Skip to content

A powerful web application engineered to handle high volumes of emergency calls, this system automatically prioritizes based on severity and quickly dispatches the nearest, most capable response teams.

Notifications You must be signed in to change notification settings

miit-daga/DisMan

 
 

Repository files navigation

Disaster Management using Generative AI

This is a prototype for the project "Disaster Management using Generative AI."

Steps to Run the Prototype Locally

  1. Clone the repository:

    git clone https://github.com/your-username/DisasterManagement.git
  2. Navigate to the project directory:

    cd DisasterManagement
  3. Run the Flask application:

    python app.py
  4. Set up Ngrok:

    • Navigate to the folder ngrok-v3-stable-windows-amd64.
    • Run the ngrok.exe file inside the folder.
  5. Start the ngrok local server:

    ngrok http 5000
    • A new server will start at localhost:5000.
    • Copy the forwarding link provided by Ngrok in the terminal window.
  6. Configure the Twilio webhook:

    • Take the forwarding link from Ngrok and append /voice to it.
      https://<ngrok-forwarding-link>/voice
      
    • Go to your Twilio console and paste this link in the Webhook URL for your active number.
  7. Save and continue on Twilio:

    • Make sure to add your Account SID and Auth Token in the app.py program (replace placeholders).
  8. Make a call to your Twilio number:

    • Call the number, and you'll hear instructions to describe your emergency.
    • After the beep, describe the emergency and end the call.
  9. Recording and Transcription:

    • The program will:
      • Download the call recording.
      • Transcribe the audio.
      • Delete the recording from the local machine after transcription.
  10. View Transcription and Disaster Type Matching:

    • The transcription will be printed in the terminal.
    • Google's gemini will analyze the information reflected by the transcribed text and return JSON format
    • The JSON format contains any name mentioned by the caller, address, disaster type
    • It will also give a sentiment analysis based on the transcribed text

About

A powerful web application engineered to handle high volumes of emergency calls, this system automatically prioritizes based on severity and quickly dispatches the nearest, most capable response teams.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages

  • JavaScript 87.2%
  • Python 9.7%
  • HTML 2.3%
  • CSS 0.8%