Comparison Study of Inertial Sensor Signal Combination for Human Activity Recognition based on Convolutional Neural Networks

TL;DR

This study evaluates various inertial sensor signal combinations for human activity recognition using CNNs, achieving near-perfect accuracy with high-dimensional signals and highlighting the importance of specific sensor placements.

Contribution

It introduces an optimized CNN-based approach for HAR that systematically compares different inertial sensor signal combinations for improved accuracy.

Findings

High accuracy (99.77-99.98%) with signals of 12 or more modalities.

Lower dimension signals perform significantly worse, with 73-85% accuracy.

Sensor placement on chest and ankle is crucial for better recognition.

Abstract

Human Activity Recognition (HAR) is one of the essential building blocks of so many applications like security, monitoring, the internet of things and human-robot interaction. The research community has developed various methodologies to detect human activity based on various input types. However, most of the research in the field has been focused on applications other than human-in-the-centre applications. This paper focused on optimising the input signals to maximise the HAR performance from wearable sensors. A model based on Convolutional Neural Networks (CNN) has been proposed and trained on different signal combinations of three Inertial Measurement Units (IMU) that exhibit the movements of the dominant hand, leg and chest of the subject. The results demonstrate k-fold cross-validation accuracy between 99.77 and 99.98% for signals with the modality of 12 or higher. The performance of lower dimension signals, except signals containing information from both chest and ankle, was far inferior, showing between 73 and 85% accuracy.