Machine Learning Applied in Identification of Anthracnose and Stem End Rot in Mangoes-Alphonso Cultivar

Computer Vision systems are essential in food inspection lines across packaging industries. High speed and accuracy in sorting fruits are necessary. Such systems are modeled in automated systems. Identifying fruit defects is essential in assessing fruit quality. Vision systems play a crucial role in mango processing industries since India contributes majorly to the export of mangoes worldwide. Defects such as Anthracnose and stem end rot are similar when processed visually through the naked eye. Differentiating them is vital as adequate agricultural practices can be induced at every stage of the development of the fruit. In this work, images of defected mangoes were collected during the season. The images were preprocessed, GLCM features were extracted, and classifiers such as SVM and KNN were applied. The system has achieved an average of 96 %and 94 %accuracy on the two classifiers SVM and KNN for Anthracnose; 96 %and 94 %accuracy with SVM and KNN two classifiers for Srem End Rot.