PHDMamba: Progressive Hybrid Mamba for Hyperspectral Image Classification

Yichu Xu¹, Chengxi Han¹, Shi Chen¹, Yao Jin¹, Yuchun Miao¹, Haonan Guo¹, Di Wang<sup>1 †

1</sup> Wuhan University, China, ^†Corresponding author

📖Overview

• PHDMamba is a Progressive Hybrid Mamba Module, which performs stage-wise modeling of long-range dependencies along both spectral and spatial dimensions. In addition, a dedicated Spectral-Spatial Interaction Module is introduced to adaptively integrate contextual spectral and spatial features.

🚀Let's Get Started!

A. Installation

Step 1: Clone the repository:

Clone this repository and navigate to the project directory:

git clone https://github.com/YichuXu/PHDMamba.git
cd PHDMamba

Step 2: Environment Setup:

It is recommended to set up a conda environment and installing dependencies via pip. Use the following commands to set up your environment:

Create and activate a new conda environment

conda create -n PHDMamba
conda activate PHDMamba

Install dependencies

Our method uses python 3.8, pytorch 1.13, other environments are in requirements.txt

Please use the following mamba version: mamba_ssm==1.2.0 !

pip install -r requirements.txt
cd kernels/selective_scan && pip install .

B. Data Preparation

Download HSI classification dataset from Google Drive or Baidu Drive (百度网盘) and put it under the [dataset] folder. It will have the following structure:

${DATASET_ROOT}   # Dataset root directory
├── datasets
│   │
│   ├── pu        # Pavia University data
│   │   ├──PaviaU.mat
│   │   ├──PaviaU_gt.mat
│   │
│   ├── houston13  # Houston 2013 data
│   │   ├──GRSS2013.mat
│   │   ├──GRSS2013_gt.mat 
│   │     
│   ├── whuhc     # Whu-HanChuan data
│   │   ├──WHU_Hi_HanChuan.mat
│   │   ├──WHU_Hi_HanChuan_gt.mat 
│   │
│   ├── other HSI Datasets   
│   │   ├ ... 
│   │    

C. Performance Evaluation

• The following commands show how to generate training samples:

python GenSample.py --train_num 15 --dataID 1
python GenSample.py --train_num 15 --dataID 3
python GenSample.py --train_num 30 --dataID 6

• The following commands show how to train and evaluate PHDMamba for HSI classification:

CUDA_VISIBLE_DEVICES=0 python  main.py --model PHDMamba --dataID 1 --epoch 200 --lr 1e-3 --decay 0 --split False --dataset_name pu --train_num 15
CUDA_VISIBLE_DEVICES=0 python  main.py --model PHDMamba --dataID 3 --epoch 200 --lr 1e-3 --decay 0 --split False --dataset_name houston13 --train_num 15
CUDA_VISIBLE_DEVICES=0 python  main.py --model PHDMamba --dataID 6 --epoch 200 --lr 1e-3 --decay 0 --split False --dataset_name whuhc --train_num 30

🤠Results Taken Away

• The visualization results are provided in the results folder.

• We'd appreciate it if you could give this repo a ⭐️star⭐️ and stay tuned.

📜Reference

if you find it useful for your research, please consider giving this repo a ⭐ and citing our paper! We appreciate your support！😊

@ARTICLE{Xu2025PHDMamba,
  author={Xu, Yichu and Han, Chengxi and Chen, Shi and Jin, Yao and Miao, Yuchun and Guo, Haonan and Wang, Di},
  journal={IEEE Geoscience and Remote Sensing Letters}, 
  title={PHDMamba: Progressive Hybrid Mamba for Hyperspectral Image Classification}, 
  year={2025},
  volume={22},
  number={},
  pages={1-5}
  }

🙋Q & A

For any questions, please contact us.