[Paper Review] Real-time Attention Span Tracking in Online Education

Rahul RK, Shantha Kumar S+2 more|arXiv (Cornell University)|Nov 29, 2021

Mind wandering and attention6 references4 citations

TL;DR

This paper proposes a real-time system that tracks student attention spans during online classes using camera and microphone inputs to analyze five non-verbal features—head pose, eye gaze, facial expressions, head movement, and speech patterns. By combining computer vision and machine learning, the system computes an attention score and delivers immediate feedback to students and educators, enabling performance and teaching quality assessment through heuristic attention metrics.

ABSTRACT

Over the last decade, e-learning has revolutionized how students learn by providing them access to quality education whenever and wherever they want. However, students often get distracted because of various reasons, which affect the learning capacity to a great extent. Many researchers have been trying to improve the quality of online education, but we need a holistic approach to address this issue. This paper intends to provide a mechanism that uses the camera feed and microphone input to monitor the real-time attention level of students during online classes. We explore various image processing techniques and machine learning algorithms throughout this study. We propose a system that uses five distinct non-verbal features to calculate the attention score of the student during computer based tasks and generate real-time feedback for both students and the organization. We can use the generated feedback as a heuristic value to analyze the overall performance of students as well as the teaching standards of the lecturers.

Motivation & Objective

• To address the growing challenge of student distraction in online education environments.
• To develop a real-time, non-intrusive method for monitoring student attention during virtual classes.
• To identify and quantify non-verbal behavioral cues that correlate with attention levels.
• To provide actionable feedback to students and educators based on real-time attention metrics.
• To establish a heuristic framework for evaluating student engagement and teaching effectiveness.

Proposed method

• The system captures real-time video and audio streams from students during online classes.
• It extracts five non-verbal features: head pose, eye gaze, facial expressions, head movement, and speech patterns using computer vision and audio processing techniques.
• A machine learning model fuses these features to compute a dynamic attention score in real time.
• The attention score is updated continuously and used to generate immediate feedback for students and instructors.
• The system leverages image processing and deep learning models for accurate detection of gaze and facial activity.
• Feedback is used as a heuristic to assess individual student engagement and overall teaching quality.

Experimental results

Research questions

• RQ1 How can non-verbal behavioral cues be reliably extracted from video and audio streams during online learning?
• RQ2 Which combination of non-verbal features most accurately reflects real-time student attention?
• RQ3 Can a real-time attention score be computed with sufficient accuracy to inform pedagogical feedback?
• RQ4 How can attention metrics be used to evaluate both student engagement and teaching effectiveness?
• RQ5 What is the feasibility of deploying such a system in real-world online education settings?

Key findings

• The system successfully computes a real-time attention score using five non-verbal features from video and audio inputs.
• The attention score provides actionable feedback for both students and educators during live online sessions.
• The approach enables heuristic-based assessment of student engagement and teaching quality through attention metrics.
• The system demonstrates feasibility for real-time monitoring in online education environments.
• The integration of visual and audio cues improves the robustness of attention estimation compared to single-modality approaches.
• The method supports scalable deployment for institutional use in e-learning platforms.