Application of machine learning methods for the differentiation of fungal diseases in strawberry based on hyperspectral image analysis

Fungal diseases have a significant negative impact on strawberry yield. Their detection and differentiation using hyperspectral measurements is possible alternative to traditional methods. In this study, leaves infected with Ramularia Tulasnei, Marssonina potentillae Dendrophoma obscurans visible symptoms of the disease were used for analysis. The reflection spectrum was recorded Photonfocus camera (wavelength range 475–900 nm, 149 channels) under laboratory conditions line scanning method. This research has aimed compare four machine learning methods: spectral angle mapper (SAM), support vector (SVM), k-nearest neighbors (KNN) linear discriminant analysis (LDA). Classification models built based full spectrum, as well 12 vegetation indices (VI) features. results demonstrated that SVM model spectra reached highest classification accuracy 94%. KNN performed slightly worse 91% accuracy. performance VIs lower than an 78–85%.