Skip to main content
Fig. 3 | Plant Methods

Fig. 3

From: A comparison of ImageJ and machine learning based image analysis methods to measure cassava bacterial blight disease severity

Fig. 3

Overview of the Support Vector Machine learning segmentation and analysis method. A Images of cassava leaves infiltrated with Xam WT, XamΔTAL20, and mock treatments were segmented and analyzed using a support vector machine learning tool. Images depict steps used to generate a classifier training mask for the machine learning tool. A mask was made by combining representative CBB infected images into one graphic and generating a binary mask in ImageJ. White lines showcase a representative water-soaked lesion within the combined leaf graphic and indicate changes at each step. The mask was used to generate a classifier (YAML) file with PhenotyperCV. B Images depict steps of machine learning processing using a CBB infected cassava leaf image. Images were uploaded into the machine learning tool and processed by gray balance color correction, thresholding, and the inoculated regions of interest were selected and labeled using a color code: Red = Xam WT, Green = XamΔTAL20 and Blue = Mock. White lines showcase a representative water-soaked lesion within the image and indicate changes at each step. C Images exhibit outputs from the machine learning image processing and include the color corrected image (left), a pseudo-colored map of the pixels classified as water-soaked (middle), and a feature prediction image (right). White lines showcase a representative water-soaked lesion within the image and indicate differences in each output image. Text separated files with shapes and color data for each inoculation spot were also generated

Back to article page