Real-time Exercise Posture and Muscle Strength Feedback System: Integrating Pose Tracking Model, ArUco Marker, and EMG Sensor

Improper posture during exercise can lead to injuries, and the use of visual feedback has been demonstrated to positively affect exercise efficacy and injury prevention. Individuals who exercise independently often follow instructional videos for guidance. To improve exercise effectiveness and minimize the risk of injury, we propose a system that provides real-time visual feedback. This system utilizes MediaPipe for movement estimation and incorporates ArUco markers to compensate for occlusions caused by exercise equipment. Furthermore, electromyography (EMG) is employed to assess whether the appropriate muscles are being activated. In instances where correct posture and muscle activation are not achieved, the system prevents further progress in the exercise repetitions and notifies the user of a failure to meet the required conditions. The experimental results demonstrated that the proposed compensation algorithm reduced the mean absolute error of posture estimation from 9.010° to 2.270°, a reduction of approximately 74.8%. Consequently, the exercise success rate significantly improved from 62.7% to 87.3%, confirming that the system is effective in enhancing exercise performance while preventing injuries.