3D Real-time Face Acupoints Recognition System Based on HoloLens 2

PubDate: January 2022

Teams: Sichuan University

Writers: Xiyu Chen; Hongyu Yang; Yulong Ji; Dongnan Chen

PDF: 3D Real-time Face Acupoints Recognition System Based on HoloLens 2

Abstract

Acupoints positioning is a significant part of Traditional Chinese Medicine (TCM) massage. However, the number of acupoints are large and their functions are complex, trainees usually need the guidance of doctors to accurately locate them. In order to improve the efficiency of acupoints location, this paper proposes a three-dimensional (3D) real-time face acupoints recognition system based on HoloLens 2. Through wearing the mixed reality device HoloLens 2, users can directly see the acupoints stacked on the human face. In this paper, we use a new face tracking technology, combined with landmark detection and acupoints positioning method in Acupoint Science to detect and track 3D facial acupoints in real time, and realize visualization on HoloLens 2. The positioning error of our system is about 2.55 mm and the frame rate is 20.59 FPS. The experimental results show that our system has good real-time and accuracy, can realize the automatic positioning of acupoints, and is conducive to the teaching and practice of TCM massage theory.