Ignis Vigilante

• 

  Home Page - Logo

• 

  Density Graph - Standard Variable Deviation

• 

  Correlation Map - Changes of Relative Variables

• 

  The hardware with the database

Inspiration

Forest fires have caused massive devastations during the last 5 years. Recently, Australia has been suffering from fires burning since October 2019. We wanted to create a hack that involves both the tool to collect the data and analyze it.

What it does

The sensors connected to the Arduino would send out statistical data using IoT protocols such as the MQTT. This data would be stored in our Firebase Cloud Firestore database which would be updated in realtime. We would then run our Machine Learning model to analyze the data and predict the likelihood of the area catching fire to prevent it. All this data would be tracked on our website for the citizens and the government of the country.

How we built it

Our hardware side uses an Arduino with a Grove shield. A Grove Temperature sensor v1.2 to track the temperature and a SparkFun Moisture sensor to track moisture content at a moderate depth. The collected data is read by server-side python code which adds all the collected data to our Firebase Cloud Firestore database.

The collected data is then run through our Machine Learning model which is trained using the sci-learn kit. The model outputs the likelihood of a fire taking place in an area based on the temperature, wind, relative humidity, etc. All this data is displayed on our web application through charts and heatmaps. The web application is built using Python and the Flask web framework.

Challenges we ran into

Training the model was our biggest challenge, as we didn't have a big dataset set to work with. We had to tweak our calculations to get better prediction accuracy based on the dataset.

Accomplishments that we're proud of

What we learned

What's next for Ignis Vigilante

Built With

• css
• firebase
• flask
• html
• machine-learning
• python