-
-
Displays the weather at that specific location to give more context to understand the condition of the plants
-
Displays the graphs over time and averages, as well as Gemini-powered insights.
Inspiration
Plants are essential to the ecosystem, yet their health is often difficult to assess—especially at scale, such as across a large campus.
What it does
The system uses sensors and a camera to evaluate plant health and detect potential diseases. Data is sent to a Google Cloud–hosted web app that displays a dashboard with temperature and deformity metrics, time-series graphs, and a map showing local weather conditions for additional context.
How we built it
A Raspberry Pi connected to sensors and a camera handles data collection and runs a lightweight classification model. The results are sent to a Flask-based backend hosted on Google Cloud, which stores data in a Snowflake database.
The frontend is built with HTML, CSS, and JavaScript, with Chart.js powering visualizations. The backend also sends plant data to the Google Gemini API to generate analyses, and supports a chatbot that provides plant care recommendations. Weather data from the OpenWeatherMap API is integrated to give environmental context. The entire application is containerized using Docker.
Challenges we ran into
API integration caused intermittent issues. We were unable to obtain a light sensor, limiting environmental data. The Raspberry Pi initially had connectivity problems, which required numerous tweaks to the Linux operating system.
Model development was also challenging: balancing size and accuracy for on-device inference required multiple iterations, and early versions suffered from overfitting.
Accomplishments that we're proud of
We successfully combined hardware and software into a cohesive system. The interface is clear and usable, and the model performs effectively despite running on constrained hardware.
What we learned
We learned how to integrate physical sensors with a full-stack web application, optimize models for edge devices, and coordinate multiple APIs within a single pipeline.
What's next for HoosLeaf.fit
Next steps include adding more sensors (such as light), deploying additional Raspberry Pi units for broader coverage, and improving model capability with more powerful hardware to detect a wider range of plant diseases.
Log in or sign up for Devpost to join the conversation.