Spectral Query Spatial: Revisiting the Role of Center Pixel in Transformer for Hyperspectral Image Classiffcation

Ning Chen, Leyuan Fang, Yang Xia, Shaobo Xia, Hui Liu, and Jun Yue

---

The code in this toolbox implements the "Spectral Query Spatial: Revisiting the Role of Center Pixel in Transformer for Hyperspectral Image Classiffcation".

More specifically, it is detailed as follow.

Citation

Please kindly cite the papers if this code is useful and helpful for your research.

N. Chen, L. Fang, Y. Xia, S. Xia, H. Liu and J. Yue, "Spectral Query Spatial: Revisiting the Role of Center Pixel in Transformer for Hyperspectral Image Classification," in IEEE Transactions on Geoscience and Remote Sensing, vol. 62, pp. 1-14, 2024, Art no. 5402714, doi: 10.1109/TGRS.2024.3361652.}

@ARTICLE{10419080,
  author={Chen, Ning and Fang, Leyuan and Xia, Yang and Xia, Shaobo and Liu, Hui and Yue, Jun},
  journal={IEEE Transactions on Geoscience and Remote Sensing}, 
  title={Spectral Query Spatial: Revisiting the Role of Center Pixel in Transformer for Hyperspectral Image Classification}, 
  year={2024},
  volume={62},
  number={},
  pages={1-14},
  keywords={Feature extraction;Hyperspectral imaging;Interference;Data mining;Current transformers;Adaptation models;Tokenization;Deep neural network;hyperspectral image (HIS) classification},
  doi={10.1109/TGRS.2024.3361652}}

How to use it?

1. Prepare HSI datasets, such as IP, PU, WH, etc. Before use, it is necessary to perform train/test splitting. The data format can refer to the provided reference data at data/Indian/Indian_10_split.mat. Store the splitted data in the corresponding folders under the data directory. For example, store the IP dataset in data/Indian. The file names of the mat files should follow the format SIGN_NUM_split.mat, where SIGN represents the data identifier and NUM represents the number of training samples for each class.

2. Modify the parameters in src/params_use/sqsformer.json according to the data file path, mainly for the data_sign and data_file parameters. For example: data/Indian/Indian_10_split.mat => data/[data_sign]/[data_file]_split.mat.

3. Run the code.

   python workflow.py
   

Others

If you want to run the code in your own data, you can accordingly change the input (e.g., data, labels) and tune the parameters in params_use.

If you encounter the bugs while using this code, please do not hesitate to contact us.

Licensing

Copyright (C) 2024 Ning Chen

This program is free software: you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, version 3 of the License.

This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program.