Real-Time Webcam-Based Eye Tracking for AI-Assisted Reading Analysis in Web Environments

Abstract

This study investigates real-time, webcam-based eye tracking as a low-cost and scalable alternative for AI-assisted reading analysis in web environments, aiming to address the limitations of hardware-dependent eye tracking systems highlighted in existing literature. The proposed framework captures gaze data through standard consumer webcams and integrates seamlessly into a browser-based architecture to monitor user attention during reading sessions. Gaze coordinates are dynamically mapped to webpage elements, enabling paragraph-level analysis of reading behavior. The methodology incorporates dwell time computation and kernel density estimation (KDE) to mitigate noise and enhance fixation stability inherent in webcam-based gaze tracking. Reading sessions are managed independently, allowing systematic collection and comparison of user interaction data. Experimental evaluations demonstrate that the system can effectively differentiate focused and distracted reading regions without requiring external sensors or intrusive equipment. The results indicate that the proposed approach provides meaningful insights into user attention distribution and reading patterns while maintaining accessibility and ease of deployment. By eliminating the dependency on specialized hardware, the system enables large-scale data collection across diverse user groups and real-world settings. Furthermore, the framework establishes a foundation for adaptive and personalized reading experiences through AI-driven content optimization. The findings support the applicability of the system in educational technologies, usability evaluation, adaptive content delivery, and human–computer interaction research, suggesting that browser-based eye tracking can serve as a practical and efficient tool for large-scale reading behavior analysis.