Abstract
Coffee rust, caused by the fungus Hemileia vastatrix, is a fungal disease that affects coffee production and quality, so its early detection is crucial to prevent massive outbreaks and protect production. This article analyzes the most effective factors, the algorithms used, the accuracy of the models, and the challenges in the detection of coffee rust, through an exploratory systematic review of 35 empirical articles obtained from Scopus, IEEE Xplore and SciELO. The review identifies that the most determinant factors for detection include humidity, temperature and the presence of shade. The most commonly used algorithms are Convolutional Neural Networks (CNN), Support Vector Machines (SVM) and Random Forest, highlighting CNN for its ability to process and analyze images with an accuracy of 99.57%, followed by Artificial Neural Networks (ANN) with 98% and SVM with 96%. However, it is concluded that challenges remain such as the need for high quality labeled datasets, variability in environmental conditions and implementation costs. This study provides a comprehensive overview of recent advances and areas for improvement in coffee rust detection, providing information for researchers, practitioners and decision makers in the agricultural sector.