Using U-Net Model to Detect Wildfire from Satellite Imagery

An experimental repository to build ML models and perform efficient wildfire smoke detection.

This repository showcases our work on using computer vision to detect wildfires. Explore the code, model, and results of our research on wildfire prevention.

Alaska Project Ideas, mentored by the researchers and collaborators of University of Alaska and supported by open-source entities and enthusiasts in Alaska.

Geospatials project meant to help generate burn severity maps & assess surface area affected by wildfires.

Wildfire risk assessment using remote sensing data - Prediction of Wildfires

A collection of modules to programmatically search for/download imagery from live cam feeds across the state of California.

End-to-end machine learning pipeline for the prediction of extreme and dangerous wildfires.

Utilizing Google Earth Engine and satellite imagery to identify wildfire occurrences.

National Forest FireBot: a Python script that scrapes incidents for any National Forest using WildCAD's WildWeb or WildWeb-E feature, and reports them to a desired output. Optionally there is a self-service SMS gateway end users can leverage

A Wildfire Detection System that integrates machine learning models with satellite imagery, camera feeds, and weather data to predict and detect wildfires effectively.

Xtinguish is an CNN Image Classfication model which helps in detecting and preventing Wildfires

Canada Wildfire Prediction Using Deep Learning

🎥🌲🔥 Improving wildfire smoke detection models by creating virtual fine-tuning data in Unreal Engine.

Machine Learning for Wildfire Detection - MLOps

Embedded Node Network that relies on node-to-node communication and internet to alert on wildfires.

Automated framework for retrieving and processing Sentinel-2 satellite imagery using the Sentinel Hub API. Focused on analyzing geospatial data, particularly SWIR composites, to visualize wildfire-affected areas and support environmental monitoring.

AI-based image classification for early wildfire detection using CNNs. Trained on a cleaned and augmented dataset to distinguish fire, smoke, and non-fire, achieving high recall and AUC. Supports real-time monitoring and emergency response applications.

end-to-end pipeline to predict next-day wildfire risk from NASA FIRMS (active fires) and Meteostat weather, train LightGBM, and visualize alerts in Streamlit.

A concise Google Earth Engine toolkit for wildfire monitoring and detection using Sentinel-2 imagery, featuring false-color composites, data annotation, and deep learning readiness.