PoseNet Based Acupoint Recognition of Blind Massage Robot

摘要

In allusion to the current common acupoint recognition is based on the edge extraction of neural network, which is affected by too many factors of human, there are certain interferences and errors in finding acupoints. This paper put forward a novel method combining posture tracking algorithm with proportional bone measurement, which can fully fit the skeleton of human body to provide a new idea for acupoint recognition of massage robot, and improve its accuracy and efficiency. The above method is simulated in Python to realize the function of massage robot to find acupoints automatically. The result shows that it takes less time to find more accurate and more quantities acupoints.