Detection of spotted wing drosophila hidden infestation in blueberry using hyperspectral imaging and machine learning
2025
Priyanka Dahiya | Christopher Kucha | Ebenezer Olaniyi | Allison Niu | Ashfaq Sial
Spotted wing drosophila (SWD) significantly threatens blueberry quality and safety by infesting fruits internally, making detection difficult using traditional methods. This study evaluated hyperspectral imaging and machine learning for detecting SWD infestation in blueberries. Blueberries were collected and divided into two groups: one infected with SWD and the other uninfected. A hyperspectral imaging system (900–1700 nm) was used to acquire images from both groups over six days. The images were processed, and the extracted spectral information was preprocessed using a combination of Savitzky-Golay first derivative and standard normal variate methods. The preprocessed data was utilized to build and compare classification models for the infected and uninfected blueberry groups using six machine learning classifiers. Support Vector Machine achieved the highest accuracy (87.99 %), followed by Linear Discriminant Analysis (87.18 %) and Partial Least Squares Discriminant Analysis (86.21 %). The study shows hyperspectral imaging can effectively distinguish between SWD-infected and uninfected blueberries.
Show more [+] Less [-]AGROVOC Keywords
Bibliographic information
This bibliographic record has been provided by Directory of Open Access Journals