Early Detection of Botrytis cinerea on Eggplant Leaves Based on Visible and Near-Infrared Spectroscopy

Abstract

Visible and near-infrared reflectance spectroscopy was applied to the early detection of Botrytis cinerea on eggplant leaves before symptoms appeared. Chemometrics methods were used to build the prediction model based on the spectral reflectance data. Owing to the complexity of the original spectral data, principal component analysis was executed to reduce the numerous wavelengths to several principal components (PCs) in order to decrease the amount of calculation and improve the accuracy. These PCs were set as input variables of back-propagation neural networks (BP-NN). The BP-NN model developed has an accuracy rate of 85% in predicting fungal infections. Furthermore, partial least squares regression was executed to obtain seven sensitive wavelengths. Based on these wavelengths, the prediction accuracy rate of the BP-NN model was 70%. This preliminary study, which was done in a closed room with restrictions to avoid intereference of the field environment, indicated that it is possible to apply spectral technology to the early detection of Botrytis cinerea on eggplant leaves. There is a potential to establish an online field application in early disease detection based on visible and near-infrared spectroscopy.