EduGaze: Gaze Based Attention Tracking System

This paper introduces EduGaze, a web-based system designed to monitor and analyze student engagement in real time during online classes. The platform is built using a scalable monorepo architecture that combines a Node.js/Express backend, a React-based dashboard for teachers, and a Chrome extension for collecting data from students. It uses Socket.io to enable a low-latency, two-way communication stream, allowing the system to continuously track attention states such as focused, distracted, and offline. These interactions are synchronized and stored in an optimized database, supporting both real-time visualization and later analysis. The system also includes an AI module powered by a locally hosted large language model, which generates automated session reports, identifies attention patterns, and suggests ways to improve student focus. Experimental results indicate that EduGaze can effectively detect engagement trends and provide meaningful insights to educators. By eliminating the need for specialized hardware and cloud-based AI services, the system offers a cost-effective, privacy-friendly, and scalable solution for enhancing online learning experiences.