AI driven gas flare monitoring application!
Table of Contents
Problem: Oil & gas extraction operations burn off excess natural gas (methane) when the extraction exceeds purification and pipeline capacities. However, in recent years flyovers show that in major basins 5% of flares are unlit and venting, while another 5% are malfunctioning with incomplete combustion. Here’s an example study from the Permian basin. Since methane is a far more potent gas than CO2, venting this gas into the atmosphere contributes significantly to global warming - burning (flaring) it reduces the warming effects by 96%.
Solution: Android application that continually monitors flare. Takes a picture every few seconds and sends to AI model to classify. If classified as unlit sends notification to all phone numbers in contact list through Twilio API.
- React Native
- Roboflow
- Twilio
Let's get you started!
-
Contact Firestart team for Roboflow api key
-
Clone the repo
git clone https://github.com/organization-x/project-firestart/
-
Install packages
npm install
-
Create twillio account and create a twilio service for sending sms (https://www.twilio.com/docs/serverless/functions-assets/quickstart/send-sms-and-mms)
-
Switch environment variables in .env
6.Follow steps for expo deployment!
Watch walkthrough video here.
Ai model with 98.2% validation accuracy. Trained on Roboflow.com.
- Team Lead: Seth Bassetti
- Product Manager: Asad Shahid (https://github.com/AsadShahid04)
- Frontend Developer: Alan Than
- Frontend Developer: Pallavi Kamat(https://github.com/pkam001)
- Machine Learning Engineer: Arhant Choudhary(https://github.com/gorpyshortlegs/)
- Machine Learning Engineer: Keilani Li(https://github.com/keil4ni)
Distributed under the MIT License. See LICENSE.txt for more information.