As of its publication in June 2022, the ARAD 1K natural spectral image data set is the largest collection of natural hyperspectral images published to date.
This repository will serve to provide access to the data set, facilitate evaluation of follow up work, as well as host support code for the data set.
This data set contains 950 publicly images and 50 “test” images which have only been released as RGB or Multi-Spectral-Filter-Array “RAW” files.
Publicly available images have been published as:
Sample images form the ARAD 1K dataset
Downloads are currently available via
(registration required)
If you use this data set, please cite:
@InProceedings{Arad_2022_CVPR_recovery,
author = {Arad, Boaz and Timofte, Radu and Yahel, Rony and Morag, Nimrod and Bernat, Amir and Cai, Yuanhao and Lin, Jing and Lin, Zudi and Wang, Haoqian and Zhang, Yulun and Pfister, Hanspeter and Van Gool, Luc and Liu, Shuai and Li, Yongqiang and Feng, Chaoyu and Lei, Lei and Li, Jiaojiao and Du, Songcheng and Wu, Chaoxiong and Leng, Yihong and Song, Rui and Zhang, Mingwei and Song, Chongxing and Zhao, Shuyi and Lang, Zhiqiang and Wei, Wei and Zhang, Lei and Dian, Renwei and Shan, Tianci and Guo, Anjing and Feng, Chengguo and Liu, Jinyang and Agarla, Mirko and Bianco, Simone and Buzzelli, Marco and Celona, Luigi and Schettini, Raimondo and He, Jiang and Xiao, Yi and Xiao, Jiajun and Yuan, Qiangqiang and Li, Jie and Zhang, Liangpei and Kwon, Taesung and Ryu, Dohoon and Bae, Hyokyoung and Yang, Hao-Hsiang and Chang, Hua-En and Huang, Zhi-Kai and Chen, Wei-Ting and Kuo, Sy-Yen and Chen, Junyu and Li, Haiwei and Liu, Song and Sabarinathan and Uma, K and Bama, B Sathya and Roomi, S. Mohamed Mansoor},
title = {NTIRE 2022 Spectral Recovery Challenge and Data Set},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2022},
pages = {863-881}
}
for the 31 channel hyperspectral images and/or
@InProceedings{Arad_2022_CVPR_demosaic,
author = {Arad, Boaz and Timofte, Radu and Yahel, Rony and Morag, Nimrod and Bernat, Amir and Wu, Yaqi and Wu, Xun and Fan, Zhihao and Xia, Chenjie and Zhang, Feng and Liu, Shuai and Li, Yongqiang and Feng, Chaoyu and Lei, Lei and Zhang, Mingwei and Feng, Kai and Zhang, Xun and Yao, Jiaxin and Zhao, Yongqiang and Ma, Suina and He, Fan and Dong, Yangyang and Yu, Shufang and Qiu, Difa and Liu, Jinhui and Bi, Mengzhao and Song, Beibei and Sun, WenFang and Zheng, Jiesi and Zhao, Bowen and Cao, Yanpeng and Yang, Jiangxin and Cao, Yanlong and Kong, Xiangyu and Yu, Jingbo and Xue, Yuanyang and Xie, Zheng},
title = {NTIRE 2022 Spectral Demosaicing Challenge and Data Set},
booktitle = {Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
month = {June},
year = {2022},
pages = {882-896}
}
for the 16 channel multi-spectral images.