Authors: Vignesh V Menon, Prajit T Rajendran, Amritha Premkumar, Benjamin Bross, Detlev Marpe

Published on: February 05, 2024

Impact Score: 8.15

Arxiv code: Arxiv:2402.03513

Summary

• What is new: Introduces ViSOR, a video super-resolution-based encoding scheme that optimizes for low latency and energy efficiency in online streaming.
• Why this is important: Existing per-title encoding schemes don’t adequately balance encoding time, energy efficiency, and perceptual quality in streaming applications.
• What the research proposes: ViSOR uses random forest models to predict perceptual quality and encoding time for various resolutions, adjusting to maintain quality while reducing bitrate and energy consumption.
• Results: Achieves a bitrate reduction of 24.65% and 32.70% for PSNR and VMAF quality metrics, respectively, and cuts storage consumption and encoding energy by 79.32% and 68.21%, promoting greener streaming.

Technical Details

Technological frameworks used: Random Forest, FSRCNN

Models used: Perceptual Quality Prediction, Encoding Time Prediction

Data used: Spatiotemporal features of video segments

Potential Impact

Online streaming platforms, video encoding solutions, and companies invested in reducing energy consumption for digital content delivery.

Want to implement this idea in a business?

We have generated a startup concept here: StreamWise.