S2Mamba

The official code for the paper "S2Mamba: A Spatial-spectral State Space Model for Hyperspectral Image Classification"

Get Started

Installation

pip install -i requirement.txt

Data Preparation

Download HSI datasets and put them into the ./data directory. For example:

  './data/IndianPine.mat'
  './data/Pavia.mat'
  './data/Houston.mat'
  './data/WHU-Hi-LongKou/WHU_Hi_LongKou.mat'

Data download link: https://pan.baidu.com/s/18wTlU9ERczOfUo8WVoHVnA?pwd=aak6 Password: aak6

Training and Testing

CUDA_VISIBLE_DEVICES=0 python demo_mamba.py --dataset='Indian' --epoches=400 --patches=7 --sess s2mamba --dropout 0.4 --lr 5e-4

CUDA_VISIBLE_DEVICES=0 python demo_mamba.py --dataset='Pavia' --epoches=400 --patches=11 --sess s2mamba --dropout 0.1 --lr 5e-4

CUDA_VISIBLE_DEVICES=0 python demo_mamba.py --dataset='Houston' --epoches=100 --patches=9 --sess s2mamba --dropout 0.1 --lr 1e-4

CUDA_VISIBLE_DEVICES=0 python demo_mamba.py --dataset='WHU_Hi_LongKou' --epoches=400 --patches=9 --sess s2mamba --dropout 0.4 --lr 5e-4

Acknowledgment

Our detection code is built upon SpectralFormer and Vmamba. We are very grateful to all the contributors to these codebases.

Citation

If you appreciate our work and find this repository helpful, please consider giving a citation:

@ARTICLE{s2mamba,
  author={Wang, Guanchun and Zhang, Xiangrong and Peng, Zelin and Zhang, Tianyang and Jiao, Licheng},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={S2Mamba: A Spatial-spectral State Space Model for Hyperspectral Image Classification}, 
  year={2025},
  pages={1-1},
  doi={10.1109/TGRS.2025.3530993}}