Awesome
FusionDN<br>
This is the code of the following paper (tensorflow):<br>
@inproceedings{xu2020aaai,
title={FusionDN: A Unified Densely Connected Network for Image Fusion},
author={Xu, Han and Ma, Jiayi and Le, Zhuliang and Jiang, Junjun and Guo, Xiaojie},
booktitle={Proceedings of the Thirty-Fourth AAAI Conference on Artificial Intelligence (AAAI)},
pages={12484--12491},
year={2020}
}
It is a unified model for multiple image fusion tasks, including:<br>
- visible and infrared image fusion<br>
- multi-exposure image fusion<br>
- multi-focus image fusion<br>
Framework:<br>
Overall procedure:<br>
<div align=center><img src="https://github.com/hanna-xu/FusionDN/blob/master/imgs/procedure.jpg" width="440" height="290"/></div><br>Intuitive description of data flow and the process of EWC:<br>
<div align=center><img src="https://github.com/hanna-xu/FusionDN/blob/master/imgs/MultiTask.jpg" width="510" height="200"/></div><br>Fused results:<br>
<div align=center><img src="https://github.com/hanna-xu/FusionDN/blob/master/imgs/res1.jpg" width="900" height="490"/></div> <div align=center><img src="https://github.com/hanna-xu/FusionDN/blob/master/imgs/res2.jpg" width="900" height="400"/></div>To train:<br>
CUDA_VISIBLE_DEVICES=0,1 python main.py (2 GPUs are needed)<br><br>
To test:<br>
CUDA_VISIBLE_DEVICES=0 python test_main.py<br><br>
Tips:<br>
Large files should be downloaded separately, including the following files: <br>
For training:<br>
- Training dataset<br> The code to create your own training dataset can be found here.
- vgg16 (put it in 'VGGnet/')<br>
For testing:<br>
- Trained model<br>
If you have any question, please email to me (xu_han@whu.edu.cn).