Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Questions about the PSNR result of the project #318

Open
BeamReflection opened this issue Nov 27, 2024 · 2 comments
Open

Questions about the PSNR result of the project #318

BeamReflection opened this issue Nov 27, 2024 · 2 comments

Comments

@BeamReflection
Copy link

I want to know how can I get the result like the picture in the project. I notice that some of them can exceed 40dB, but i can just only get a 30dB result. Of course it will increase the bpp, but that's fine.
Just increase/decrease the value of LAMBDA is OK? It seems that the bpp is out of control. Can I arbitrarily choose the bpp?

@YodaEmbedding
Copy link
Contributor

YodaEmbedding commented Nov 27, 2024

Depends on the image contents and image resolution.

A single point on the RD curve represents a single model at a particular quality. The point is generated by taking the average over 24 images of size 768x512 from the Kodak dataset.

@BeamReflection
Copy link
Author

Depends on the image contents and image resolution.

A single point on the RD curve represents a single model at a particular quality. The point is generated by taking the average over 24 images of size 768x512 from the Kodak dataset.

Thanks for your answer.
I have another question. For a certain point on the RD curve, my bpp value is much larger at the same PSNR value. I hardly changed the parameters of the model, just changed lambda, and I used the CLIC training set.
Maybe I didn't have enough epochs? I selected 13,000 images from CLIC and trained for 500 epochs.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants