METHOD OF PREPARING A SET OF PHOTOGRAPHS FOR MACHINE ANALYSIS FOR DETECTING AND CLASSIFYING PATHOLOGIES OF CROPS Russian patent published in 2022 - IPC G01C11/00 A01G7/00 G06T1/40 

FIELD: photography.

SUBSTANCE: invention relates to methods of preparing a set of photographs for machine analysis and can be used in preparing photographs for detecting and classifying pathologies of crops. Method of photographing is determined. If the main signs of the disease are concentrated on separate parts of the plant, then these separate parts of the plant are photographed; if the nature of the disease is determined for the whole plant, then the whole plant is photographed; if these are micro-lesions, clearly structured fragments of specific parts of plants are photographed through a magnifying glass or microscope. At that, determining the characteristic area of damage of the leaf, fruit, berry, bush, which should occupy at least 50–60 % of the image, second part of the image should include an informative environment — healthy areas, if any. Photographs of healthy plants and their parts are taken. Photographs of the healthy specimens should correspond to the photographs of the specimens with diseases. Shooting is performed during the whole phenological cycle so that photographs reflect all phenological phases of plants development. Shooting mode in photographic equipment is set to "macro", the focusing mode is set to single-point, and the parameters of the image resolution and its quality are set to maximum. Basic data set containing photographs of diseased and healthy specimens is formed, which serves to form training and test data samples. Areas of growth of photographed plants are divided into uniquely identifiable homogeneous areas. Photographs with pathology are divided into categories according to disease activity. Photographs with pathology are divided according to the performed treatment. Photographs of the main data set are arranged in folders arranged in a hierarchical structure. Surveys are associated with coordination referencing. Data set with photographs is associated with weather and climatic data relevant to the survey areas. Obtained photographs are stored in a digital format in a machine-readable form on a data medium.

EFFECT: minimizing errors and simplifying the process of forming a set of photographs for machine analysis, enabling timely detection and elimination of pathologies of crops.

1 cl