Netra AI

AI-Powered Video Stream Security Application

Back to Portfolio

The Problem

Traditional video surveillance systems require manual review of hours of footage to find specific events, making them inefficient and time-consuming. When incidents occur, users often need to sift through extensive video recordings to understand what happened, which is both frustrating and impractical.

Manual Video Review

Users often need to manually review hours of surveillance footage to find specific events, which is time-consuming and inefficient. This was the exact challenge faced when someone stole something in front of a family member's house.

AI-Powered Solution

Netra AI transforms video monitoring with instant event querying and real-time alerts. Users can ask natural language questions about events and receive immediate, detailed responses.

Key Features

Instant Event Querying
Ask natural language questions about events in your video feed and get immediate answers
Real-Time Alerts
Receive instant notifications with detailed descriptions of important events as they happen
Speech Recognition
Voice-to-text functionality for easy communication with the system, making it accessible for all users
Motion Detection
Automatic detection and recording of motion events with configurable sensitivity settings
AI Analysis
Advanced AI processing using Google Gemini to analyze video content and extract meaningful insights
Smart Storage
Efficient storage and retrieval of alert messages using ChromaDB for quick access to historical data

How It Works

1
Video Processing
The system continuously monitors video streams using OpenCV for motion detection and records 5-second clips when activity is detected.
2
AI Analysis
Google Gemini AI analyzes the video content to understand what's happening and generates detailed descriptions of events.
3
Natural Language Queries
Users can ask questions like "Did someone come to my house?" and receive instant, accurate responses based on the analyzed video data.
4
Real-Time Alerts
Important events trigger immediate alerts with detailed descriptions, keeping users informed of significant activities in real-time.

Technologies Used

Python: Core programming language for the application
Google Gemini: AI model for video analysis and content understanding
OpenCV: Computer vision library for motion detection and video processing
DeepGram: Speech-to-text functionality for voice interactions
ChromaDB: Vector database for storing and retrieving alert messages
Reflex: Full-stack framework for frontend and backend development

Future Plans

Real-Time Text Alerts: Send detailed event descriptions via text messages, such as "Your daughter has arrived home" or "There is a fire outside your home."
Enhanced AI Training: Train the AI model on video data from different countries and diverse populations for better global applicability.
Improved Security: Implement Multi-Factor Authentication and encryption to enhance privacy and security of user data.