Netflix / Netflix/vmaf

Flushing doesn't compute metrics for last frame when frames are submitted out of order

Open
#910 2 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

Dominant language
C
Stars
5.5k
Forks
833
Avg merge
13h 8m
Merged PRs (30d)
2

Description

If I submit frames out of order, e.g. the last few frames are

3970
3974
3972
3973

When I try to compute a pooled vmaf_neg score for the entire sequence (after flushing), the function fails because frame 3974 does not have metrics computed for it. Looking at the output JSON, the last frame looks like this:

    {
      "frameNum": 3974,
      "metrics": {
        "integer_motion": 0.790710,
        "integer_adm2_egl_1": 0.974544,
        "integer_adm_scale0_egl_1": 0.966092,
        "integer_adm_scale1_egl_1": 0.960806,
        "integer_adm_scale2_egl_1": 0.970672,
        "integer_adm_scale3_egl_1": 0.983781,
        "integer_vif_scale0_egl_1": 0.707988,
        "integer_vif_scale1_egl_1": 0.951758,
        "integer_vif_scale2_egl_1": 0.970953,
        "integer_vif_scale3_egl_1": 0.978867
      }
    }

Notably it's missing the integer_motion2 metric.

My understanding is that after https://github.com/Netflix/vmaf/pull/726 this usage of the API should be valid.

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 reproducing the out-of-order sequence 3970, 3974, 3972, 3973 and flushing before computing the pooled vmaf_neg score. Inspect the flush path and the API behavior described after PR #726; done means the final frame includes integer_motion2 and pooled scoring succeeds without failing.

Written by the indexing model from the issue text.

Assessment

Tech stack
c
Domain
computer-vision
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
45/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.