search

google / mipnerf

Apache License 2.0
894 stars 109 forks source link

average error matrics in paper #35

Open dlutzzw opened 2 years ago

dlutzzw commented 2 years ago

hi,thanks for your exciting work.I have two question about avg matrics.

first，I'm confused about the meaning of avg matric，will there be more advantages？

second，I find that when you compute MSE from PSNR，the implementation of the code in is different from that in the paper. so I'm a bit confunsed. Can you help me ?

jonbarron commented 2 years ago

The average metric is only useful if you want to look at a single number instead of 3 numbers. I found it useful when hyperparameter tuning, but you can definitely just ignore it if you are happy to look at PSNR/SSIM/LPIPS separately.

I think the paper has a bug in the math for PSNR, how embarrassing! There was also a bug in the code for computing DSSIM that I fixed in our follow-up release, where it should be sqrt((1-ssim)/2). This is our latest code release if you'd like something with more unit tests: https://github.com/google-research/multinerf/blob/main/internal/image.py

dlutzzw commented 1 year ago

thank you for your reply, I will be careful of these points.

At 2022-09-13 01:47:21, "jonbarron" @.***> wrote:

The average metric is only useful if you want to look at a single number instead of 3 numbers. I found it useful when hyperparameter tuning, but you can definitely just ignore it if you are happy to look at PSNR/SSIM/LPIPS separately.

I think the paper has a bug in the math for PSNR, how embarrassing! There was also a bug in the code for computing DSSIM that I fixed in our follow-up release, where it should be sqrt((1-ssim)/2). This is our latest code release if you'd like something with more unit tests: https://github.com/google-research/multinerf/blob/main/internal/image.py

— Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you authored the thread.Message ID: @.***>