Publication

Abstract : Considering home security being more prominent nowadays, its proper surveillance and alerts at the right time warrant utmost importance. Our project focuses on an enhanced home security system that integrates the surveillance systems with powerful machine learning tools that guarantee a flawless responsible home safeguard system. To implement a real-time home security system for human intrusion detection, a remote monitoring video surveillance system based on Wi-Fi was developed. The video captured from the camera is further segmented and pre-processing techniques like Histogram of Oriented Gradients (HOG) and HAAR cascade algorithm are used for smartly eliminating false alarm due to animals. The former extract features from the input image whereas the latter is a cascade classifier that is used for identifying the objects in an image as propounded by Viola-Jones. The combination of these features is fed to SVM for multi-stage classification. The system then identifies the intruder automatically alerts the home resident and sends an alert message to the user's mobile using IoT. In this chapter, the proposed work is implemented using Arduino, PIR sensor, wi-fi camera and evaluated in Python and Matlab2019. The power of this video surveillance system is efficiently enhanced with the use of proximity sensors. Experiment results show that the integration of hardware with effective machine learning techniques can attain remote surveillance with high reliability and accuracy.