This repo can reproduce the main results in Tabel 1. and Tabel 2. of our paper. All the source code and pre-trained models will be released to the public for further research.

We implement the core operators shift_3d and shift_3d_back in HSI reconstruction by cuda for faster speed. Hope it is helpful for the community.

• Python 3 (Recommend to use Anaconda)

• PyTorch >= 1.10

• NVIDIA GPU + CUDA

• Python packages:

  pip install -r requirements.txt

• Build cuda operators

  python setup.py develop

Download the dataset from https://github.com/mengziyi64/TSA-Net, put the dataset into the corresponding folder 'code/datasets/', and recollect them in the following form:

|--datasets
    |--cave_1024_28
        |--scene1.mat
        |--scene2.mat
        ：  
        |--scene205.mat
    |--CAVE_512_28
        |--scene1.mat
        |--scene2.mat
        ：  
        |--scene30.mat
    |--KAIST_CVPR2021  
        |--1.mat
        |--2.mat
        ： 
        |--30.mat
    |--TSA_simu_data  
        |--mask.mat   
        |--Truth
            |--scene01.mat
            |--scene02.mat
            ： 
            |--scene10.mat
    |--TSA_real_data  
        |--mask.mat   
        |--Measurements
            |--scene1.mat
            |--scene2.mat
            ： 
            |--scene5.mat

If you want to train any other model with RND-SCI, please refer to the following format:

# RND_[model]
python simu_train.py --method rnd_[original model name] --outf ./exp/simu_rnd_[original model name] / --seed 42 --gpu_id 0

for example:

# RND-SAUNet
python simu_train.py --method rnd_saunet_1stg --outf ./exp/simu_rnd_saunet_1stg/ --seed 42 --gpu_id 0 

# RND_MST
python simu_train.py --method rnd_mst --outf ./exp/simu_rnd_mst/ --seed 42 --gpu_id 0 

If you have original pre-trained model without/ with RND-SCI framework and want to train with RND-SCI, please refer to the following format:

python simu_train.py --method rnd_[original model name] --outf ./exp/simu_rnd_[original model name] / --pretrained_model_path  [your model with/without RND-SCI path] --seed 42 --gpu_id 0 

Please use checkpointing (--cp) when running out of memory. Refer to 'utils/simu_utils/simu_args.py' to use more options.

a). Test our models on the HSI dataset. The results will be saved in 'code/test/' in the MatFile format. For example, we test the RND-SAUNet:

python simu_test.py --method rnd_saunet_1stg --outf ./test/simu_rnd_saunet_1stg   --pretrained_model_path [your model with/without RND-SCI path]

b). Calculate quality assessment. We use the same quality assessment code as DGSMP. So please use Matlab, get in 'code/analysis_tools/Quality_Metrics/', and then run 'Cal_quality_assessment.m'.

c). If you want test the other models with RND-SCI , please change the model your want to test in above step a).

d). If you want to plug and play a model with RND, whether the pre-trained model uses RND-SCI or not, you can directly use the command that refers to a).

If you want to train any other model with RND-SCI, please refer to the following format:

# RND_[model]
python real_train.py --method rnd_[original model name] --outf ./exp/real_rnd_[original model name] / --seed 42 --gpu_id 0 --isTrain

for example:

# RND-SAUNet
python real_train.py --method rnd_saunet_1stg --outf ./exp/real_rnd_saunet_1stg/ --seed 42 --gpu_id 0 --isTrain

# RND_MST
python real_train.py --method rnd_mst --outf ./exp/real_rnd_mst/ --seed 42 --gpu_id 0 --isTrain

a). Test our models on the HSI dataset. The results will be saved in 'code/evaluation/testing_result/' in the MatFile format. For example, we test the RND-SAUNet:

python real_test.py --method rnd_saunet_1stg --outf ./test/real_rnd_saunet_1stg  --pretrained_model_path [your model with/without RND-SCI path]

b). Calculate quality assessment. We use no reference image quality assessments (Naturalness Image Quality Evaluator, NIQE ). So please use Matlab, get in 'code/analysis_tools/Quality_Metrics/', and then run 'NIQE_metric.m'.

c). If you want test SAUNet-1stg or the others , please change the model your want to test in above step a).

d). If you want to plug and play a model with RND, whether the pre-trained model uses RND-SCI or not, you can directly use the command that refers to a).

If we want to get inference fps of RND-SAUNet, run the following commond:

python test_fps.py --method rnd_saunet_1stg --outf ./test/real_rnd_saunet_1stg --gpu_id 0

Please mask sure that the GPU is not occupied by another program before running the commond. Other models are similar to this.

Afer you finish the training of model, please run these commands:

cd analysis_tools/
python tranining_time [your training log path]

You can get the Params and FLOPS of models at the begin of training. Or use following commonds (for instance, we get these values of RND_SAUNet. Other methods are similar):

python test_fps.py --method rnd_saunet_1stg

This project is based on MST. Thanks for the wonderful works.

RND-SCI is released under the MIT License.

If you find our paper and code useful in your research, please consider giving a star ⭐ and citation 📝 :)

@article{RNDSCI,
  title={A Range-Null Space Decomposition Approach for Fast and Flexible Spectral Compressive Imaging},
  author={Wang, Junyu and Wang, Shijie and Zhang, Ruijie and Zheng, Zengqiang and Liu, Wenyu and Wang, Xinggang},
  journal={arXiv preprint arXiv:2305.09746},
  year={2023}
}