Rehabilitation Training Evaluation and Correction System Based on BlazePose

Abstract

We apply computer vision to rehabilitation training in medicine. According to the survey, many diseases require rehabilitation training after treatment. However, due to reasons such as a lack of relevant knowledge of patients and large differences in patients’ physical condition, it is difficult for patients to perform rehabilitation training correctly. Improper rehabilitation training prolongs the recovery time and causes secondary injury or even leaves sequelae in serious cases. Rehabilitation training at this stage relies on the doctor’s instruction, ignoring the problem that the patient’s movements are deformed and unguided during the rehabilitation training process. The proposed system corrects the user’s movements during the rehabilitation training process, monitors the user’s heart rate in real-time, and suggests a way of improvement and training. It also provides users with professional and personalized rehabilitation training assistance.