The practice of yoga necessitates precise alignment in asanas to maximize benefits and minimize the risk of injuries. For beginners or those practicing without a teacher, maintaining proper alignment can be challenging. The project aims to develop a computer vision-based system to guide users in yoga poses, providing innovative and technological support for self-learning and safe practice.

The system utilizes a dataset comprising images and videos displaying a variety of yoga poses.

Advanced neural networks are employed for pose detection, incorporating technologies such as MediaPipe for pose capture and OpenCV for image processing. These networks are trained to identify and assess yoga poses, offering real-time feedback on alignment and suggestions for corrections.

Personal Project WIP in collaboration with Annalisa Porcelli for the Machine Leraning for Vision and Multimedia Course | DAUIN Department – Politechnic of Turin