This study developed a web-based system that automatically identifies straight neck posture from video and evaluated its effectiveness. Straight neck posture may be caused by prolonged use of smartphones and computers and can lead to health problems such as shoulder stiffness and migraines. Skeletal detection technique such as OpenPose was used to detect skeletal key points, and the angle of the head was calculated from the coordinates of the nose and neck using trigonometric functions. Five videos of normal posture, five videos of straight neck posture, and labeled reference images published by a clinic were analyzed. Based on the results, approximately 50 degrees was selected as the classification threshold(Figure 1). The system automatically performs skeletal detection, angle calculation, posture classification, and result display after a video is uploaded. In an experiment using 15 videos, the system agreed with visual assessment in 13 cases, achieving an agreement rate of 86.7%. Further refinement of the threshold and expansion of the dataset are required.
