Early detection of grapevine diseases using pre-trained Convolutional Neural Networks

Abstract

This paper proposes to apply pre-trained Convolutional Neural Networks (CNN) for the early detection of two common grapevine diseases: peronospora and o´ıdio. These diseases present similar symptoms and are of great viticultural importance. Our objective is to train a CNN using transfer learning techniques to accurately detect the presence of early symptoms of the diseases under study. To achieve that, we’ll design a pipeline that starts with data acquisition in the field and finalizes with the early disease identification, including class definition, labeling, image preprocessing and training process of the CNN, employing edge computing-based service computing paradigm to overcome some inherent problems of traditional mobile cloud computing paradigm.