ABSTRACT
Intervention skills of experts in various fields are often developed through their own experiences. From this viewpoint, the purpose of this paper is to develop an automatic detection technique that can objectify the subjectivity of experts and extract the necessary data on the spot, so that experts can reflect on their own teaching and use it to explain their interventions. In this paper, we focus on the features of gait and posture essential for healthy walking, which are said to influence physical and mental fitness. To understand the state of gait, we used a posture detection tool "OpenPose" to extract as features "Grounding rate" and "Whether gait is uniaxial or biaxial," which can be used as common measures among the viewpoints considered important by the experts. In order to extract these two viewpoints as features, we developed an evaluation method using automatic detection. As a result of comparing each method and evaluation method, we have confirmed the effectiveness of the posture detection technology for objectifying the skills to assess gait and posture of the expert and the usefulness of the analysis program we have created for visualizing the traits of interventions by the expert. We have also obtained the valuable findings that can lead to improve the intervention programs for encouraging healthy walking useful in the various practical fields.
