/C2H-data

C2H-data

We have released our C2H-data and Synthetic-data proposed in paper 'Reconstruction of Hyperspectral Data from RGB Images with Prior Category Information'.

The 30-channel hyperspectral image data (HSI) is saved as .tiff file, we have provided its corresponding RGB data and given carefully labeled object bounding boxes.
Google Drive
BaiDu Drive password: cy5p

The 128-channel raw hyperspectral image data can be obtained from：
BaiDu Drive password: vriu

.tiff image data can be read as follow python script:

import numpy as np
from libtiff import TIFF
def tiff_to_numpy(tiff_image_name): 
    tif = TIFF.open(tiff_image_name, mode = "r")
    idx = 0
    for im in list(tif.iter_images()):
        im=np.expand_dims(im,axis=2)
        if idx==0:
            image=im
        else:
            image=np.concatenate((image,im),axis=2)       
        idx = idx + 1
    return image

If you use these datasets in your researches or works, please cite our paper as follow:

@article{yan2020reconstruction,  
  title={Reconstruction of Hyperspectral Data from RGB Images with Prior Category Information},  
  author={Longbin Yan, Xiuheng Wang, Min Zhao, Maboud Farzaneh Kaloorazi, Jie Chen and Susanto Rahardja},  
  journal={IEEE Transactions on Computational Imaging},  
  year={2020},  
  publisher={IEEE}  
}