Fig. 7

Classification accuracy of intact bunches depending on random forest classifier complexity. Boxplot of the predicted surface area affected for the three main categories of the experiment based on pixel-wise segmentation of LDA projected hyperspectral images (VNIR only). a Pixel-wise pure spectral classification with Random Forest, b texture-based spatial-spectral segmentation with 10 trees versus c Random Forest with 50 trees. Severely diseased bunches can be detected with high accuracy, while discrimination between healthy and infected is challenging in a few cases. Classification accuracy increases with the complexity (number of decision trees) of the Random Forest classifier. Results of the analysis of hyperspectral images are comparable and correspond well to qPCR results (see Fig. 4)