Demeter

Inspiration

We were deeply moved by 2021 famines in Africa and decided to attempt to be part of a solution to world hunger. There is currently enough food on the planet to feed everyone and no one should have to starve. We want to inform farmers and predict famines before they take lives.

What it does

Our project collects data on soil moisture and temperature through a DHT11 sensor to help farmers realize what needs to be fixed to improve crop yield. We visualize this data through flask onto a website with Chart.js integrated. This allows the data to be seen easily and accessible to all. Finally we run our neural network on the data we collect to let the farmer know the best crop to grow in that area and potential yield rates.

How we built it

We built Demeter with two Arduino boards, two DHT11s and a whole lot of code. Our Arduino board takes in the direct data from the DHT11s and uses serial communication to send the data to our python file. The python file reads the Arduino serial monitor and processes it into a readable form and graph visualization using matplotlib, putting it on our website. Run our neural network based on data collected to find the optimal plant for that spot.

Challenges we ran into

We couldn't find the right training data for the neural networks we wanted to make. We wanted to get data that has humidity, temperature, and yield but unfortunately we couldn't find anything with those parameters.

Accomplishments that we're proud of

We were proud of the neural network we made, which takes in give soil health metrics and outputs a optimal crop to grow. We are also proud of using matplotlib to display data in a user friendly graphs and visualizations. In addition, we are proud of creating a website that changes in live time to data from sensors. We are also proud of creating an idea that can have a real impact.

What we learned

We learned how to use scikit-learn to create neural networks, how to use flask to create a live changing website, and how to use matplotlib to show data in a user friendly graph.

What's next for Demeter

We have a couple more things we could work on for Demeter: adding more sensors into our Arduino like soil nutrient sensors and more, use new soil readings to diagnose diseases in the plants, finding/making a dataset + neural network to test for droughts by observing moisture levels from our data, updating our user interface and making a mobile version. We eventually want to have sensors spread throughout vast areas to predict droughts and forest fires in a state and country-wide school.