Spill

Empowering sensor data for mapping with analysis for a brighter and more sustainable future!

Comment

Inspiration

The inspiration for this project started when we initially did our research into how impactful gas leaks can be to the environment. These natural gases not only are valuable reason but also highly explosive. In order to help with this project, we looked to ML models in order to try and detect leaks.

What it does

This project is built two-fold, there is a web app portion that seeks to inform the directors about the problem through why it matters. The end of it takes in a file of sensor data and can be used to predict where gas leaks originate from to explain where the problems lie.

How we built it

For our web app, we used React with Tailwind CSS in order to build a fully functioning web app that appeals to the users as well. On the backend, we used Python with an assortment of libraries such as pandas, sklearn, and geopandas in order to build and map our model. We also used Flask in our backend as a connection.

Challenges we ran into

Our greatest challenge was that we had skilled individuals on both the front end and the back end but we lacked critical knowledge in the middle section, the connection of the front end and the back end. We fixed this issue by learning how to use Flask through the use of tutorial videos and StackOverflow.

Accomplishments that we're proud of

One of our accomplishments that we're extremely proud of was that our model on the classification scale has a Root Mean Squared Error (RMSE) of just .165%

What we learned

We learned a lot of things through this project including:

  • React Rerouting
  • How to use various models utilizing APIs
  • Committing is very important
  • Stylizing with Tailwind CSS
  • Communication is key

What's next for Spill

Right now, Spill is scalable to the factory given to us. In the future, we plan on using the temperature of the surrounding areas along with potentially taking in the rain and other natural causes in order to predict the gas leaks. If this works, then we can apply our model to our factories to gather an even better prediction model!

Built With

Share this project:

Updates