Classification of Coffee Disease: A Systematized Approach Based on Chronicity and Transfer Learning

The most sustainable tropical crop in terms of export revenue is coffee, which additionally acts as a major source of revenue for the nations where it is grown. As a crucial agricultural commodity in the global market, it is essential to increase coffee production while maintaining quality. To meet global demands, farmers need to expand and effectively manage coffee production. However, various diseases can reduce the productivity and quality of coffee plants, posing significant challenges. Detection in early stage of these diseases is crucial to mitigating their impact. Despite frequent contributions to research aimed at enhancing the coffee industry, most coffee farming is carried out by smallholder farmers who lack access to advanced resources and high-tech equipment. Coffee bean quality is highly sensitive to diseases caused by fungi, bacteria, insects, and environmental factors, all of which can significantly affect the financial success of the entire production chain. This study our goal is to accurately classify and detect coffee leaf diseases using images. Hence we used some Transfer Learning models like CNN, Resnet-50, VGG-16 and Efficient net. The accuracy of the Efficient Net is 93.12% which is 1.18% and 4.84% and 2.61%superior to Resnet-50, CNN and VGG-16 models.