Computer vision for monitoring and accounting Pyrenophora teres of winter barley

I.V. Arinicheva, G.V. Volkova, Ya.V. Yakhnik, I.V. Arinichev

Abstract. The traditional practice of diagnosing a disease and determining the economic threshold of harmfulness is based on visual assessment. But it is difficult and requires highly qualified specialists, whose timely departure is not always possible, especially to small farms. A fundamentally new and extremely promising approach to diagnose the development of net leaf spot of barley is an approach based on automatic (without the participation of a human expert) recognition of the pathogen and the degree of its development from an image or series of images. The article proposes the use of an innovative approach to diagnosing the development of net blotch (Pyrenophora teres) of winter barley, based on advanced computer vision technologies. This approach involves a two-step image analysis process designed to improve the efficiency and accuracy of plant disease diagnosis. The first step uses two convolutional neural networks to perform two key tasks: separating barley leaf blades from the image background and segmenting net spot lesions. This allows for precise identification of affected areas, which is critical for subsequent analysis. At the second stage, a quantitative assessment of the degree of damage occurs, based on counting the pixels of affected and healthy areas of the leaf. The ratio of the areas of the affected areas to the total leaf area is determined, which provides an accurate and objective assessment of the degree of disease development. This method demonstrates significant advantages over traditional visual diagnostic methods, including increased accuracy and objectivity, as well as an accelerated analysis process. Field and laboratory studies were carried out in 2021–2023 at the sites of the Federal State Budgetary Institution Federal Research Center for Biological Plant Protection.

Keywords: winter barley, barley diseases, diagnosis of disease development, net spot, pathogen, protection of grain crops, phytosanitary monitoring, computer vision, artificial intelligence

For citation. Arinicheva I.V., Volkova G.V., Yakhnik Yа.V., Arinichev I.V. Computer vision for monitoring and accounting Pyrenophora teres of winter barley. News of the Kabardino-Balkarian Scientific Center of RAS. 2024. Vol. 26. No. 2. Pp. 72–79.  DOI: 10.35330/1991-6639-2024-26-2-72-79