Awesome
MambaHSI: Spatial-Spectral Mamba for Hyperspectral Image Classification
<div align="center">Yapeng Li, Yong Luo, Lefei Zhang, Zengmao Wang*, Bo Du*
</div> <div align="center"> <a href='https://ieeexplore.ieee.org/abstract/document/10604894'><img src='https://img.shields.io/badge/TGRS-Paper-blue'></a> </div>š Introduction
<div align="center"> <img src="./images/Motivation.png" alt="Motivation" width="55%"> </div>- To our best knowledge, the MambaHSI is the first image-level hyperspectral image classification model based on SSM, which can simultaneously model long-range interaction of whole image and integrate spatial and spectral image information.
- MambaHSI demonstrates the great potential of Mamba to be the next-generation backbone for hyperspectral image models.
š Getting Started
Installation
conda create -n MambaHSI_env python=3.9
conda activate MambaHSI_env
conda install pytorch==1.13.1 torchvision==0.14.1 torchaudio==0.13.1 pytorch-cuda=11.7 -c pytorch -c nvidia
pip install packaging==24.0
pip install triton==2.2.0
pip install mamba-ssm==1.2.0
pip install spectral
pip install scikit-learn==1.4.1.post1
pip install calflops
Data Preparation
The dataset can download Google Drive and BaiduNetdisk.
data
āāā UP/
āāā PaviaU.mat
āāā PaviaU_gt.mat
...
āāā Houston/
āāā Houston.mat
āāā Houston_GT.mat
...
āāā HanChuan/
āāā WHU_Hi_HanChuan.mat
āāā WHU_Hi_HanChuan_gt.mat
...
āāā HongHu/
āāā WHU_Hi_HongHu.npy
āāā WHU_Hi_HongHu_gt.npy
Training:
python train_MambaHSI.py --dataset_index 0
python train_MambaHSI.py --dataset_index 1
python train_MambaHSI.py --dataset_index 2
python train_MambaHSI.py --dataset_index 3
šļø Main Results
<details open> <summary><font size="4"> Pavia University Results </font></summary> <img src="./images/PaviaU_results.png" alt="PaviaU" width="100%"> </details> <details open> <summary><font size="4"> Houston Results </font></summary> <img src="./images/Houston_results.png" alt="Houston" width="100%"> </details> <details open> <summary><font size="4"> HanChuan Results </font></summary> <img src="./images/HanChuan_results.png" alt="HanChuan" width="100%"> </details> <details open> <summary><font size="4"> HongHu Results </font></summary> <img src="./images/HongHu_results.png" alt="HongHu" width="100%"> </details>Citation
If you find this project helpful for your research, please kindly consider citing our paper and give this repo āļø:
@ARTICLE{MambaHSI_TGRS24,
author={Li, Yapeng and Luo, Yong and Zhang, Lefei and Wang, Zengmao and Du, Bo},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={MambaHSI: Spatial-Spectral Mamba for Hyperspectral Image Classification},
year={2024},
volume={},
number={},
pages={1-16},
keywords={Hyperspectral Image Classification;Mamba;State Space Models;Transformer},
doi={10.1109/TGRS.2024.3430985}}
Acknowledgement
Part of our MambaHSI framework is referred to CVSSN and SSFCN. We thank all the contributors for open-sourcing.