LeSA: Applying a Computer Vision- based AI Approach to Monitoring Sustained Attention in Online Learning

LeSA: Applying a Computer Vision-based AI Approach To Monitoring Sustained Attention In Online Learning doi: 10.52987/edc.2025.004 Pichaya Promla King Mongkut’s University of Technology North Bangkok, Department of Computer Education Bangkok, Thailand pichaya.promla@gmail.com Somkid Saelee King Mongkut’s University of Technology North Bangkok, Department of Computer Education Bangkok, Thailand somkid.l@fte.kmutnb.ac.th ABSTRACT As online learning continues to expand, its flexibility in time and location supports individual learning differences and lifelong learning, contributing to sustainable social and economic development. However, online learners require greater self-discipline and responsibility due to limited interaction and reduced instructor oversight, making it difficult to monitor engagement and adapt teaching strategies. One of the most significant challenges in this environment is sustaining student attention, which is essential for effective learning. A lack of sustained attention can lead to decreased motivation and higher dropout rates, impacting overall learning outcomes. This study explores the use of computer vision-based AI to analyze behavioral patterns that indicate sustained attention in online learning. By tracking key behavioral indicators, including gaze direction, facial expressions, blinking patterns, and eye closure frequency, this approach provides valuable insights into student engagement. The Learner Sustained Attention (LeSA) model was developed and evaluated using four machine learning algorithms to determine the most effective method for identifying attention patterns. Findings indicate that integrating computer vision-based AI into online education provides a scalable and data-driven solution for real-time attention monitoring. The Random Forest model demonstrated the highest accuracy (84.07%), effectively identifying behavioral patterns associated with sustained attention. These insights can support instructors in adapting teaching methods, enhancing student engagement, and reducing dropout rates. This study contributes to the ongoing innovation in educational technology, reinforcing the role of AI-driven behavioral tracking in improving learning experiences and instructional effectiveness in online education KEYWORDS: Online Learning, Computer Vision, Sustained Attention, Artificial Intelligence, Educational Innovation, Behavioral Patterns, Engagement Tracking, Learning Analytics