SMLNet - GitLite

KO | EN

gitlite — search

#security #nextjs #typescript #generative-ai #ios #linux #web #shell #proxy #nodejs

SMLNet

(★ 15)

SMLNet: A SPD Manifold Learning Network for Infrared and Visible Image Fusion (IJCV 2025)

File Explorer

ZIP Download

- metrics-deeplabv3.py
- metrics-yolo.py
- readme.txt
- - SMLNet.model
- - README.md
- __init__.py
- kernel.py
- optimizer.py
- spd.py
- utils.py
args_fusion.py
fusion_strategy.py
net.py
README.md
test.py
test_robust.py
train_autoencoder.py
utils.py
utils_robust.py

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

Command Glossary

Commands referenced in this DOCs, explained below.

🔍

`python`

View Details ▼

Python language interpreter.

python

Start a REPL (interactive shell):

python {{path/to/file.py}}

Execute a specific Python file:

python -i {{path/to/file.py}}

Execute a specific Python file and start a REPL:

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

🧩🚀SMLNet: A SPD Manifold Learning Network for Infrared and Visible Image Fusion
====
Accetped by 
[IJCV 2025]
🔗"(https://doi.org/10.1007/s11263-025-02578-1)"

🔍Highlights
---
● **First SPD Manifold Learning for Fusion**  
Our work is the first to introduce Riemannian manifold networks (SPD manifolds) into image fusion tasks, enabling geometrically consistent modeling of cross-modal correlations.

● **Manifold-Aware Attention**  
We propose a novel SPD Attention Module (SPDAM) that dynamically weights cross-modal features on the manifold space, enhancing complementary information fusion while suppressing redundancies.

● **Superior Performance & Efficiency**  
Extensive experiments show SMLNet outperforms state-of-the-art methods in fusion quality (e.g., EN, VIF) and computational efficiency, with proven gains in downstream tasks like object detection.

🖥️Environment
----
python==3.12.7

pytorch==2.5.1

pytorch-cuda==12.4

scipy==1.13.1

numpy==1.26.4

pillow==10.4.0

tqdm==4.66.5

⚙️Training
----
```bash
python train_autoencoder.py
```

✔️Testing
----
For inputs with standard image dimensions, please use the following command for testing:
```bash
python test.py
```

To enhance the practical applicability and overall robustness of the framework, particularly when handling images of non-standard or irregular dimensions, we have introduced a new testing script:
```bash
python test_robust.py
```
This script is highly optimized for robustness. The updated testing pipeline incorporates a more flexible data processing mechanism to ensure stable inference across diverse real-world scenarios. It effectively accommodates varying aspect ratios and irregular resolutions while maintaining consistent manifold-based feature fusion.

📖Citation
----
If you are interested in our work, please cite it in the following format:
```bash
@article{kang2025smlnet,
  title={SMLNet: A SPD Manifold Learning Network for Infrared and Visible Image Fusion},
  author={Kang, Huan and Li, Hui and Xu, Tianyang and Wu, Xiao-Jun and Wang, Rui and Cheng, Chunyang and Kittler, Josef},
  journal={International Journal of Computer Vision},
  volume={133},
  number={12},
  pages={8642--8663},
  year={2025},
  publisher={Springer}
}
```

⬇️Model Download
----
The vgg16 model can be found in https://pan.baidu.com/s/14YYYrDZ1RM3yqFbYNnbQbw, and the password is: usd6