Spectral Partitioning Residual Network with Spatial Attention Mechanism for Hyperspectral Image Classification
This repository is the implementation of our paper: Spectral Partitioning Residual Network with Spatial Attention Mechanism for Hyperspectral Image Classification.
If you find this work helpful, please cite our paper:
@ARTICLE{9454961,
author={Zhang, Xiangrong and Shang, Shouwang and Tang, Xu and Feng, Jie and Jiao, Licheng},
journal={IEEE Transactions on Geoscience and Remote Sensing},
title={Spectral Partitioning Residual Network With Spatial Attention Mechanism for Hyperspectral Image Classification},
year={2021},
volume={}, number={},
pages={1-14},
doi={10.1109/TGRS.2021.3074196}}
Only Python3 is supported. We recommend you to create a Python virtual environment and then run the following command to install dependencies.
pip install -r requirement.txt
CUDA and cuDNN are optional
You can download hyperspectral image datasets at https://www.ehu.eus/ccwintco/index.php?title=Hyperspectral_Remote_Sensing_Scenes, and move the files to ./Datasets
folder
To train a model, simply run main.py
, for example:
python main.py --dataset PaviaU --model HPDM-SPRN --runs 10 --patch_size 7 --percentage 0.01 --data_aug
To get colored results, run eval.py
. The colored results can be found in the results
folder. For example:
python eval.py --dataset PaviaU --model HPDM-SPRN --patch_size 7 --weights (saved model path)
- DFFN (HPDM-DFFN)
- DHCNet (HPDM-DHCNet)
- BASSNet (HPDM-BASSNet)
- SSRN (HPDM-SSRN)
- FDSSC (HPDM-FDSSC)
- SPRN (HPDM-SPRN)
Part of our codes references to the project DeepHyperX.