Development of Human Pose Recognition System by Using Raspberry Pi and PoseNet Model
Kosei Yamao, Ryosuke Kubota · 2021
New sports tools, which can be used anywhere, are needed for sports, recreation and/or health. In this paper, we propose a human pose recognition system developed on a single-board computer. The proposed system is constructed by PoseNet model, which is a neural networks-based pose estimation method, implemented by a Raspberry Pi. The proposed system detects 17 key-points of a human body from a camera. Then a set of key-points is matched to that of key-points registered beforehand. The validity and the effectiveness of the proposed system have been verified by some experiments.