Ensemble Classification and Feature Extraction Based Plant Leaf Disease Recognition

Abstract

Plants are a significant source of energy for both humans and animals. Plant leaves are the principal means by which plants communicate with the earth's atmosphere. As a result, it falls on academics and academicians to investigate the situation and devise strategies for identifying disease-infected leaves. This will enable farmers all across the world to take prompt action to prevent their crops from becoming irreparably damaged. Detecting diseases by hand may not be the best option, therefore a robotic technique to discover leaf illnesses could be a boost to the agricultural industry while also increasing crop productivity. This work focuses on using ensemble classification in conjunction with hybrid Law's mask, LBP, GLCM to improve classification results. The suggested method demonstrates that an ensemble of the chosen classifiers can outperform individual classifiers. Because ensemble classification has shown to be more accurate, the features used are also important in achieving the best results. The studies were carried out on the PlantVillage dataset's sick leaf photos of bell pepper, potato, and tomato.