Lettuce Root Development and Monitoring System Using Machine Learning in Hydroponics

Roots of a hydroponically grown lettuce are usually susceptible to waterborne diseases like Pythium disease (root rot). Visual inspection of a plant root is needed in a hydroponic system to determine the health status of the plant. However, removing the plants in its growing environment will stress...

Full description

Saved in:
Bibliographic Details
Main Authors: De Los Santos, Bryan B., Don, Anna Grace M., Gotengco, Angelito A., Millena, Jose Luis M., Romero, Rhaizel Gel C., Agustin, Eugene V., Beano, Mary Grace P., Mandayo, Ericson A., Medina, Oliver A., Sigue, Anna-Liza F.
Format: Conference Proceeding
Language:English
Subjects:
agriculture
Decision Tree
Hydroponics
image processing
lettuce
Machine learning
Pythium disease
root monitoring
Temperature distribution
Temperature measurement
Temperature sensors
Visualization
Water heating
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Roots of a hydroponically grown lettuce are usually susceptible to waterborne diseases like Pythium disease (root rot). Visual inspection of a plant root is needed in a hydroponic system to determine the health status of the plant. However, removing the plants in its growing environment will stress the plants that can cause stunted growth, and is susceptible to further complications once the roots are exposed outside the environment. Also, studying the development of the lettuce roots on its freshness could help to set standards in determining its shelf life and could lead to a longer life. The project intends to develop an automated root monitoring system for detecting root rot and changing water temperature to avoid extreme heat that damages the root system, affecting the entire plant. The Peltier module was employed to keep the water temperature stable between 18°C - 26°C. If the water temperature exceeds this range, the system will react based on what is best for the plant. The researchers used a camera to monitor the roots across the three grow boxes for the root detecting method. The researchers also implemented image pre-processing, specifically cropping and resizing, to improve image quality in preparation for image processing, and the Decision Tree method for machine learning. The model detected root rot with 91 % accuracy. The system was able to accurately identify and locate roots as well as predict their health status.
ISSN:2159-3450
DOI:10.1109/TENCON55691.2022.9977465