Yoga Pose Estimation using Artificial Intelligence

• Published in: 2023 International Conference on Data Science and Network Security (ICDSNS) pp. 1 - 6
• Main Authors: R, Janardhana D, H, Shashwatha P, V, Manasa, R, Kavya H, R, Pruthvi
• Format: Conference Proceeding
• Language: English
• Published: IEEE 28.07.2023
• Subjects: Accuracy; Body Landmarks; Deep learning; Media pipe; Mental health; Network security; Open pose; Pandemics; Pose estimation; Schedules; Training
• DOI: 10.1109/ICDSNS58469.2023.10245467

The COVID-19 pandemic has had a significant impact on people's physical and mental health worldwide. As countries slowly recover from the pandemic, the importance of yoga has become increasingly apparent. One of the most significant benefits of yoga is its ability to improve mental health. Studies have shown that practicing yoga regularly can reduce stress, anxiety, and depression, which are all common symptoms of the pandemic. Furthermore, the pandemic has resulted in a significant increase in sedentary lifestyles, which can lead to a variety of health problems. Yoga is an excellent way to stay active, as it can be practiced at home and requires no special equipment. Numerous studies have put forth algorithms and employed various techniques to identify yoga postures using machine learning or deep learning methodologies. Our proposed work utilizes Machine Learning and Deep Learning techniques to recognize yoga postures and is designed to benefit yoga practitioners by allowing them to monitor their postures and improve their technique through continuous practice with guidance. Key points on the human body are detected using the Media Pipe framework and OpenCV, and a trained model is used to classify the yoga pose and provide feedback to the user. The proposed model outperforms with state-of-the-art methods in detecting the yoga pose more accurately with 99.53% accuracy. Publicly available datasets used for training the model and user can choose their preferred yoga posture for training.