Shi Guo

(1) The blur effect and not working well on some high noisy image. The denoising results may be over smoothing in some cases. These drawbacks may because that the realistic...

Hi, I actually met this problem a long time ago. I can not clearly remember how to solve this problem. But it caused by the failure of the installation of...

You can try different versions of Matlab. I'm not sure. But I suggest you take care of the system path problem. Since the code needs to find the installed Matconvnet....

https://github.com/GuoShi28/CBDNet/issues/24#issuecomment-509852896 这个基本是我代码的实现，你可以参考下。

Hi, It seems like to be a common environment problem. You can refer to [https://github.com/GuoShi28/CBDNet/issues/3](https://github.com/GuoShi28/CBDNet/issues/3).

hi, I test using matconvnet using 1080Ti.

Actually, you can directly compare. CBDNet is much slower for the following reasons:(1) deeper network, UNet structure. (2) ffdnet uses downsampling first to further save running time. Suggestions: I don't...

Yes, I think sqrt(temp_x) is correct and I did not notice this error when coding. I also suggest you may also need to adjust the range of sigma_s and sigma_c...

In the paper, for synthetic noise level, we have two noise level, \sigma_s and \sigma_c. For the clean raw image L, the image level is \sigma_c + \sigma_s*L.

@oneTaken Sorry for the late reply. It's actually different. For python code, I refer to other paper and complete the isp process in CBDNet, including color transformation, white balance and...