FusionMamba

• Code for the paper: "FusionMamba: Efficient Remote Sensing Image Fusion with State Space Model", TGRS, 2024.

• First application of the state space model (SSM) in the hyper-spectral pansharpening and hyper-spectral image super-resolution (HISR) tasks.

• State-of-the-art (SOTA) performance in pansharpening, hyper-spectral pansharpening, and HISR tasks.

Paper

• For a detailed understanding of our method, please refer to the paper: FusionMamba: Efficient Remote Sensing Image Fusion with State Space Model.
• This paper has been published in the IEEE Transactions on Geoscience and Remote Sensing (TGRS).

Method

FusionMamba Block

We expand the single-input Mamba block to accommodate dual inputs, creating the FusionMamba block, which can serve as a plug-and-play solution for information integration.

Experimental Results

Get Started

Dataset

• Datasets for pansharpening: PanCollection. We recommend downloading datasets in the h5py format. The testing toolbox can be found here.

• Datasets for hyper-spectral pansharpening: HyperPanCollection. We recommend downloading datasets in the h5py format.

• Dataset for HISR: the CAVE dataset. You can find this dataset on the Internet.

• Results for FusionMamba and a serious of methods on PanCollection: https://pan.baidu.com/s/1ARGLyvGKn57-eCl041Gk3g, key: 6271.

Installation

1. Clone the repository:

git clone https://github.com/PSRben/FusionMamba.git

1. Install the Mamba implementation by following the instructions in the Mamba-block directory.

2. Install other packages:

pip install einops h5py opencv-python torchinfo scipy numpy

Usage

• This repository is only for the pansharpening task.

• The model weights trained on the WV3 dataset for 400 epochs can be found in the weights directory.

# train
python train.py --train_data_path ./path_to_data/train_WV3.h5 --val_data_path ./path_to_data/valid_WV3.h5
# test
python test.py --file_path ./path_to_data/name.h5 --save_dir ./path_to_dir --weight ./weights/epochs.pth

Citation

@ARTICLE{10750233,
  author={Peng, Siran and Zhu, Xiangyu and Deng, Haoyu and Deng, Liang-Jian and Lei, Zhen},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={FusionMamba: Efficient Remote Sensing Image Fusion With State Space Model}, 
  year={2024},
  volume={62},
  number={},
  pages={1-16},
  doi={10.1109/TGRS.2024.3496073}}

Contact

We are glad to hear from you. If you have any questions, please feel free to contact siran_peng@163.com.