NETRA

NETRA is an AI-powered vision assistant designed to empower the visually impaired. By leveraging the power of Google's Gemini AI, NETRA provides real-time identification of objects, currency, and text, communicating findings through synthesized speech in a user-friendly Hinglish format.

Features

• Real-time Object Recognition: Instantly identifies objects and scenes using the devise's camera.
• Currency & Text Detection: specialized capabilities for recognizing currency notes and reading text.
• Hinglish Audio Feedback: detailed yet concise audio descriptions in Hinglish (Hindi + English), tailored for the Indian context.
• Voice-Guided Interface: Provides auditory cues for system status (online, scanning, success, error).
• Vibration Feedback: Haptic feedback for user interactions and alerts.

Tech Stack

• Frontend: React (v19)
• Build Tool: Vite
• AI Model: Google Gemini 2.5 Flash Lite
• Routing: React Router DOM

Getting Started

Follow these instructions to set up and run the project locally.

Prerequisites

Ensure you have Node.js installed on your machine.

Installation

1. Clone the repository (if applicable) or navigate to the project directory.

2. Install the dependencies:

   npm install

Environment Setup

Create a .env file in the root directory of the project and add your Google Gemini API key:

VITE_GEMINI_API_KEY=your_api_key_here

Note: You need a valid API key from Google AI Studio to use the generative AI features.

Running the Application

Start the development server:

npm run dev

The application will be available at http://localhost:5173 (or the port shown in your terminal).

Usage

1. Grant Permissions: Allow the browser to access your camera upon launching the app.
2. Scan: Tap anywhere on the screen (or the video feed) to capture an image.
3. Listen: The app will process the image and speak out the description in Hinglish.