Detection of Bacterial Infection in Melon Plants by Classification Methods Based on Imaging Data

Area of Science:

• Plant pathology
• Agricultural science
• Biotechnology

Background:

• Dickeya dadantii causes significant economic losses in global crop yields.
• Infection by D. dadantii in melon leaves leads to necrotic spots, chlorotic halos, and tissue necrosis, with symptom severity depending on bacterial dose and time.

Purpose of the Study:

• To investigate the use of advanced imaging techniques combined with machine learning for early and accurate detection of D. dadantii infection in melon leaves.
• To evaluate the efficacy of these methods for potential application in plant breeding and precision agriculture.

Main Methods:

• Utilized variable chlorophyll fluorescence, multicolor fluorescence, and thermography to capture spatial and temporal data on metabolic and stomatal alterations in infected leaves.
• Applied machine learning algorithms to analyze numerical data from imaging techniques for disease detection.

Main Results:

• Mathematical algorithms achieved high accuracy (96.5–99.1%) in classifying mock-infiltrated versus bacteria-infiltrated areas.
• Models trained on infiltrated areas demonstrated high performance (up to 96% accuracy) in classifying whole diseased leaves.

Conclusions:

• Machine learning analysis of imaging data provides a reliable and scalable method for detecting D. dadantii in melon leaves.
• This approach holds promise for integration into plant breeding programs and precision agriculture for disease management.