Pose Estimation and Correcting Exercise Posture

Abstract

Posture plays a crucial role in maintaining both physical and mental well-being. Incorrect posture during exercises can lead to injuries and reduce workout efficiency. Traditional posture detection methods rely on sensor-based and image-processing approaches, but they often require wearable devices or manual supervision. This study proposes an AI-powered exercise posture correction system using pose estimation techniques, specifically OpenPose, a multi-stage CNN model. The system detects key body joints from images or videos, analyzes posture alignment, and provides real-time corrective feedback to improve form. By leveraging computer vision and deep learning, the proposed solution offers an automated, non-invasive, and efficient method for monitoring and improving exercise posture, benefiting applications in fitness training, rehabilitation, and injury prevention.