Akvelon | Fighting Wildfires with AI: Akvelon Ranks in Top 10 of Open Source AI Competition
Akvelon’s team recently developed an application that uses #AI to predict and prevent wildfires, placing them in the Top 10 of the Wildfire & Bushire Challenge hosted by @h2oai
AI, Artificial Intelligence, App, Application, Machine Learning, AI-driven applications, Data, Wildfires, Competition
36165
post-template-default,single,single-post,postid-36165,single-format-standard,ajax_fade,page_not_loaded,,qode-theme-ver-8.0,wpb-js-composer js-comp-ver-6.8.0,vc_responsive,elementor-default,elementor-kit-34246
 

Fighting Wildfires with AI: Akvelon Ranks in Top 10 of Open Source AI Competition

12 May Fighting Wildfires with AI: Akvelon Ranks in Top 10 of Open Source AI Competition

Akvelon’s engineers are constantly looking for ways to challenge their technical skills beyond the work day, often competing in competitions across technologies in their free time. Recently, one team of Akvelon software engineers participated in the Wildfire & Bushire Challenge hosted by H2O.ai. Participants were tasked with developing open source AI applications that build wildfire and bushfire prediction models. 

Challenge Goal

The H2O.ai Fights Fire Challenge is a global collaborative effort to improve the world that we share, and  aims to create new AI applications that can be used to help save lives, nature, and property from wildfires and bushfires. This problem is incredibly relevant these days, because wildfires  threaten lives, communities, wildlife, and forests around the world every year. With the impact of global climate change, it is only getting worse. 

Key Objectives

The participants needed to develop open source AI applications that help in some way with the following:

  • Predicting the behavior of wildfires
  • Predicting where wildfires will start
  • Reducing the loss of life and property from wildfires

Akvelon’s team developed an application that placed them in the top 10 of all competitors, ranking in 7th place.

Powering the Advantage

For the challenge, Akvelon’s team prepared models based on weather and additional satellite features. They used a stratified split by region and fire occurrence, and finished with LGBM and Catboost blending. 

Judges for the competition required that all teams explain their prediction, so Akvelon used the shape values for this task to provide compelling visualization of the most important features and how it influenced the final prediction. 

Akvelon’s AI solution was able to predict fires all around the world and visualize them on a map. The probability of fires occurring was also provided, making it possible to understand which fires will occur with the highest probability.

Akvelon’s auto-updated AI system constantly monitors weather and satellite data and provides fire predictions with a color visualization on an interactive map. You can see how it works in the picture above.

Results of the Competition

Beyond ranking in the top 10 of the Wildlife & Bushfire Challenge, Akvelon’s team is proud to report that H2O.ai is sharing the applications developed by our team and other participants with numerous organizations that are working to fight wildfires and bushfires around the world. The open source AI application that we developed is being used by first responders, local leaders, businesses, and the public to prevent and contain wildfires, helping to reduce the loss of lives, forests, resources, and property. 

 

 

More Insights

Developing Intelligent Apps with Spice.AI

Akvelon Machine Learning Engineer has prepared an overview of a new platform, Spice AI, that has been designed to make the process of creating intelligent applications much easier. Read more here

Akvelon’s Approach to High-Resolution Face Attribute Manipulation

Our app development team has recently developed even more tools for manipulating images of people’s faces on a deeper level. Read more here

Akvelon Develops Auto Comment, an AI-powered Code Documentation Extension

A closer look at Akvelon’s Auto Comment, an AI-powered Code Documentation Extension. Read more here



Akvelon

Akvelon

We Make BIG Software Happen. Contact Us