Machine-Learned Participant Tiles for Inclusive Meeting Displays
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Cloud-based meeting systems often suffer from inefficiencies in resource utilization and user engagement due to misidentification of speakers, poor camera placement, and lack of inclusivity for remote participants, leading to distractions and reduced attention during meetings.
Innovation Solution
Implementing machine learned models to generate and arrange tiles representing meeting participants, determining active speakers, and optimizing tile display to enhance engagement and inclusiveness by analyzing participant behavior and sensor data in real-time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If multiple camera locations are used to capture meeting participants, then coverage of all participants is improved, but device complexity and cost increase
Solution Approach 1:
The patent uses virtual avatars generated from 3D models to represent remote participants in the meeting space. Instead of placing physical cameras at every participant location, the system creates digital copies (avatars) that replicate participants' appearances and behaviors, eliminating the need for multiple physical camera installations while maintaining complete participant visibility.
Solution Approach 2:
The system introduces a central processing system that receives video feeds from a single camera array and generates virtual representations of all participants. This intermediary processing layer consolidates multiple camera functions into a single physical camera system while producing the equivalent visual information that would require multiple cameras distributed throughout the meeting space.
2Reliability
If all participants are displayed simultaneously in meeting spaces, then inclusivity is improved, but attention and engagement of participants decrease
Solution Approach 1:
The patent segments the meeting display into multiple virtual tiles, each representing a different participant or group of participants. This segmentation allows the system to selectively highlight and expand individual participants when they are speaking or need attention, while maintaining smaller tiles for other participants in the background, thus balancing inclusivity with individual attention.
Solution Approach 2:
The system dynamically adjusts the size, position, and visibility of participant tiles based on real-time factors such as who is speaking, engagement levels, and meeting context. Active speakers receive larger tile representations and enhanced visual attention, while inactive participants are displayed in reduced detail, creating a dynamic display that adapts to meeting needs and maintains participant engagement.
3Loss of information
If video links transmit all meeting content to all participants, then information completeness is improved, but network bandwidth requirements and computational resources increase
Solution Approach 1:
The system extracts and transmits only the essential visual information needed for each participant's representation rather than transmitting complete video feeds to all participants. By processing and selecting only the necessary visual data to create accurate virtual avatars and tiles, the system reduces network bandwidth requirements and computational resources while maintaining information completeness for the meeting.
Data Source
AI summary
Techniques for a service provider network to manage a meeting between individuals located in a conference room with one or more other individuals remote from the conference room are discussed herein. An Enterprise system can implement one or more machine learned models to generate tiles that represent different individuals associated with a meeting. A same or different machine learned model can automatically arrange a series of tiles for display that, when presented collectively, promotes inclusiveness and attention for both in-room and remote participants of the meeting. The meeting management techniques can include determining which content to include on a display device based on evaluating changes in behavior of the meeting participants over time.