shayu296 / hsi_baseline

A Simple and Efficient Reconstruction Backbone for Snapshot Compressive Imaging

A Simple and Efficient Reconstruction Backbone for Snapshot Compressive Imaging

This repository contains the test code and pre-trained models for the arXiv paper by Jiamian Wang, Yulun Zhang, Xin Yuan, Yun Fu and Zhiqiang Tao.

Data

The benchmark simulation test dataset can be accessed here. Make sure to specify the test_path before evaluation. The coded apture can be accessed here. Make sure to specify the mask_path before evaluation.

Three versions of proposed method

Checkpoint of model v1: w/o rescaling pairs can be accessed here Similarly, pre-trained models of the other two versions are also been released. Makre sure to specify the model_path as desired pre-trained model.

Citation

If you find new baseline useful, please cite the following paper

@article{wang2021new,
  title={A new backbone for hyperspectral image reconstruction},
  author={Wang, Jiamian and Zhang, Yulun and Yuan, Xin and Fu, Yun and Tao, Zhiqiang},
  journal={arXiv preprint arXiv:2108.07739},
  year={2021}
}

For any questions, please submit an issue or contact [email protected].