Machine Learning and Computer Visualization for Monocular Biomechanical Analysis

The gap between machine learning-assisted digital processing and modern scientific analysis has been closed by advances in computer vision technologies. In the realm of biomechanics, tactics that combine traditional and machine-assisted methods have demonstrated remarkable effectiveness in enhancing the electromyographical sensor observations; nevertheless, these methods are limited by specialized, multi-source arrays. This work is an effort to apply multiple different technologies to a single optical sensor in order to obtain comparable outcomes. The goal of this work is to provide an overview of the research, development, and use of a monocular feature extraction pipeline that can be used to enhance current sports biomechanics analysis. We will look at the methods that related publications have provided and how we have incorporated these approaches into our framework.