A Combined One-Class SVM and Template-Matching Approach for User-Aided Human Fall Detection by Means of Floor Acoustic Features.

Diego Droghini,Daniele Ferretti,Stefano Squartini,Francesco Piazza,Emanuele Principi

doi:10.1155/2017/1512670

Diego Droghini, Daniele Ferretti + Show 3 more

Open Access

https://doi.org/10.1155/2017/1512670

Copy DOI

Abstract

The primary cause of injury-related death for the elders is represented by falls. The scientific community devoted them particular attention, since injuries can be limited by an early detection of the event. The solution proposed in this paper is based on a combined One-Class SVM (OCSVM) and template-matching classifier that discriminate human falls from nonfalls in a semisupervised framework. Acoustic signals are captured by means of a Floor Acoustic Sensor; then Mel-Frequency Cepstral Coefficients and Gaussian Mean Supervectors (GMSs) are extracted for the fall/nonfall discrimination. Here we propose a single-sensor two-stage user-aided approach: in the first stage, the OCSVM detects abnormal acoustic events. In the second, the template-matching classifier produces the final decision exploiting a set of template GMSs related to the events marked as false positives by the user. The performance of the algorithm has been evaluated on a corpus containing human falls and nonfall sounds. Compared to the OCSVM only approach, the proposed algorithm improves the performance by 10.14% in clean conditions and 4.84% in noisy conditions. Compared to Popescu and Mahnot (2009) the performance improvement is 19.96% in clean conditions and 8.08% in noisy conditions.

Highlights

The ageing of population is posing major concerns in governments and public institutions, since it will consistently increase the demand for healthcare services and the burden on healthcare systems [1]
This paper proposed a combined One-Class SVM (OCSVM) and templatematching classifier to discriminate human falls from nonfalls in a semisupervised framework
Fall signals are captured by means of a Floor Acoustic Sensor, and Mel-Frequency Cepstral Coefficients (MFCCs) and Gaussian Mean Supervectors (GMSs) are extracted from the acquired signal

Summary

Introduction

The ageing of population is posing major concerns in governments and public institutions, since it will consistently increase the demand for healthcare services and the burden on healthcare systems [1]. “analytical methods” classify an event as a fall or nonfall by thresholding the acquired signals or the features extracted from them [5]. These methods require manual tuning of their hyperparameters for different operating scenarios and subjects. “machine learning” methods learn to discriminate falls from nonfalls directly from the data [5]. They can be divided into “supervised methods,” which require a labelled dataset for training, and “unsupervised methods,” which base their decision on a normality model built from nonfall events only. Their weakness is that certain events deviate from normality as the human fall (e.g., the fall of an object), and they may produce false alarms

Methods

Results

Conclusion