Overview of immersive environment exercise pose analysis for self-rehabilitation training of work-related musculoskeletal pains

Exercise therapy is a protracted, difficult, and tiresome procedure in rehabilitating various physical disabilities due to recent working conditions in software, healthcare professionals, teaching communities, etc. From the perspective of selfrehabilitation training, the technological impact creates numerous solutions to analyze whether the patient/human is performing the exercises properly without any mistakes, as the physicians expect. In this chapter, initially, we explore a variety of physical disabilities due to the working environment, which is elaborated in existing works of literature. Second, review the various implementations of augmented reality (AR) and virtual reality (VR) and how it predicts the correctness of various exercise poses, pose estimation techniques, pros, and cons, and summarize the techniques employed in the immersive visual exercise pose analysis and the outcomes of its experiments through implications of different deep learning algorithms. The system's performance can be increased by optimizing depth analysis to result in the identification of more petite body part movements, adding more features, such as contour identification and more meta-attributes for particular points in 3D reconstruction on participants, scaling up the computational power, and focusing on the current model's refinement to achieve more accuracy and development of a multi-stage ensemble process. Without large datasets for analysis, the system's efficiency is relatively low. To maximize accuracy, real-time enhancement is necessary. More real-time data are required for training and testing to improve the best-effort solutions for classifying the user's exercise pose as proper or improper. To assist the research methodology on imperfections analysis in developing a new plan for future investigations, the correctness of exercise pose mistakes is visualized, and instructing the participants do it properly through audio feedback also.