This is a personal pytorch-based implement of HybridSN by Jupyter notebook.
The HybridSN is spectral-spatial 3D-CNN followed by spatial 2D-CNN.The 3D-CNN facilitates the joint spatial-spectral feature representation from a stack of spectral bands. The 2D-CNN on top of the 3D-CNN further learns more abstract level spatial representation.
The paper link: "HybridSN: Exploring 3-D–2-D CNN Feature Hierarchy for Hyperspectral Image Classification"
Official keras code: gokriznastic/HybridSN
pytorch 1.9.0
scikit-learn 1.0.2
spectral 0.22.2
torchinfo 1.6.1
Fig1. IndianPines (a) RGB (b) groundtruth (c) predition
Fig2. Pavia University (a) RGB (b) groundtruth (c) predition
Fig3. Salinas Scene (a) RGB (b) groundtruth (c) predition