AI-based Intrusion Detection System for IoT-Enabled Smart Homes

Over the past few years, there has been a surge in demand for intelligent home and office security systems based on rising security needs. This project introduces an AI-powered Smart Intrusion Detection System that combines motion sensing, object detection, and face recognition for real-time monitoring and threat detection. The system employs a PIR motion sensor to monitor activity in the environment and a camera module to record frames on detection. Object detection is carried out through a TensorFlow Lite (TFLite) model — SSD MobileNet V2, that has been trained on the COCO dataset, detecting the presence of humans or other objects in motion with high effectiveness on low-resource devices such as the Raspberry Pi 4. After detection of a person, the DeepFace framework utilizing the VGG-Face model crosschecks the identity of the person by matching their face with pre-stored images of authenticated users. The suggested system operates without an internet connection, can analyze data in real time, and can be connected to MQTT and Telegram alerts for remote alerting. The study demonstrates the creation of intelligent surveillance systems that can autonomously distinguish between authorized users and potential intruders by integrating deep learning algorithms with cost-effective embedded hardware.