Google Industry Seminar: Video Processing in the New Age of AI
Video processing and compression are being re-envisioned in the age of AI. Traditional video codecs, which rely on rigid, pre-defined rules, are being augmented and, in some cases, replaced by AI-driven approaches. These new methods leverage machine learning to intelligently analyze video content, allowing for more adaptive and efficient compression. We are going to discuss AOM’s new codec AV2, its low- and high-level features that enable significantly smaller file sizes with no perceptible loss in quality, a crucial development for streaming and storage. The shift to AI has also transformed how we evaluate video quality. Traditional metrics, while directionally useful, don’t always align with human perception, especially for user generated content (UGC). They fail to capture what’s most important for machine vision tasks. We will talk about new AI-based quality metrics that are being developed. They correlate better with a human’s subjective experience and a machine’s ability to perform tasks like object recognition. Along the way, we’ll cover large scale industrial infrastructure challenges and the ways to achieve high reliability and accuracy.
Organizer
Balu Adsumilli (YouTube)
Speakers
Title: AV2 Overview
Speaker: Jianle Chen (Chrome)
Abstract:
This talk overviews AV2, the next-gen AOM video coding standard. It will cover its goals, key technical features, and potential impact. It would highlight AV2’s aim to surpass its predecessor, AV1, in compression efficiency and performance, enabling better video quality at lower bitrates. The talk will delve into specific coding tools employed by AV2, the testing methodologies used to evaluate them. The talk might also discuss the near future work/project of AOM Codec Working Group, especially the AI extension of AV2.
Bio:Dr. Jianle Chen received his B.S. and Ph.D. degrees in EE from Zhejiang University, Hangzhou, China, in 2001 and 2006, respectively. He is a software engineer in the Open Video team at Google since 2021, working next generation video codec research and development efforts. He was formerly with Samsung Electronics Company Ltd., Qualcomm Technologies, Inc., San Diego, CA, USA, focusing on the research of video technologies. Since 2006, he has been actively involved in the development of various video coding standards, including the VVC and HEVC standards, and their extensions and in the Joint Video Experts Team (JVET).
Title: AV2 complexity and high level features
Speaker: Ville-Mikko Rautio (YouTube)
Abstract:
The talk explores the AV2 video codec, highlighting its reasonable complexity in both SW/HW while providing much better compression efficiency over its predecessor, AV1. Furthermore, it introduces new high level functionalities which will enable diverse use cases.
Bio:Ville-Mikko Rautio is Engineering Manager at YouTube and Google in charge of Google Video Coding Unit (VCU), including research and development. He has been involved in the research and development of hardware codecs at Google since 2010. Prior to his tenure at Google he worked on video and ASIC startup ecosystem in Finland. Ville-Mikko holds a M.Sc. in Computer Science from Oulu University (1998).
Title: Video Quality models
Speaker: Yilin Wang (YouTube)
Abstract:
This talk overviews video quality research at Google/YouTube. We’ll introduce how industry applications drive our video quality model development, highlighting our existing accomplishments and ongoing explorations. We’ll also share insights on leveraging large language models (LLMs) for video quality assessment, discussing potential benefits and remaining challenges.
Bio:Dr. Yilin Wang is a staff software engineer in the Media Algorithms team at YouTube/Google. He spent the last ten years on improving YouTube video processing and transcoding infrastructures, and building video quality metrics. Beside the video engineering work, he is also an active researcher in video quality related areas, and published papers in CVPR, ICCV, TIP, ICIP, etc. He received his PhD from the University of North Carolina at Chapel Hill in 2014, working on topics in computer vision and image processing.