HUMAN ACTIVITY RECOGNITION AND FALL DETECTION USING VIDEO AND INERTIAL SENSORS

Abstract

Falls are a crucial problem for elderly people. Early detection of falls may prevent or attenuate possible negative consequences for elderly people. There is a number of scientific articles on the topic of detecting falls using machine learning techniques. While some of them focus on fall detection systems based on scalar body sensors, others apply vision based detection. The goal of this thesis is to try to perform a fusion of inertial sensor based and video sensor based modules to provide a more robust solution, as each method has their own drawbacks in terms of both performance and feasibility.