EcoMonitor

The Project

EcoMonitor is an AI-based project that incorporates various aspects of software and hardware with the goal of raising plants. The project consists of a device and app that work together to determine the health status of a chosen plant and notify the user with watering reminders dependent on the moisture level of the soil. With real-time feedback, EcoMonitor provides a seamless experience that guarantees a long life for one’s plants, as it utilizes a camera that sends information to the AI model, a moisture detector within the soil, and LED lights that change color depending on the health of the plant. This project was created because the challenge of raising plants is often underestimated. Myself and others have all had difficulty keeping houseplants alive for extended periods of time due to the responsibility and time required for such a small item, causing it to be easily overlooked throughout the mayhem of one’s day to day lives. Raising plants promotes a multitude of benefits, including improving one’s air quality to reduce risk of cardiovascular or respiratory issues, boosting mental health, and improving self-esteem. Some challenges found throughout this project included training the AI model and transferring data from the raspberry pi system to the app. To train the model, we initially attempted to find an existing database that included images of assorted houseplants, sorted into healthy, over-watered, and under-watered. Unfortunately, no dataset seemed to appear for the hydration of houseplants, so we had to manually find numerous pictures and sort them into each category, along with converting them all to the same file type and pixel size. In addition to training the AI model, we had to go through a third party database to store the information gathered by the AI. We decided to use firebase to store the information and transfer it to our app. Initially, our app was being programmed using flutterflow, which is a version of flutter (an app making program) that involved less code. However, because of this, it has less capabilities and could not connect to the data that we had to transfer to the app. Therefore, we changed programs to flutter, which even though it was more program intensive, allowed for the AI data to appear in the app for real-time changes. In the future, EcoMonitor hopes to expand further, including being able to detect different diseases that can affect plants, more specifically to be focused in the agricultural sector rather than private use. This can be utilized to be aware of crop diseases and improve efficiency when farming. Furthermore, we hope to have a wider dataset to train our AI model to be more accurate and consistent when identifying plant health. As our current dataset is just a collection of images from the internet, there aren’t as many pictures covering as many plants, leading to more variability and less consistency when running.

View this project on YouTube