microsoft / microsoft/DCVC

Cannot replicate the reported results in the paper

Open
#167 0 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
Python
Stars
829
Forks
138
Avg merge
1d 2h
Merged PRs (30d)
1

Description

Hi,

Could you share the bitrate and Y, U, V psnr values that are supposed to be obtained when running the inference with HTL, HTS and LD models? Though following all of the instructions, compared to VTM RA configuration with are observing between 80% to 120% BD-rate loss.

One additional observation is that, when using the HTL model, the luma and chroma PSNR exhibit a periodic pattern: one out of every eight frames has significantly higher quality, while the following seven frames show PSNR values that are roughly 5 dB lower. Is this behaviour expected?

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by rerunning the documented inference for the HTL, HTS, and LD models and compare bitrate plus Y, U, and V PSNR against the VTM RA configuration. Check whether the HTL output consistently shows the reported one-in-eight-frame quality pattern. Done means the expected results are documented or the cause of the BD-rate and periodic-quality discrepancy is identified.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
machine-learning
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Active
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.