search

GuoShi28 / CBDNet

Code for "Toward Convolutional Blind Denoising of Real Photographs", CVPR 2019
Apache License 2.0
501 stars 92 forks source link

A question about the loss #7

Closed XSLXANDY closed 6 years ago

XSLXANDY commented 6 years ago

Hello, I noticed you used L2 norm instead of F norm in your paper. image image In the loss function, it seems the variable x and y are matrices because I haven't seen any sum operation. Notably, with regard to vectors, L2 norm is equal to F norm. But for matrix, L2 norm is totally different from F norm. I want to confirm whether you used L2 norm of the matrix because papers often use F norm rather than L2 norm in image denoising. Can you help me? Thanks a lot!

GuoShi28 commented 6 years ago

You are right. I follow the L2 norm experiment setting of other papers.

XSLXANDY commented 6 years ago

Thank you very much for your consideration and time. But I'm still wondering why you choose the trace norm rather than F norm. Can you list your reference papers about the L2 norm?

GuoShi28 commented 6 years ago

I am not sure I understand your question in the right way. preview works, such as DnCNN, FFDNet are both using L2 norm. I did not do the experiment with F norm in denoising, so I can not say L2 norm is a better choice. There a research discussed the impact of different loss functions in image restoration (Paper). You can also refer to this paper and see whether its helpful to your understanding.

XSLXANDY commented 6 years ago

I know what you mean. I appreciate your assistance very much!