US20260099976
2026-04-09
Physics
G06T13/40
A novel system enhances video representation in network-based meetings by replacing low-quality video feeds with animated avatars. The system assesses video quality based on head pose, facial feature visibility, and image clarity. When a video feed fails to meet these thresholds, the system generates an avatar using a pre-captured image of the participant. This avatar is animated with realistic facial expressions and lip movements, synchronized with the participant's speech, ensuring continuous engagement in the meeting interface.
The application is situated within the realm of video processing for online meetings, focusing on improving visual representation of participants. It describes methods for dynamically evaluating video quality, particularly for in-room attendees, and generating photorealistic avatars to replace subpar video feeds. This approach facilitates high-quality visual engagement between in-room and remote participants by intelligently substituting live video with animated avatars.
Network-based meetings have become essential for business communication, allowing collaboration among geographically dispersed participants. These meetings typically involve a mix of in-room and remote attendees, with video conferencing systems capturing live video feeds of in-room participants. However, these systems often use a single camera, leading to challenges such as poor angles, partial occlusion, and inadequate lighting, which can hinder visual clarity and engagement for remote attendees.
The described techniques enhance video quality and participant representation in network-based meetings through dynamic video processing and avatar generation. The system evaluates in-room video feeds in real-time, replacing poor quality feeds with photorealistic avatars. This process involves capturing live video, segmenting participants, assessing video quality, and animating avatars with synchronized lip movements and expressions. The system ensures consistent visual representation, addressing challenges like poor camera angles and lighting conditions, thus improving hybrid meeting efficacy.
Conventional meeting systems face challenges like camera placement limitations, varying participant distances, and lighting conditions, affecting video quality. The described system addresses these by employing real-time video analysis to evaluate video feeds against quality thresholds. When a feed is suboptimal, it generates a photorealistic avatar using pre-enrolled images, adapting to current participant appearance. The system seamlessly transitions between live video and avatars, maintaining visual engagement and effectively handling multiple participants.