-
At EdgeConf the other week it was pointed out that when people encode JPEG images they often pick a fairly arbitrary quality level. Generally one based on habit than one that makes the given image loo…
-
https://github.com/hustvl/4DGaussians/assets/40021525/53cf365f-f143-40e0-910b-8f4446ecc4e7
-
The codebase for VMAF is quite large, but it would be nice to have a pure Rust implementation. This is by no means high priority, just if someone (probably me) gets bored.
-
This is the time to calculate psnr and ssim when I inference the model, but when I calculate the metrics afte…
-
Hi guys, is there any one who have tested encoder with kodak test?
I have used the example binary file to encode the pictures, and compute the PSNR with respect to different bpp.
, SSIM, LPIPS, IS metrics for quantitative evaluation and where exactly did you implement in the code for quantitative evaluation?
-
in losses.py,
self.perceptual_loss_model = VGGPerceptualLoss(device)
self.alpha1 = 1.00
self.alpha2 = 0.06
self.alpha3 = 0.05
self.alpha4 = 0.5
…
-
**Is your feature request related to a problem? Please describe.**
Current testing pipeline is different for each metric and sometimes doesn't cover all potential scenarios.
**Describe the solutio…
-
I don't fully understand which methods checkpoints refer to? Are they Enh-POELIC, HFLIC or neither? If they don't, can you share them? I would like to compare with them on CLIC 2022 test set.
-
Hi, I have some more questions here:
1. Do the pretrained models contains CCN-light?
2. What's the meaning of the number - 4, 8, 14 - in the model name? Like ```4_caffemodel```, ```8_caffemodel```, …