This is the code of the paper titled as "UNIFusion: A Lightweight Unified Image Fusion Network".
- Download all these files.
- If you want to test your own images on our model, "matlab_code_for_creating_base_and_detail_layers/main.m" is ready for you to generate the base and detail layers.
All the necessary parameter settings can be found at "args_fusion.py".
We provide a series of testing files for different fusion tasks.
e.g. Infrared and visible image fusion task Run the following code:
python test_imageTNO.py
The fusion results will be presented at the "outputs" folder.
Training dataset can be found at this website: https://pjreddie.com/projects/coco-mirror/
Put the images at the "train2014" folder.
python train.py
- Python 3.7.3
- torch 1.7.1
- scipy 1.2.0
If you have any questions, please contact me at chunyang_cheng@163.com.
Most code of this implementation is based on the DenseFuse: https://github.com/hli1221/densefuse-pytorch
If this work is helpful to you, please cite it as:
@article{cheng2021unifusion,
title={UNIFusion: A Lightweight Unified Image Fusion Network},
author={Cheng, Chunyang and Wu, Xiao-Jun and Xu, Tianyang and Chen, Guoyang},
journal={IEEE Transactions on Instrumentation and Measurement},
volume={70},
pages={1--14},
year={2021},
publisher={IEEE}
}