Development of posture evaluation system using 2D camera

In this study, we propose a system and analytical method of quantitatively evaluating posture during movement using a two-dimensional (2D) camera. To verify the effectiveness of the system and analytical method, young subjects performed a specific movement in front of the 2D camera in a normal state and a state that imitated the posture of the elderly, and each was filmed. The video was analyzed using the 2D human pose estimation algorithm OpenPose, and age-related differences in movements were detected. When we compared young people and the elderly based on the total whole-body change, they could not be classified. However, when comparing young people and elderly based on the total ankle change, they could be classified. This result suggests that focusing on the ankle is more appropriate than focusing on the entire body when evaluating the movement and posture of the elderly using a 2D camera. Because the system proposed here can perform measurements using a 2D camera, which is highly popular, its application is expected to systems for analyzing the walking posture of elderly people at home.