Convolutional Neural Network-Based Human Activity Recognition For Edge Fitness and Context-Aware Health Monitoring Devices

Abstract

The vital signs can vary significantly depending on the daily physical activities, which may not be due to defects of the organs. Under remote human health monitoring applications, for reliable disease diagnosis, recorded biomedical signals such as electrocardiogram (ECG), photoplethysmogram (PPG) must be indexed with physical activity information, as it is unknown to the physicians and also to the computer aided diagnostic system. Since deep learning networks were explored for various vital signs extraction and disease classification, we present an effective convolutional neural network (CNN) based human activity recognition (HAR) method by exploring suitable hyper-parameters. CNN based methods are evaluated by using the acceleration signals taken from the standard HAR benchmark database, University of California, Irvine (UCI) with accuracy (ACC), model size (in kB) and processing time (PT), and also implemented on the Raspberry Pi 4 (R-Pi-4) to study real-time feasibility. Evaluation results showed that higher HAR accuracy can be achieved with activation function of exponential linear unit for the 2-layer CNN and segment size of 2.5 seconds (ACC of 89.05% and PT of 0.142 msec), the 4-layer CNN and segment size of 1 second (ACC of 91.66% and PT of 0.541 msec), and the 6-layer CNN with segment size of 2 seconds (ACC of 91.18% and PT of 1.672 msec). Results demonstrated that selection of an optimal number of layers, and hyper-parameters plays a major role in achieving higher accuracy with lower computational time on both PC-CPU and R-Pi computing platforms, which were not addressed in the past studies. IEEE