Related Experiment Video
Updated: Jun 15, 2025
05:03
Remote Sensing Evaluation of Two-spotted Spider Mite Damage on Greenhouse Cotton
Published on: April 28, 2017
8.5K
Machine learning-based hyperspectral wavelength selection and classification of spider mite-infested cucumber leaves.
Boris Mandrapa1, Klaus Spohrer2, Dominik Wuttke3
1Institute of Agricultural Engineering, Tropics and Subtropics Group, University of Hohenheim, Stuttgart, Germany. boris.mandrapa@uni-hohenheim.de.
Experimental & Applied Acarology
|August 23, 2024
Summary
Machine learning accurately detects two-spotted spider mites in cucumbers using hyperspectral imaging. This method offers a promising, practical approach for early pest identification in greenhouse agriculture.
Area of Science:
- Agricultural Science
- Plant Pathology
- Computer Science
Background:
- Two-spotted spider mite (Tetranychus urticae) is a significant greenhouse pest, causing substantial crop damage, particularly in cucumbers.
- Spider mite feeding alters leaf spectral properties, creating potential for optical detection methods.
Purpose of the Study:
- To apply machine learning methods for the detection of spider mite infestations in cucumber leaves using hyperspectral data.
- To identify effective wavelengths and assess the accuracy of machine learning algorithms for classifying healthy and infested leaves.
Main Methods:
- Collected hyperspectral data from cucumber leaves under controlled conditions.
- Employed three feature selection methods to identify effective wavelengths.
- Utilized three supervised machine learning algorithms to classify leaf infestation status.
Main Results:
- Achieved classification accuracy exceeding 80% for all feature selection and classification method combinations.
- High accuracy was maintained even when using a reduced set of ten or five wavelengths.
- Demonstrated the effectiveness of machine learning in distinguishing between healthy and spider mite-infested cucumber leaves.
Conclusions:
- Machine learning is a powerful tool for image-based detection of spider mites in cucumbers.
- The identification of a limited number of effective wavelengths suggests significant potential for practical, cost-effective applications in pest management.
