Abstract
This study examined the implementation of a gesture-controlled presentation system in STEM education and its association with student participation and classroom interaction. The system enables hands-free control of presentation content through computer vision and a supervised machine learning model for gesture classification. The study involved 84 undergraduate students (experimental group, n = 42; control group, n = 42) across four STEM subjects: computer science, biology, chemistry, and physics. In the experimental group, gesture-based presentation control was used during lessons and student presentations, whereas the control group used conventional presentation methods based on standard input devices. Data were collected through structured classroom observations, and differences between groups were analyzed using the chi-square ( χ ²) test. The results showed significantly higher levels of student participation in the experimental group, whereas no statistically significant changes were observed in the control group ( p < 0.05). Classroom observations also indicated more continuous interaction with presentation materials and smoother instructional flow during presentation activities. The gesture recognition system achieved a classification accuracy of 0.93 with an average response latency of 94 ms, supporting reliable real-time classroom use. Unlike previous studies that primarily evaluated gesture recognition from a technical perspective, this study examined its implementation in authentic STEM classrooms involving both teacher-led instruction and student presentations. The findings suggest that gesture-controlled presentation systems may serve as an additional instructional tool to support interactive teaching practices and student participation in STEM education.