GuardianEye

AI-Powered Home Safety and Incident Detection System

Overview

GuardianEye is an intelligent home safety monitoring system that uses real-time computer vision and AI reasoning to detect potential risks, emergencies, or unusual behaviors from live camera feeds. When the system detects a dangerous situation (such as a fall, smoke, or hazardous motion), it allows the security monitortrigger notifications — for example, sending SMS alerts via Twilio — to keep users informed and safe. GuardianEye integrates cutting-edge AI tools, cloud infrastructure, and a modern web interface to provide a seamless safety monitoring experience.

Features

Real-Time AI Vision

• Captures live video frames from a webcam using MediaPipe.
• Performs intelligent visual analysis via Google Gemini AI using the Strands framework.
• Detects ambiguous or dangerous situations (e.g., falls, unsafe movements, water spills).

Intelligent Incident Analysis

The backend runs a Gemini 2.5 Flash model prompt-engineered to identify safety incidents in the following format:

[
  {
    "incident": "Fall detected",
    "emergency_level": "high",
    "summary": "A person appears to have fallen near the kitchen area.",
    "suggestion": "Check on the individual immediately and call emergency services if necessary."
  }
]

Real-Time Alerts

• Integrates Twilio SMS API to send immediate text alerts to caregivers or homeowners when an emergency is detected.
• GuardianEye Alert: Possible fall detected in the living room. Please check immediately.

Modular Architecture

• Frontend (React + TypeScript): clean interface for live camera view and status dashboard.
• Backend (FastAPI + Python): handles AI inference and Twilio notifications.
• AI Service Layer: leverages Google Gemini through Strands SDK for scalable, interpretable reasoning.

Setup Guide

1. Clone the Repository

$ git clone https://github.com/ryanraen/GuardianEye.git
$ cd GuardianEye

2. Backend Setup

$ cd backend
$ python -m venv venv
$ source venv/bin/activate
$ pip install -r requirements.txt

3. Create a .env file in the backend directory:

GEMINI_API_KEY=your_google_genai_key
TWILIO_ACCOUNT_SID=your_twilio_sid
TWILIO_AUTH_TOKEN=your_twilio_auth_token
TWILIO_MESSAGING_SERVICE_SID=your_twilio_service_sid

4. Run the backend:

python routes/app.py

5. Frontend Setup

cd frontend
npm install
npm start

Tech Stack

Layer Technology

• Frontend: React, TypeScript, MediaPipe, TailwindCSS
• Backend: FastAPI, Python, Strands SDK, Google Gemini

Notifications

• Twilio API

AI Model

• Gemini 2.5 Flash

Runtime

• Frontend: Node.js
• Backend: Uvicorn