Smart and precision agriculture requires deploying modern agricultural technologies based on high-performance machines, fertilizers, novel irrigation methods, new species, and hybrids of agricultural crops. These technologies are aimed at supporting the productivity and sustainability of modern agriculture. In this paper, we study the advantages of novel technologies and algorithms of data analysis for the early identification of plant infection. We consider the possibility of improving the agricultural monitoring using innovation system that combine unmanned aerial vehicles with machine learning technologies for the identification of areas of possible infection of wheat. We are focusing on the wheat pathogen that is largely distributed in many areas of the world and can lead to devastating epiphytotics. The block diagram that represents the learning stage of detection is developed and presented. Also, a prototype of a modular visual data analysis system was developed that implements the concept of simulation training based on preliminary classification by brightness characteristics. The system allows assessing the potential of simple image pre-processing algorithms before implementing full-fledged neural architectures or deep learning.