Unobtrusive and Non-Invasive Human Activity Recognition using Kinect Sensor

The main objective of this paper is to present a depth imaging human activity recognition system that is basically used to monitor and detect normal activities of the person in the indoor environment. The Proposed system runs in real-time and is capable of detecting moving falls, sitting fall accurately and robustly without taking into account any false positive activities. Human activity recognition with high accuracy is a huge challenge in the research world. Fall detection is an important technology in health care and elderly person activity monitoring. The proposed method is feasibility and workability are illustrated throughout the experimental result that shows the perfect human tracking and activity detection. System works well with low lighting conditions; lighting does not affect the accuracy of detecting activities. Moreover, the proposed method is able to avoid false positive as: lying down, retrieve something from the floor.