APPLICATION OF DECISION TREES TO THE FALL DETECTION OF ELDERLY PEOPLE USING DEPTH-BASED SENSORS

Piotr Bilski, Paweł Mazurek, Jakub Wagner, Wiesław Winiecki

Abstract

The paper presents application of the Decision Tree (DT) to the fall detection of elderly people monitored by the infrared depth sensors. The decision making system works on data acquired by the sensor, recording movement of the person and raising the alarm if his/her behaviour suggests that the accident occurred. From the measurement data so-called morphological features are extracted, further processed by the DT. Various configurations of the classifier have been verified, proving its usefulness to solve the presented task, but also revealing disadvantages.

Keywords

fall detection, decision tree, depth sensor, binary classification

Download

IMEKO-WC-2015-TC18-349.pdf

DOI

-

IMEKO TC

TC18 - Measurement of Human Functions

Back to the proceedings