Digital innovations in grain production: methodological principles of using artificial intelligence technologies

The paper discusses an innovative approach to monitoring grain ecosystems, based on the use of neural network technologies. A classification of the tasks of monitoring the phytosanitary condition of crops in a grain field has been carried out, and the corresponding intellectualization tools have been identified. The main attention is paid to the problems of detection, classification and development of diseases in crops, for the effective solution of which it is proposed to use computer vision methods, including a complex of convolutional architectures GoogleNet, DenseNet, U-Net, which have shown high performance in classification and segmentation problems on test samples of images of wheat diseases, obtained as a result of three years of field experiments in the Krasnodar Region. The results of the study show that the use of neural network methods in the process of monitoring grain ecosystems contributes to the effective solution of complex problems associated with diagnostic procedures, allowing to reduce the level of uncertainty in the decision-making process, which is especially important under the influence of environmental factors with a high level of randomness and variability. The main barrier to the intellectualization of production processes is the lack of methodology for working with artificial intelligence, big data and other digital technologies at different levels of management in the agricultural sector of the economy, which affects not only issues of technical implementation and implementation of artificial intelligence, but also organizational aspects, including work with data, staffing the intellectualization process, information infrastructure, defining the roles and responsibilities of participants in the process, as well as the integration of intelligent solutions with the agricultural solutions module of the national platform “Digital Agriculture”.