A Study on Digital Human Body Image Processing and Training for AI-Based Acupoint Localization Guiding

Objectives : This study aims to develop an AI-based acupoint localization guiding system that enables automatic detection and visualization of acupoint locations in traditional Korean medicine.The system is designed to be applicable in clinical, educational, and self-care contexts, facilitating digital transformation in Korean medicine.Methods : Standard acupoint locations defined by WHO/WPRO were used to extract anatomical landmarks.Deep learning models including YOLOv8 (object detection), HRNet (pose estimation), and Pix2Pix (image translation) were applied.The system was integrated with mixed reality (MR) devices (HoloLens) and a mobile web interface to visually guide users in real time.Results : The proposed system successfully predicted acupoint coordinates relative to body landmarks and provided real-time visual guidance through MR and mobile platforms.It demonstrated practical utility in various scenarios such as clinical treatment support, medical education, and home-based self-care.Conclusions : This research presents a significant case of applying AI-based digital technology to acupoint theory in Korean medicine.The system lays a foundation for developing acupoint-based digital therapeutics (DTx), AI-assisted diagnostic tools, and global standardization of acupuncture point systems.