Because falls are the second leading cause of injury deaths, especially in the elderly according to WHO statistics, there have been a lot of studies on developing a fall detection and warning system. Many approaches based on wearable sensors, cameras, Infrared sensors, radar, etc., have been proposed to detect falls efficiently. However, it still faces many challenges due to noise and no clear definition of fall activities. This paper proposes a new way to extract 44 features based on the time domain, frequency domain, and Hjorth parameters to deal with this. The effect of the proposed feature set has been evaluated on several classification algorithms, such as SVM, k-NN, ANN, J48, and RF. Our method achieves a relative high performance (F1-Score metric) in detecting fall and non-fall activities, i.e., 95.23% (falls), 99.11% (non-falls), and 96.16% (falls), 99.90% (non-falls) for the MobileAct 2.0 and UP-Fall datasets, respectively.

Pedestrian fall detection attracts a lot of research due to its importance in building a warning system to avoid negative consequences. There are many techniques for fall detection that have used different devices such as cameras, environmental sensors, wearable sensors, etc. However, the popularity of wearable devices with many embedded sensors has motivated research on detecting abnormal activity such as falling based on sensor data. High accuracy of fall detection in daily life activities is always a practical challenge. This paper investigates different classifiers using a combined feature set that computes from Time, Frequency, and Hjorth domains using the wearable device’s accelerometer data only. This study also employs Particle Swarm optimisation (PSO) to find the parameters of the Random Forest classification, which aims to improve the accuracy of the fall detection system. The experimental results on the UP-Fall dataset have shown a better performance than the study of L. Martínez-Villaseñor et al. Combining the RF classifier with PSO also improves the stability and performance of the model.

Pedestrian fall detection attracting a lot of research due to its importance in building a warning system to avoid negative consequences. There are many techniques for fall detection that have using different devices such as cameras, environmental sensors, wearable sensors, etc. However, the popularity of smartphones with many embedded sensors has motivated research on detecting abnormal activity such as falling based on sensor data. High accuracy of fall detection in daily life activities is always a practical challenge. This paper put forward a new feature set that pulls out from time, frequency and Hjorth domains, which calculate from the smartphone’s accelerometer data. This study also employs Particle Swarm Optimization (PSO) to optimize the parameters of the Random Forest classification, which aims to improve the accuracy of the fall detection system. Experimental results on the MobiAct 2.0 dataset have shown that the put forward system has improved detection efficiency from 96.9% to 98.5% in F-measure. In addition, the detection efficiency in falls is 17% to 26.7% higher on the F-measure than the methods proposed by the authors C. Chatzaki et al.