Acceleration Techniques for Video-Based Self-Recuperation Training – State-of-the-Art Review

In recent years, advancements in optimized Deep Learning (DL) models have strived to speed up self-recuperation training effectively. Video-based acceleration techniques enhance the self-recuperation process concerning fast recovery, increase patient engagement, individual progress tracking, optimized exercise plans, remote monitoring, effective feedback generation for correction of exercise poses, and reduced healthcare costs. This article explored the state-of-the-art comparison of various acceleration techniques for hardware and software in non-invasive self-recuperation. Also, we have deliberated some optimized DL methods to be deployed in the edge computers of users and developed accelerated hardware to achieve high precision outcomes with low computation and memory consumption on a home-based self-recuperation training system. Finally, this article addresses challenges and suggestions for future research directions on home-based self-recuperation acceleration techniques.