MethaneGuard AI '26

Build the AI that smells trouble before the planet does.

Prize · 2000 cash prize for the team and wooble goodies worth 2000 for all builders of the team.

Evaluation only · Open

MethaneGuard AI '26 is an AI and Computer Vision hackathon focused on one of the world's most urgent climate challenges—detecting methane leaks before they become environmental and financial disasters. Participants will build intelligent systems capable of analyzing satellite imagery, drone footage, and environmental sensor data to identify methane emission hotspots, estimate leak severity, and help organizations prioritize inspections. The goal is to transform raw environmental data into actionable intelligence that enables faster decisions, reduces emissions, and strengthens regulatory compliance.

What you can win

2000 cash prize for the team and wooble goodies worth 2000 for all builders of the team.

1.1st place
- Wooble Branded Borosilicate Glass Bottle , Branded premium Pen, Branded cushioned diary, Notepad and a premium branded Box
- Talent pool shortlist and Interview Opportunities
₹2,000

Also included

Expert review
Your work is reviewed by experienced practitioners.
Participation certificate
Get a verifiable certificate the moment you submit your work.

Overview

Methane doesn't announce itself. It's invisible, odorless, and quietly does more short-term climate damage than CO2 — while leaking out of pipelines, storage tanks, and refineries that nobody is watching closely enough. A single undetected leak can run for weeks, and by the time anyone notices, the environmental damage — and the regulatory fine — is already done. Right now, energy companies are trying to monitor thousands of kilometers of infrastructure using inspection schedules and hope. Satellites and sensors are already generating oceans of data. What's missing is the intelligence to actually read it. You're building the eyes that never blink — an AI system that catches the leak nobody was looking at.

The brief

Methane leaks represent one of the most overlooked yet significant contributors to climate change. Every year, thousands of small leaks remain undetected across oil and gas infrastructure because continuous physical inspection is impractical. These leaks quietly release large volumes of methane into the atmosphere, resulting in environmental damage, financial losses, regulatory penalties, and increased operational risk.

Although modern satellites, drones, and environmental sensors continuously collect vast amounts of data, extracting meaningful insights from these datasets remains a major challenge. Traditional monitoring approaches rely heavily on manual analysis, scheduled inspections, or simple threshold-based alerts, all of which are slow, resource-intensive, and prone to both missed detections and false positives.

Organizations need intelligent systems capable of processing environmental data in real time, accurately identifying methane emission hotspots, estimating the likelihood and severity of leaks, and helping inspection teams focus on the locations that matter most.

Your challenge is to design an AI-powered methane detection platform that can:

Detect methane emission hotspots from satellite imagery, drone imagery, or sensor datasets. Differentiate genuine methane leaks from environmental background noise. Estimate the probability, location, and severity of each detected leak. Prioritize inspection efforts using AI-generated risk scores. Present actionable insights through an intuitive environmental monitoring dashboard. Improve environmental compliance while reducing unnecessary inspections and operational costs.

The ultimate objective is to build an intelligent early-warning system that enables organizations to detect methane leaks before they become costly environmental and operational disasters.

Recommended datasets

NASA AVIRIS Methane Plume Dataset Best for: Computer Vision / methane plume detection Data: Hyperspectral airborne imagery containing methane plume observations. Good for: Detecting and locating methane emissions from imagery. Recommended as the main dataset for your hackathon.
TROPOMI / Sentinel-5P Methane Data The Copernicus Data Space Ecosystem provides Sentinel-5P Level-2 methane (CH₄) data globally, with archive data available from April 2018 onward. Download / Explore Sentinel-5P Methane Data - https://dataspace.copernicus.eu/explore-data?utm_source=chatgpt.com

What to select: Mission: Sentinel-5P Product: Level-2 Methane (CH4) Timeliness: OFFL / NTC for archived data Format: NetCDF (.nc)

You can also use the Copernicus Browser to select a geographic area and date and download products. A free account is currently required.

Deliverables

Working AI Solution capable of detecting methane leak hotspots from satellite imagery, drone imagery, or environmental sensor data.
Environmental Monitoring Dashboard displaying detected leaks, confidence score, risk level, and inspection priority.
GitHub Repository containing the complete source code with setup instructions and documentation.
Project Documentation (PDF or README) explaining the problem, dataset, approach, model, results, and future improvements.
Demo Video (3–5 Minutes) showcasing the solution, key features, and end-to-end workflow.
Presentation Deck (Maximum 10 Slides) covering the problem, solution, architecture, results, and impact.

Evaluation criteria

AI Model Performance & Accuracy 30
Technical Implementation 20
Innovation & Problem Solving 20
Dashboard & User Experience 15
Documentation, Demo & Presentation 15