Real-Time Video Quality Prediction on Live, High-Motion Studio/Professional-Grade Content

2023 Cycle
distorted photo of kids on 6th street in austin, tx

PI

A.C. Bovik, Department of Electrical and Computer Engineering

SUMMARY
A diagram showing a video encoding process for broadcast, internet, cable, satellite, 4G/5G, and WiFi distribution. The image begins with a video of a person wearing clown makeup, labeled ‘RAW VIDEO,’ passing through ‘Select Encode Parameters’ and ‘Quality Control (VMAF)’ before becoming ‘H264/HEVC VIDEO.’ It leads to a TV screen, a tablet, and a smartphone displaying the same video. The diagram includes the text ‘2021 Technology and Engineering Emmy Award’ with the National Academy of Television Arts & Sciences logo.

Predicting perceptual video quality is a hard problem that has been successfully addressed in many scenarios, such as quality control of streaming and sharing of videos. However, videos continue to “get bigger” along every dimension including frame rate, bit depth, color gamut, resolution, and fusion with generative methods. This work addresses new quality issues that arise and how to deal with them to optimize perceptual video quality vs bandwidth consumption.