Shu Kong

icon indicating a website link aimerykong.github.io icon indicating an email [email protected]

icon company Robotics Institute @ CMU icon location Pittsburgh, PA, US icon location seeing the open-world, learning forever

Results 65 comments of Shu Kong

conv2 input

Hi, Thanks for your interest. I didn't provide the list of image pairs because the txt file that stores the pairs is >200MB, meaning millions pairs are sampled for testing...

Will you provide the code and model for JPEG compression artifact reduction?

Thanks for your interest! The released model is for image deblurring. It is not for image denoising (i.e., removing JPEG artifacts). So you may not want to fine-tune it. I'd...

Trained discriminator not working after netD.eval() or shuffle = False

Happy to help diagnose the issue. Can you send along two curves of AUROC at epochs: one with netD.eval() and the other without? The shuffle thing should not be crucial....

Trained discriminator not working after netD.eval() or shuffle = False

Thanks for the curves! I recall that I observed the same in my experiments, that's why I didn't use .eval(). I conjecture that this is because the discriminator is trained...

Trained discriminator not working after netD.eval() or shuffle = False

Thanks for the discussion! This is a fair point and I agree that there can be information leakage in the testing implementation. That made me think about setting batch size...