HybridSN: Exploring 3-D–2-D CNN Feature Hierarchy for Hyperspectral Image Classification

---

Introduction

• This is a keras implementation of the IEEE GRSL paper 'HybridSN: Exploring 3-D–2-D CNN Feature Hierarchy for Hyperspectral Image Classification'.
  

• See https://github.com/gokriznastic/HybridSN for the original implementation in the jupyter notebook form.
  

• I have adapted the code into the python script form, making it easier for running directly.
  

Environment & Requirements

• CentOS Linux release 7.2.1511 (Core)
  
• python 3.6.5
  
• keras 2.3.1
  
• spectral
  

Usage

Download hyperspectral data.

Download the Inidan Pines, the Salinas and the Pavia University and put them into ./dataset:

Train&Test the Inidan Pines:

python hybridsn.py -GPU 0 

You can change the namespace and conduct experiements under your preferred setting.

References

https://github.com/gokriznastic/HybridSN