Abnormal Behavior Forecasting in Smart Homes Using Hierarchical Hidden Markov Models

Abstract

Computer vision techniques having the capability to inspect human behavior are gaining fame. The conception of intelligent abnormal human activity visual identification has raised the standards of monitoring systems, situation appreciation, homeland protection, and intelligent automated surroundings. Numerous scholars are submitting through past years their inspection regarding behavior detection. Nevertheless, abnormal human activity is dissimilar in itself owing to the factors as (a) the central meaning of anomaly (b) feature demonstration of an anomaly, (c) its application, and henceforth (d) the dataset. Human behavior detection in automatic control of home appliances is the mission of spotting a human being’s behavior patterns so as to construct harmless surroundings for that human being. It is beneficial in constructing surroundings for older grown human being or to aid any human being in his/her everyday life. The motive of the put forward study is to establish a model detects the behavior of grown-up people living in a smart house, to inspect abnormal behavior and alert relatives or a caretaker aware if assistance is required. The foremost goal of exploiting hierarchical hidden Markov models is foretelling whether the contemporary activity is normal or abnormal. Hierarchical hidden Markov models are being exploited as it is a statistical method that works well with a diminutive dataset or unsatisfactory training data. In the lead, the statuses of the hierarchical hidden Markov models generate chain of surveillance symbols rather than solitary surveillance symbols as the same case as it is for the standard hidden Markov models statuses. Results acquired throughout that model were encouraging as they decreased both cost and computational time.