Random Forest Classification in Cocoa Pods Desease

Cocoa, scientifically known as Theobroma cacao, is a tree-like crop that has its origins in South America. Cocoa is a significant plantation product that serves as a crucial source of revenue and foreign currency. However, cocoa production had a sustained decrease from 2012 to 2017. The cocoa production in 2012 amounted to 740,500 tons. However, in 2019, the quantity is limited to just 659,776 tons. Pests and illnesses may lead to a decline in cocoa production. Given the advancements in technology, it is feasible to develop a system capable of categorizing diseases detected in cocoa pods. This study employs the Random Forest algorithm for the classification procedure. Prior to the classification procedure, image processing is performed using the grayscale approach, followed by image segmentation using the inverse binary threshold and closure techniques. In addition, the color extraction is performed using the hue saturation value (HSV) approach, while the texture extraction is done using the gray level co-occurrence matrix (GLCM) method. The texture extraction process incorporates dissimilarity, contrast, correlation, ASM (Angular Second Moment), energy, and homogeneity as key properties. The cocoa diseases include pods rot, anthracnose, helopeltis, and borer. There are a total of 360 data points, which are split into 80% training data and 20% testing data. According to the test findings in this research, the system can accurately diagnose the different kinds of diseases with a precision of 94.4%.