The Performance of Quality Metrics in Assessing Error-Concealed Video Quality
Bibliographic record
Abstract
In highly-interactive video streaming applications such as video conferencing, tele-presence, or tele-operation, retransmission is typically not used, due to the tight deadline of the application. In such cases, the lost or erroneous data must be concealed. While various error concealment techniques exist, there is no defined rule to compare their perceived quality. In this paper, the performance of 16 existing image and video quality metrics (PSNR, SSIM, VQM, etc.) evaluating errorconcealed video quality is studied. The encoded video is subjected to packet loss and the loss is concealed using various error concealment techniques. We show that the subjective quality of the video cannot be necessarily predicted from the visual quality of the error-concealed frame alone. We then apply the metrics to the error-concealed images/videos and evaluate their success in predicting the scores reported by human subjects. The errorconcealed videos are judged by image quality metrics applied on the lossy frame, or by video quality metrics applied on the video clip containing that lossy frame; this way, the impact of error propagation is also considered by the objective metrics. The measurement and comparison of the results show that, mostly though not always, measuring the objective quality of the video is a better way to judge the error concealment performance. Moreover, our experiments show that when the objective quality metrics are used for the assessment of the performance of an error concealment technique, they do not behave as they would for general quality assessment. In fact, some newly developed metrics show the correct decision only about 60% of the time, leading to an unacceptable error rate of as much as 40%. Our analysis shows which specific quality metrics are relatively more suitable for error-concealed videos.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.007 | 0.031 |
| Meta-epidemiology (narrow) | 0.002 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.000 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".