Increased focus on personalized learning has highlighted the need for real-time monitoring of student engagement. Understanding attention levels during instruction helps improve teaching effectiveness and learning outcomes. However, existing methods rely on manual observation or periodic assessments, which are subjective and lack consistency. These approaches fail to capture moment-to-moment variations in engagement. Conventional systems using basic video tracking or facial detection lack robustness in variable lighting, head pose changes, and classroom dynamics. They are also limited in providing timely, actionable insights. This study presents FocusTrack, a real-time engagement monitoring system that utilizes facial cues and behavioral indicators for accurate classification. The system processes video frames locally and provides continuous engagement feedback. Two annotated datasets—EngageFace (150 hours, classroom-based) and StudyFocus (90 hours, home-based)—were developed to capture diverse learning scenarios. Each dataset includes labels for gaze direction, drowsiness, and facial cues. Experimental results show accuracy levels of 97.0% and 95.5% across the two datasets, outperforming conventional models. The system also maintains latency under 60 ms on CPU- based setups. FocusTrack offers a scalable, privacy-aware solution for continuous engagement monitoring in real-world educational environments. It provides instructors with objective feedback to adapt teaching strategies dynamically.