Machine Learning Deteksi Jatuh Menggunakan Algoritma Human Posture Recognition

Abstract

The increase in the number of elderly people in Indonesia is quite high. One thing that must be considered is that there are elderly people who live alone without family at home. This has a high risk, especially for biological aspects, that is, if something unwanted happens, one of them is falling, so a fall detection system is needed to monitor the condition of the elderly at home. In this study, a human fall detection system has been designed using image processing with input from the camera to detect falls using the Human Posture Recognition Algorithm. To obtain digital images using the Image Acquisition technique and using the Human Posture Recognition Algorithm algorithm to detect falls in the video. In this research, a human fall detection system has been designed using image processing with input from the camera, the Motion History Image (MHI) method and the Approximated Ellipse. This method will produce parameter values C_motion, Sigma_theta, and Sigma_rho which will be used as a reference for fall detectors. The results of this study indicate an accuracy of 95.33% for data on conditions of falling or not falling.