GTMFuse: Group-attention transformer-driven multiscale dense feature-enhanced network for infrared and visible image fusion

⭐ This code has been completely released ⭐

⭐ our article ⭐

If our code is helpful to you, please cite:

@article{mei2024gtmfuse,
  title={GTMFuse: Group-attention transformer-driven multiscale dense feature-enhanced network for infrared and visible image fusion},
  author={Mei, Liye and Hu, Xinglong and Ye, Zhaoyi and Tang, Linfeng and Wang, Ying and Li, Di and Liu, Yan and Hao, Xin and Lei, Cheng and Xu, Chuan and others},
  journal={Knowledge-Based Systems},
  volume={293},
  pages={111658},
  year={2024},
  publisher={Elsevier}
}

To Train

python train.py

To Test

Downloading the pre-trained checkpoint from best_model.pth and putting it in ./checkpoints.
python test.py

HBUT dataset

Downloading the HBUT dataset from HBUT

overall network

Results

MSRS Dataset

- Four representative images of the MSRS test set. In alphabetical order they are infrared image, visible image, GTF, FusionGAN, SDNet, RFN–Nest, U2Fusion, LRRNet, SwinFusion, CDDFuse, DATFuse, and GTMFuse.

RoadScene Datasset

- Four representative images of the RoadScene test set.

TNO Datasset