Physical Activity Repetition Count Extraction From Sparse mmWave Radar Point-Cloud

Abstract

Contactless fitness tracking using mmwave radars is gaining a lot of momentum fueled by radar on a chip solution. Apart from accurate classification of physical activity, repetition count of activity is also an important parameter to effectively track the fitness routine, calorie burnout and provide statistics of the workout to the user. In this paper, we present a data processing method using alpha shape and locally weighted scatter-plot smoothing to reliably extract the repetition count of physical activities using the sparse mmwave radar point cloud data. The presented method is able to overcome the challenges of sparseness and non-uniform sampling of the mmwave radar point cloud data.