Demeter

Inspiration

We have all had the experience of having a houseplant die for no apparent reason. Gardening is a hobby that not only requires practice, but an immense amount of knowledge of what factors impact plant health. This knowledge barrier is what inspired Demeter. Demeter is designed to support those who either aren't as experienced with planting, or those who simply don't have the time to do the research. Because we believe everyone deserves to have a green thumb

What it does

Demeter is a two-part system, one being our daylight exposure sensor and the other our AI-powered webapp. The daylight exposure sensor measures the lux (unit of measure for light covering an area) in any area of interest. It collects this data over time at set intervals to paint a clear picture of exactly how much sun is in an area and at what times.

This data is then sent to our webapp, where it is sent alongside local forecast data collected from Google Maps' Weather Service and preferences provided by our user (i.e., plant type, watering level, indoor/outdoor planting) to Google's Gemini. Gemini then selects a plant that would thrive in the provided environment and satisfy the preferences of the user.

How we built it

The daylight exposure sensor uses an Arduino board and a photoresistor (a resistor whose resistance changes based on the presence of light). Through some calculations, we are able to approximate lux levels based on the level of resistance.

For the webapp, the frontend is created in React (Vite), the backend in Python (Flask), with MongoDB in the background storing data from the daylight exposure sensors. To collect local forecast data, we decided to use Google Maps' Weather Service API, and to ultimately analyze and recommend plants, we decided to use Google's Gemini AI models. Together, these components provide a fluid and dynamic user experience that is highly customized to each of their needs and situations.

Challenges we ran into

When we were imagining this concept, we hoped to be able to wirelessly communicate the data from the sensors to a computer that could upload it to the database. Unfortunately, this was not possible with the materials provided to us so we had to adjust our plans for data collection a little bit.

Another challenge was limiting the scope in terms of what data to collect. In gardening, there are hundreds of factors at play, and we could have used 5-6 different sensors and be able to make a good argument as to why they were necessary. To avoid scope creep, we decided to define the problem in terms of what was necessary to make an educated decision, rather than what information we would want to have at all times.

Accomplishments that we're proud of

One deadline we set early on was completing the sensor prototype before Saturday morning, so that we could spend a good portion of the day collecting real data to use on our webapp for our demo. We smashed that goal and were able to collect enough data to turn our vision into a proof of concept.

What we learned

From this project, we learned how important it is to have a concrete vision of what the end product should be. There were many points where we could have easily gone off the rails and added tons of irrelevant features that failed to address the problem we wanted to solve. It was our ability to tune out scope creep and keep each other focused that allowed us to finish this project before deadline.

What's next for Demeter

From here, we would want to look into shrinking the size of the sensor, looking into wireless solutions, and maybe even making a custom 3D-printed chassis.