence (IJCAI), 2024.

We strongly recommend that you use Conda as a package manager.

# create virtual environment
conda create -n CAF python=3.8
conda activate CAF
# select and install pytorch version yourself (Necessary & Important)
# install requirements package
pip install -r requirements.txt

This code natively supports the same naming for ir/vi image pairs. An naming example can be found in ./data folder.

# Test: use given example and save fused color images to ./result.
# If you want to test the custom data, please modify the data path in 'eval.py'.
# The default is to use Det_final to pre-train the model, If you want to change the model, please modify the model path in 'eval.py'.
python eval.py

If this work has been helpful to you, we would appreciate it if you could cite our paper!

@inproceedings{CAF,
  title={Where Elegance Meets Precision: Towards a Compact, Automatic, and Flexible Framework for Multi-modality Image Fusion and Applications},
  author={Liu, Jinyuan and Wu, Guanyao and Liu, Zhu and Ma, Long and Liu, Risheng and Fan, Xin},
  booktitle={Proceedings of the 33rd International Joint Conference on Artificial Intelligence},
  year={2024}
}