Image classification-based system for disease diagnosis in coffee crops: A case study in the province of Rodriguez de Mendoza, northern Peru

Coffee is an important product in the world economy and its productivity and quality are affected by diseases such as Rust (Hemileia vastatrix), anthracnose (Colletotrichum kahawae) and rooster’s eye (Mycena citricolor), in this sense, the objective of this research was to determine the level of accuracy of a system based on image recognition for the diagnosis of diseases in coffee crops, which will help in the timely detection of diseases. Convolutional neural networks (CNN) were used, which have demonstrated their efficiency and accuracy in the field of image classification and pattern recognition, based on Artificial Intelligence (AI), which allowed the recognition of diseases in coffee crops through a mobile web application. As a result, high accuracy was obtained to detect anthracnose, rooster’s eye and rust diseases (92%, 94%, 91%) at an overall accuracy of 92%, an F1 score of 0.92%. With this help, coffee growers can identify diseases more quickly, increasing coffee production. In this sense, it is of great importance to implement future projects, incorporating participatory research methods for its massification that will allow the diagnosis of coffee diseases in real time.