Yoga Pose Detection with Deep Learning and Computer Vision
Amisha Srivastava, K. Gayathri, Senthil Kumar Thangavel, Pullela Meghana, S. Vishvajit, B Senthil Kumar, Kanagasabapathi Somasundaram · 2024
An approach to do real-time monitoring of Yoga Asanas using Deep Learning and Computer Vision approaches. Convolutional neural networks (CNN) and long short-term memory (LSTM) are combined to create a hybrid deep learning model. Human pose recognition can be used to create a selfinstruction exercise system that enables people to learn and perform exercises appropriately on their own as these resources are not always readily available. This project discusses several machine learning and deep learning algorithms to precisely identify yoga positions on pre-recorded films as well as in realtime, laying the groundwork for developing such a system. These kinds of applications are useful during times of lockdown, such as the lockdown we experienced in 2020 due to the coronavirus epidemic, when people’s freedom of movement is severely constrained and they may use such programmers’ quite easily from home.