Eye Gaze Feedback for Remote Meeting Presenters
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video conferencing systems lack the ability to effectively provide personalized feedback to presenters on participant attention, as they cannot accurately determine eye gaze angles in unconstrained environments with varying head poses and illumination, limiting the presenter's ability to adapt their presentation in real-time.
Innovation Solution
The system uses consumer-level webcams to capture images, extracts facial landmarks and eye information, and employs convolutional neural networks to estimate eye gaze angles, providing feedback to presenters on regions of interest in the presentation content, allowing for real-time adaptation of the presentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If consumer-level webcams are used to capture images in unconstrained environments, then the system becomes cost-effective and easier to operate, but the accuracy of eye gaze angle determination deteriorates due to varying head poses and illumination
Solution Approach 1:
The system dynamically adjusts parameters such as illumination compensation and head pose estimation to maintain measurement accuracy despite varying environmental conditions. By changing parameters like lighting normalization and geometric correction factors, the system achieves reliable eye gaze angle determination using consumer-level webcams in unconstrained environments
Solution Approach 2:
The system introduces intermediate processing steps including head pose estimation and facial landmark detection as mediators between the raw webcam images and the final eye gaze angle calculation. These intermediary computations compensate for the limitations of consumer-level cameras by adding computational layers that correct for head orientation and lighting variations
2Productivity
If the system processes eye gaze angle information in real-time during remote meetings, then the presenter can adapt their presentation in real-time, but the computational complexity and processing time increase
Solution Approach 1:
The system performs preliminary actions by pre-processing facial landmark detection and head pose estimation before the actual eye gaze angle calculation. By preparing these intermediate results in advance during the video frame capture phase, the system reduces the computational burden during real-time gaze tracking, enabling faster response times without sacrificing accuracy
Solution Approach 2:
The system segments the complex task of eye gaze angle determination into distinct modular components: facial landmark detection, head pose estimation, and gaze angle calculation. This segmentation allows each module to be optimized independently and processed in parallel where possible, reducing overall system complexity while maintaining real-time performance
3Loss of information
If the system provides detailed feedback on participant attention, then the presenter can personalize content and improve engagement, but the amount of information to be processed and acted upon increases
Solution Approach 1:
The system implements feedback mechanisms that provide the presenter with actionable insights about participant attention. By analyzing eye gaze angle data and generating summarized feedback reports that highlight key attention patterns and regions of interest, the system delivers comprehensive information in a digestible format that guides presenter decisions without overwhelming them with raw data
Data Source
AI summary
A method of providing feedback to a presenter in a remote meeting includes capturing images of remote participants in the remote meeting using cameras associated with computing devices that display content presented by the presenter. Eye gaze angle information for at least one of the remote participants is determined based on the captured images. At least one region of interest in the displayed content is identified based on the eye gaze angle information. Feedback is provided to the presenter including an indication of the identified at least one region of interest.