Smart Yoga Instructor for Guiding and Correcting Yoga Postures in Real Time

Area of Science:

• Computer Science
• Health & Fitness

Background:

• Yoga practice is increasingly popular globally, emphasizing the need for correct posture.
• Access to qualified yoga instructors can be limited, hindering effective practice.
• Personalized guidance is crucial for safe and beneficial yoga asana execution.

Purpose of the Study:

• To develop an interactive system that serves as a personal yoga instructor.
• To enable individuals to practice yoga with proper posture correction in their preferred environment.
• To leverage technology for accessible and real-time yoga posture feedback.

Main Methods:

• Utilized a camera to capture user yoga poses.
• Employed a pre-trained deep learning model for yoga posture classification against benchmark poses.
• Integrated audio feedback for real-time posture correction guidance.

Main Results:

• The system accurately classifies and provides corrective feedback for yoga postures.
• Real-time processing ensures a rapid response time of a few seconds.
• The system currently supports five common yoga asanas: Ardha Chandrasana, Tadasana, Trikonasana, Veerabhadrasana, and Vrikshasana.

Conclusions:

• The developed system effectively acts as a personal yoga instructor, enhancing practice accessibility.
• AI-powered posture analysis and audio feedback offer a viable solution for remote yoga training.
• This technology can support individuals in achieving correct yoga form and maximizing practice benefits.