Real time direction-sensitive fall detection system using accelerometer and learning classifier

Abstract

Continuous fall recognition framework screens the day by day action of particularly elderly individuals to enroll somebody's assistance as quick as conceivable if there should be an occurrence of crisis. This paper presents a real-time fall detection using a single 3D commercial accelerometer (3DCA) and support vector machine learning algorithm (SVMLA). In past, two machine learning (ML) based calculations SVMLA and k-Nearest Neighbors (K-NN) were executed for mandate fall discovery in reproduction. Among the two strategies, SVMLA give better exhibitions which prompts 96.45% of exactness utilizing PCA mean and standard deviation highlights, surpassing the exhibitions detailed in the writing. The performances of the developed system in real time are also evaluated and they are found same accuracy, precision and recall. When applied to experimental data from 13 male subjects, the real time system discriminates between falls and activities of daily living (ADL) with same level like simulation. The system utilizes privacy preserving sensor. The system is reliable, user friendly and cost effective with less technical error rate and high classification accuracy.