Performance Monitoring for Exercise Movements using Mobile Cameras

Despite numerous devices targeted to fitness tracking, the strength training domain has often been overlooked and understudied. In this paper, we propose a smartphone camera based approach to track users' strength training workouts, as well as metrics pertaining to their form and performance. Our goal is to detect the repetitions in a workout without requiring user intervention or any training data from the user. Unlike many existing systems, our proposed system is scalable, low-cost, and widely accessible. We gather data from two sources for 5 exercises across 25 subjects. We compute performance metrics such as range of motion, velocity, and duration from each repetition with median errors less than 10%. These results demonstrate that commercial off the shelf smartphone cameras can be used to accurately detect and count repetitions in user movements, as well as to compute rep-by-rep user performance.