SMART HOME SECURITY SYSTEM FOR INTRUDER DETECTION USING YOLO V3 ALGORITHM

Abstract

Empirical studies suggest that environment is the main factor influencing crime patterns so that closed circuit television (CCTV) becomes the choice to reduce the risk of crime. However, CCTV is less effective because it requires high bandwidth & storage and cannot provide notifications. Therefore, a technology called the Internet of Things (IoT) appears so that CCTV or webcam can work together with sensors to detect the presence of intruders and provide notifications. This study proposes a system that detects intruders and sends notifications to homeowner regardless of time and place. This system is usually referred to as a smart home security system. The authors proposed the smart home security system for intruder detection using YOLO V3 algorithm. This research is implemented using magnetic door sensor and alarm or buzzer and a webcam connected to Raspberry Pi 3B+. This proposed system will send notifications using telegram. We have implemented the proposed system and found the result that most accurate intruder detector is the YOLO V3 algorithm with an accuracy of 98.58%.