AI – Integrated Smart Yoga Mat

Author Name : Sika K., Adithyan E., Balaji D, Mathan K.

ABSTRACT

The Real-Time Yoga Pose Detection and Feedback System offers an innovative approach to enhancing self-guided yoga practice by deploying an AI integrated smart yoga mat with embedded flex sensors and a webcam to collect real-time data on body alignment, joint angles like elbows, shoulders, knees, and mat pressure distribution, which is processed locally on a standard laptop using Media Pipe for pose estimation and OpenCV for video rendering. Advanced algorithms interpret this data to detect posture misalignments, providing immediate feedback through a 16x2 LED display showing mat occupancy "Mat Occupied/Free" and posture alerts like "Your posture is wrong", alongside audio feedback "You done correctly/wrongly" via the Speech Synthesis API and tactile feedback through vibration motors, all controlled by an Arduino microcontroller. Accessible via a Flask web interface on any modern browser, the system delivers comprehensive session summaries and posture analysis reports with ~95% accuracy in pose classification and low latency ~120 ms per frame, enabling practitioners to make informed adjustments for consistent improvement. By promoting real-time posture correction and reducing injury risk, the system enhances safety, accessibility, and effectiveness in yoga practice, positioning itself as an indispensable tool for practitioners of all skill levels, yoga instructors, and wellness centres.