Help! I can't get dataset.

[StiphyJay]

Hello, the url of RGB91 dataset your paper provided is void.I can't open the link. Could you help me for this problem? Thank you very much! I would appreciate for your help.

[zhuyu-06]

Hello, I can't open the RGB91 dataset link, too. I wonder know if you've got it?
Thank you!

[aleksandaratanasov]

Would be lovely to have the RGB91 dataset.

[aleksandaratanasov]

I was able to find the dataset. Following paper reference [11] (https://ieeexplore.ieee.org/document/5466111) I was able to find

https://github.com/tingfengainiaini/sparseCodingSuperResolution/tree/master/Data/Training

Since I wasn't sure I kept digging and found the original conference presentation by the authors of TEN:

https://www.gputechconf.co.kr/assets/files/presentations/1-1650-1710_DL_Contest_%EC%9E%90%EC%9C%A0%EC%A3%BC%EC%A0%9C_%EB%8C%80%EC%83%81.pdf

On slide 17 a couple of samples of the Train Data (the so called RGB 91) is displayed. Comparing the images from the git repo I have posted above and the samples it looks legit. In addition the paper states "which consider the luminance channel in YCbCr color space, we only used the gray-channel in our experiments ...", which is also visible in the samples in the presentation.

So get the dataset, convert it to grayscale and you are good to go. As the paper further states they did not do any augmentation such as flipping, rotation etc.