Google-Tencent Industry Workshop: Advances in Open Visual Coding and AI (video)
Talks:
AV2 standardization, applications and practical encoder progress, Jianle Chen, Urvang Joshi, Google
This talk presents a brief overview of AV2, the latest AOMeida video coding standard. It will cover its goals, key technical features to achieve coding gain, and its new high level functionalities which will enable diverse use cases. The talk will also present the ongoing AOM work for the practical AV2 encoder development, including the goals, milestones and the current progress.
Video Coding in the Era of AI, Leo Zhao, Tencent
Neural networks are reshaping image and video compression, from enhancing individual coding tools to enabling fully learned compression frameworks. This tutorial provides an overview of three major paradigms: neural-network-enhanced conventional codecs, autoencoder-based compression, and implicit neural representations (INR), highlighting representative techniques and recent advances in each category. We also discuss their key trade-offs in coding efficiency, computational complexity, and practical deployment, and provide perspectives on the future evolution of neural image and video coding.
Beyond Conventional Video Coding: Coding for Machines and Volumetric Visual Media, Shan Liu, Tencent
Conventional video coding has primarily focused on efficiently representing visual content for human consumption. With the rapid development of AI and emerging immersive applications, new coding paradigms are needed to support machine intelligence and volumetric visual media. This presentation provides an overview of recent advances in Video Coding for Machines (VCM) and Volumetric Visual Media (VVM), highlighting their key concepts, technical challenges, and ongoing standardization activities. We will also discuss how these emerging technologies may extend video coding beyond traditional 2D human-oriented applications.
