Real time fuel classification from drone imagery for improved wildfire prediction

Wildfires are naturally occurring phenomena that are necessary for the health of boreal forests. However, fires can cause infrastructural damage and loss of life if they spread into places where people, towns, and other assets are at risk. Due to the impact of wildfires on society, attempts have been undertaken to assess wildfire potential, spread, and magnitude, as well as to anticipate their behavior after ignition. Predicting wildfire risk is essential when making forest management decisions and selecting the most effective wildland fire response tactics to mitigate their negative effects. This project aims to develop user-friendly, cost-effective software to enhance the capabilities of existing fire management software (developed by Fire AI) utilizing artificial intelligence systems that have already demonstrated their effectiveness and performance. This project will allow for the automatic identification of hazardous fuels in near real time around active wildfires.

Faculty Supervisor:

Jeff Boisvert

Student:

Partner:

Fire AI

Discipline:

Computer science

Sector:

Agriculture

University:

University of Alberta

Program: