-
Dashboard
-
Analytics Summary
-
Analytics Comparison
-
Prediction Map
Inspiration
The recent wildfires in California have showcased the devastating impact of wildfires on communities, and the environment. This inspired us to take action and apply modern solutions to wildfire prediction and resource deployment optimization for Québec. It's time to find the fire!
What it does
On FireFind website, we showcase the wildfire dataset based on their severity. Then, we combust an algorithm to optimize the resources spent fighting the fire: One optimized for time efficiency, the other for cost efficiency. We also train and deploy our own Neural network AI to predict potential wildfire.
How we built it
React.js & Node.js with Tailwindcss for the website. Python Jupyter for NN AI training and testing.
Challenges we ran into
Training on the dataset was difficult because there are 99% no fire occasion and we were overfitting our models for the given dataset. In order to fix it we have to filter our data a lot. It was our first time doing front-end so the project was literally on fire.
Accomplishments that we're proud of
Interactive map. AI model. And just chef kiss UI design.
What we learned
The forefront of front-end programming, since most of us are back-end developers. Manipulating dataset to avoid extreme cases and optimization.
What's next for FireFind
We found the fire, so we want to really improve our optimization in order to apply in more practical situation. Maybe as a Control system over IP, or as a scheduler. Guess it's time for FireFight huh.
Log in or sign up for Devpost to join the conversation.