In posture analysis group, we conduct research on posture analysis using the open-source library OpenPose, which is a technology for skeletal detection. OpenPose is capable of detecting skeletal structures in real-time from images and videos. It extracts each point within the skeleton from the image and outputs their coordinates. In addition to real-time video analysis, it can also analyze multiple individuals simultaneously.
The following studies related to posture analysis have been conducted so far:
– For individuals who are inexperienced in darts or wish to improve their scores, we analyze the position and height of the arm to investigate the optimal form for throwing darts. (Fig. 1)
– We have developed a web system that analyzes the coordinates of the neck and nose from images to determine whether an individual is suffering from forward head posture (straight neck). (Fig. 2)
– We have conducted research to quantify sign language movements, enabling learners to objectively review their own movements. (Fig.3)
These studies demonstrate that posture analysis using OpenPose can be applied in various fields such as sports and education, and further developments are anticipated in the future.
  Fig.1 Posture analysis for darts |   Fig. 2 Calculation between nose and neck for “straight neck” |   Fig. 3 Posture analysis for sign language |
