Development of Ai-Based Posture Monitoring System to Assist Yoga Training

In the modern era of remote fitness and wellness, maintaining correct posture during exercises and yoga practices is vital for maximizing the benefits and preventing injuries. However, monitoring one's posture can be challenging, often leading to improper form and reduced effectiveness. To address this issue, this project presents an innovative real-time posture monitoring system using computer vision and pose estimation techniques. The system employs the Mediapipe library to detect and analyze the user's body posture through a camera feed. It begins by capturing a reference posture from an image and calculating its landmark coordinates. These coordinates serve as the baseline for correct posture. As the user performs exercises in front of the camera, their live posture is continuously evaluated against the reference posture using Euclidean distance as a similarity metric. A crucial contribution of this project lies in its ability to provide immediate feedback to the user regarding their posture. The system visualizes landmarks on the user's body, highlighting correct and incorrect positions in distinct colors. If the similarity score between the live posture and the reference posture surpasses a certain threshold, the landmarks appear in green, indicating good posture. Conversely, red landmarks indicate discrepancies, prompting the user to adjust their form. Experimental results showcase the effectiveness of the proposed system in real-time posture assessment. Users can actively observe their posture and make necessary adjustments, fostering improved exercise outcomes and reduced risk of injuries. By offering a convenient and immediate feedback mechanism, this project bridges the gap between remote fitness practices and professional guidance, ultimately promoting safer and more effective workout routines