Muscle Vision: Real Time Keypoint Based Pose Classification of Physical Exercises
This work addresses the need for accessible fitness tracking tools, but it is incremental as it applies existing pose recognition techniques to a specific domain.
The authors tackled the problem of real-time classification of physical exercises from live video by developing a lightweight keypoint and time series-based pipeline, resulting in a web application that performs pose recognition and classification with live display.
Recent advances in machine learning technology have enabled highly portable and performant models for many common tasks, especially in image recognition. One emerging field, 3D human pose recognition extrapolated from video, has now advanced to the point of enabling real-time software applications with robust enough output to support downstream machine learning tasks. In this work we propose a new machine learning pipeline and web interface that performs human pose recognition on a live video feed to detect when common exercises are performed and classify them accordingly. We present a model interface capable of webcam input with live display of classification results. Our main contributions include a keypoint and time series based lightweight approach for classifying a selected set of fitness exercises and a web-based software application for obtaining and visualizing the results in real time.