hjjl

The application of machine learning in plant disease detection has gained tremendous momentum in precision agriculture over the last few decades. This can be attributed to the catastrophic implications of diseases, and to the availability of higher–quality measurements coupled with modern algorithms and the increased possibility to fuse multiple sources of images into a dataset. These images can either be from satellite imagery, sensors, or cameras positioned in fields. Image processing algorithms together with machine learning classifiers are the most viable and widely used automated methods to examine, identify, and classify plant leaf disease visual symptoms by performing analysis on the colored images. Segmentation is a key essential process in this context and basically involves two stages: The first stage involves separating the leaf (foreground) from other surrounding parts (background) of the image. The second stage, which is also the most challenging, involves separating the disease symptom region (s) from the leaf. This second region is normally referred to as the region of interest or ROI from which feature parameters used to train machine learning classifiers are extracted. Various approaches have been developed through computer-aided segmentation categorized as: