Yoga Pose Detection Using Deep Learning

Yoga helps improve the body, mind, and overall health. But practicing yoga the wrong way can be harmful. Long ago, people learned yoga under the guidance of an experienced teacher (Guru). But these days, it is hard to find a good teacher due to time and location constraints. That is why it is important to learn the right yoga poses right from the beginning. Deep learning can help by recognizing yoga poses and providing feedback for corrections. Mediapipe is a system that can detect human postures in real time. In this study, used Mediapipe to build a model to recognize yoga poses. For image feature extraction and pose classification, VGG16 model and SVC (Support Vector Classifier) are used as Convolutional Neural Network (CNN) accurately recognizes yoga postures. The goal of this study is to develop a model that can accurately classify different yoga postures. In this research suggested a simple neural network model which checks whether a yoga posture is performed correctly or not. This framework classifies different yoga postures with high accuracy and improved efficiency compared to traditional methods. These findings highlight the possibility of developing an automated system to support monitoring of yoga training and improvements. The implemented system gives 98% accuracy