NeVES: Real-Time Neural Video Enhancement for HTTP Adaptive Streaming

NeVES: Real-Time Neural Video Enhancement for HTTP Adaptive Streaming

IEEE VCIP 2025

December 1 – December 4, 2025

Klagenfurt, Austria

[PDF]

Daniele Lorenzi, Farzad Tashtarian, Christian Timmerer

Abstract: Enhancing low-quality video content is a task that has raised particular interest since recent developments in deep learning. Since most of the video content consumed worldwide is delivered over the Internet via HTTP Adaptive Streaming (HAS), implementing these techniques on web browsers would ease the access to visually-enhanced content on user devices.

In this paper, we present NeVES, a multimedia system capable of enhancing the quality of video content streamed through HAS in real time.

The demo is available at: https://github.com/cd-athena/NeVES.