Real-Time Sitting Posture Detection for Students in Classroom Environment

Discomfort in the lower back, neck, and almost everywhere in the body can easily be noted among several people, particularly among students who spend long hours seated in classrooms. Extended periods of sitting in fixed postures can lead to strain and musculoskeletal issues, affecting focus and overall well-being. Real-time posture detection is necessary to prevent long-term health problems caused by poor posture and lifestyles, such as extended periods of sitting and low levels of physical activity. This paper proposes a system for real-time detection of students' sitting postures using the MediaPipe framework, specifically employing the angle heuristics approach. MediaPipe BlazePose is able to detect up to 33 keypoints, compared to the detection of fewer keypoints with most standard approaches. The system can recognize common postures like forward leaning, backward leaning, slouching, or crossing arms/legs.