This AI pose-assessment smart mirror uses a single camera and edge computing to extract human skeletons. It provides two branches: gait analysis quantifies steps, stride length, speed and body sway for rehab tracking and fall-risk interpretation; fitness assessment quantifies angles, angular velocity, repetitions and reach distance to track muscular endurance, flexibility, balance and cardiopulmonary endurance, generating sarcopenia alerts and personalized exercise suggestions

This technology overcomes the limits of visual inspection, single indicators and poor quantitative tracking by introducing task-oriented AI routing. The gait branch uses YOLO-Pose to analyze lower-limb stability, while the fitness branch uses YOLO-Pose for movement quality and AlphaPose + MediaPipe for fine hand/foot localization and reach measurement. It detects subtle decline and compensatory movements, with edge computing for real-time privacy protection

This technology can be deployed by scenario, with field validation shown in Fig. 3. The gait branch supports hospital rehab and long-term care, helping stroke, Parkinson's and lower-limb degeneration cases track rehab outcomes and interpret fall risk. The fitness branch suits community centers, senior fitness centers and homes, supporting sarcopenia alerts, movement-quality tracking, personalized advice and telecare, reducing manpower and tracking costs and improving access.