Attention Based Video Summaries of Live Online Zoom Classes
Project Overview
The document discusses the innovative Help-Me-Watch system, which leverages generative AI to enhance educational experiences by creating personalized video summaries of online classes based on students' attention levels during live Zoom sessions. This technology aims to address the challenges of distractions that students face, ensuring they can revisit key lecture points they may have missed. Additionally, the system provides valuable aggregated feedback to professors regarding class engagement, enabling them to refine their teaching strategies for better learning outcomes. By prioritizing student privacy and adhering to GDPR guidelines, the Help-Me-Watch system exemplifies a forward-thinking application of AI in education, ultimately fostering an environment that supports both student retention of information and instructor effectiveness. The findings indicate that such AI-driven tools can significantly improve the overall educational experience, making learning more accessible and tailored to individual needs.
Key Applications
Help-Me-Watch system
Context: University-level online classes, targeting students attending live Zoom sessions
Implementation: Students install the Help-Me-Watch app, which tracks their attention during live classes. After class, personalized video summaries are generated based on their attention data.
Outcomes: Students receive personalized video summaries that help in revision and understanding of missed content. Professors gain insights into class engagement levels.
Challenges: Challenges include ensuring accuracy in attention measurement, addressing potential distractions, and managing the fragmented nature of summaries.
Implementation Barriers
Technical
Accurate measurement of student attention levels using facial analysis may be affected by environmental factors and individual differences.
Proposed Solutions: Using robust algorithms for attention measurement and considering individual baseline attention profiles to improve accuracy.
Pedagogical
The transition from traditional to online learning has not fully considered pedagogical effectiveness in online environments.
Proposed Solutions: Developing new teaching methods and materials suited for online formats, alongside the use of tools like Help-Me-Watch.
Privacy
Concerns regarding student monitoring and data privacy.
Proposed Solutions: Ensuring GDPR compliance and protecting student data through anonymization and secure data practices.
Project Team
Hyowon Lee
Researcher
Mingming Liu
Researcher
Hamza Riaz
Researcher
Navaneethan Rajasekaren
Researcher
Michael Scriney
Researcher
Alan F. Smeaton
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Hyowon Lee, Mingming Liu, Hamza Riaz, Navaneethan Rajasekaren, Michael Scriney, Alan F. Smeaton
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai