Hyperspectral imaging for pest symptom detection in bell pepper.

Area of Science:

• Agricultural Science
• Remote Sensing
• Plant Pathology

Background:

• Automated pest monitoring is crucial for integrated pest management.
• Hyperspectral imaging (HSI) is an emerging technology for detecting plant stress and pests.
• Automated image analysis holds potential for practical pest monitoring.

Purpose of the Study:

• Investigate HSI for noninvasive identification and distinction of bell pepper plants infested with Myzus persicae and Frankliniella occidentalis.
• Evaluate the use of a modified spraying robot as an autonomous platform for HSI data acquisition.
• Develop and train a decision algorithm for pest detection using HSI data.

Main Methods:

• Utilized HSI across a 400-2500 nm spectrum on bell pepper plants.
• Infestations with M. persicae and F. occidentalis were conducted under netted conditions.
• Trained an XGBoost algorithm using images of single plants, dissected leaves, and whole plants.
• Adapted a spraying robot into an autonomous platform for greenhouse data collection.

Main Results:

• Specific wavelengths correlated with insect damage patterns.
• M. persicae and F. occidentalis were distinguishable from each other and healthy controls on single leaves and small whole plants under confined conditions.
• Pest detection under greenhouse conditions showed a lower fit compared to manual monitoring.

Conclusions:

• HSI effectively distinguishes sucking pests on bell pepper leaves and plants under controlled environments.
• Wavelength reduction techniques enable multispectral camera application in commercial greenhouses.
• Automated platforms show promise, but algorithms require further refinement for real-world greenhouse pest detection.