Smart technologies and intelligent computing for AI-driven gesture recognition in rehabilitation, yoga, and learning

In this paper, we propose a lightweight and easy-to-use MediaPipe-based computational pipeline for static hand gesture recognition with applications in rehabilitation, yoga, and educational setups. The proposed system detects 21 anatomical hand landmarks and uses them to extract a set of geometric features such as angles and distances to recognize yoga mudras with an average accuracy of 92%, under different illumination and partial occlusion scenarios. Interact with corrective feedback in real-time to practice Safe Rehab, perform yoga mudras accurately, and learn through gestures interactively. The exploitation of the economical, portable and robust model are expected to meet healthcare and educational technology needs for Industry 6.0. The results indicate that the proposed framework achieves dependable hand posture detection and feedback precision in multi-session scenarios, which paves the way to intelligent, AI-based digital wellness systems.