A lightweight spectral-spatial convolution module for hyperspectral image classification

PyTorch implementation of A lightweight spectral-spatial convolution module for hyperspectral image classification.

Basic Usage

import torch
from LSSCM import LSSCM_ResNet
# Take the Indian Pines dataset as an example, the number of classes and spectral channels are 16 and 200, respectively.
model = LSSCM_ResNet(num_classes=16, channels=200)
model.eval()
print(model)
input = torch.randn(100, 200, 11, 11)
y = model(input)
print(y.size())

Paper

A lightweight spectral-spatial convolution module for hyperspectral image classification

Please cite our paper if you find it useful for your research.

@article{meng2021lightweight,
  title={A lightweight spectral-spatial convolution module for hyperspectral image classification},
  author={Meng, Zhe and Jiao, Licheng and Liang, Miaomiao and Zhao, Feng},
  journal={IEEE Geoscience and Remote Sensing Letters},
  volume={19},
  pages={1--5},
  year={2022},
  publisher={IEEE}
}

Reference Paper

• GhostNet: More features from cheap operations https://arxiv.org/abs/1911.11907

Reference Code

• ghostnet https://github.com/iamhankai/ghostnet.pytorch