The accurate execution of yoga postures is essential for reaping the full physical, physiological, and mental benefits of the practice. Traditional learning requires expert supervision, yet many practitioners, especially those engaging through digital platforms, lack real-time correction mechanisms. This review investigates recent experimental studies that integrate Artificial Intelligence (AI) for posture detection and correction in yoga practice. It focuses solely on empirical studies applying AI models such as PoseNet, MediaPipe, OpenPose, and hybrid neural networks, while excluding survey-based or conceptual works. Findings suggest that AI systems can classify and correct poses in real time, enhance accessibility to safe yoga practices, and reduce the risk of injury, especially in home-based practice settings. The review also identifies current limitations and areas for future research in making AI-enabled yoga instruction more adaptive, accurate, and user-centric.

Artificial Intelligence, Yoga, Posture Correction, Pose Estimation, Deep Learning, Real-Time Feedback