Machine Learning-Based Home Surveillance System for Enhanced Security

View at publisher Summarise with VIRAT
Authors: Dulhara Anjana Langappuli, Tilani Gunawardena
Cited by: 0
References: 9

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper presents Sentinel, an innovative home security system that integrates machine learning for real-time threat detection with Internet of Things technologies for environmental monitoring. The system employs three custom machine learning models: a face recognition model using Multi-Task Cascaded Convolutional Networks (MTCNN), achieving 94.39% accuracy and surpassing You Only Look Once (YOLO) at 85%, a weapon detection model based on YOLOv8n with 85.4% accuracy and a custom anomaly detection model using MobileNetV2, achieving 94.32% accuracy for violence detection. IoT sensors monitor air quality, temperature, and gas leaks, while a mobile application enables real-time video streaming, instant alerts, and remote control of a smart door lock. By combining multiple machine learning models with IoT in a cost-effective design, Sentinel offers a scalable and comprehensive solution to modern home security challenges.

Fields of Study & Subjects

IoT-based Smart Home Systems Advanced Neural Network Applications Video Surveillance and Tracking Methods

Publication details

Journal / source
IEEE International Conference on Industrial and Information Systems (ICIIS)
Pages
61–66
Publication date
16 January 2026
Language
EN