Machine Learning Classification and Segmentation of Forest Fires in Wide Area Motion Imagery

Abstract

Numerous models and simulations exist for characterizing and predicting wildland fire behavior. The U.S. Forest Service (USFS) and other organizations have devoted decades of research to identifying the parameters that affect fire movement, rate of spread, and direction of spread across geographic terrain. While this research is invaluable to the firefighting community, due to computational constraints, these models do not run in real time or against imagery at time-of-collect, and therefore do little to assist the firefighter and first responders on the ground during a wildland fire event. We present the first part of a multi-step automated computational methodology to characterize fire behavior and rate of spread in real time across any geographic terrain. This first step is the classification and segmentation of the wildland fire in Wide Area Motion Imagery (WAMI) using Machine Learning (ML) methods. The continuation of this research, detailed herein, will involve training a more robust, purpose-built Recurrent Neural Network architecture incorporating many of the parameters the USFS has been studying for decades. The goal of this research is to deploy models using ‘lite’ ML frameworks on edge devices, mounted on collection platforms for real-time decision support for firefighting operations as imagery and other data are being collected during a wildland fire event.