Wildfire Classification

• 

  One of the computer generated labels. The one shown specifically is of smoke. The computer is more accurate because of all the dots.

• 

  One of the computer generated labels. The one shown specifically is of fire. The computer is more accurate because of all the dots.

• 

  These are the computer generated labels. This model got 100% of the wildfire correct

• 

  The index part of the website

Inspiration

The Into the Inferno and the ML Workshop with Peter Griggs helped to inspire us. We also used what we learned from the React Website to build the site. We thought it would be cool to help with the different wildfires going on. We saw how much damage went on because of the fires in California and Australia, and we wanted to maybe help with that.

What it does

Our project has 2 parts - a website and a deep learning model. The website is a great place with many links to learn about wildfires, and also our deep learning model. The model itself takes in satellite images of wildfires, and it classifies the different parts. It can use 3 labels to classify it - Fire, Smoke, & Fire/Smoke Mix.

How we built it

We used repl.it to build the website. We used a Next.js and coded it in React. To build the model, we used Zendo. All images were taken from Google Images.

Challenges we ran into

At first our model classified everything as fire, and it didn't even do so correctly. So, we had to make there be several different labels, and also we had to upload and label many more photos for the training sets. We could not upload images through the react website, so we had to make a seperate html website and link them together to get the website to work better.

Accomplishments that we're proud of

We got a functional and running website! It was very cool to have that. Also, we made our first deep learning model. It also feels good that maybe our project will help someone.

What we learned

We learned 2 brand new softwares (Zendo & Repl.it), and 2 new programming languages (html, react). We made our whole project doing new things that we learned either on our own, or in workshops.

What's next for Wildfire Classification

We will keep training the model to help make it more and more accurate. This will allow for peak image classification. Also, we will continue to work on and improve the website. This will allow for a better UI experience. We will also register a domain to help with getting our website out there.

Workshops

ML Workshop with DeepAI - Discord: Collin#0030 ALL EYES on the INFERNO - Discord: Collin#0030 Intro to React - Discord: Collin#0030

Built With

• javascript
• next.js
• react
• repl.it
• zendo