Human activity recognition using support vector machine for automatic security system

Abstract

Home is a place for human beings to live and also to socialize with the society. In this rapid technology development, a home security system is a must to prevent crimes. One of the innovation is home automation which enables controlling the doors automatically. RGB-D camera can detect human body movements which later can be used to recognize the activity through the Support Vector Machine (SVM) algorithm. The result of activity prediction can be used as home automation input for the home security system. In this paper, a security system analysis has been carried out which is beneficial for the users in maintaining the security system through combining the RGB-D Camera and Skeleton Tracking which then classified by using SVM algorithm. The result shows that the optimal data resulted through 1 meter distance with 100% accuracy, while testing variable C of SVM resulting in C = 2 as the best score for C variable with 92% accuracy.