Vision-Based Edge Computing for Autonomous Plant Phenotyping and Resource Management

Author: Theresa Murphy

Field of Study: Mechanical Engineering

Faculty Mentors: James Pitarresi

Easel: 97

Timeslot: Midday

Abstract: Most indoor gardening relies on simple timers, but plants are living things that don't always follow a schedule. This project replaces guesswork with a system that watches and reacts to a plant's behavior in real-time. Using an NVIDIA Jetson Nano and a standard webcam, the system acts as a digital eye, specifically trained to spot the droop of a leaf, which is the first sign that a plant is thirsty. By combining this visual intelligence with sensors for humidity and light, the Jetson takes full responsibility for the plant's environment. It doesn't just water on a timer; it triggers a precision pump only when it "sees" the plant is stressed and manages an LED grow light to ensure perfect growth. This research moves away from static automation toward a truly responsive, hands-free way to grow food and flowers in urban spaces, making professional-grade care accessible to everyone.