Prospects for digitalization of monitoring of Lepidoptera phytophages of grapes

[摘要] The effectiveness of phytophagous insect control systems depends on the accuracy of the chosen method of monitoring their numbers and the available information on the dynamics of the population number. As part of the automation of the processes of collecting and analyzing data on the biological activity of pheromone preparations of the phytophage of grapes Lobesia botrana Den. et Schiff., the method of counting the number of pest butterflies on the glue inserts of pheromone traps has been improved by the use of digital cameras and MCView software. According to the validation results, it was found that, relative to the actual data, the improved method ensured the accuracy of counting butterflies at a density of 45-321 imago/insert at an average of 99.1%, which exceeded the results of counting by the traditional visual method by an average of 5%. It is shown that partial method automation made it possible to reduce the influence of a particular researcher, as a subjective factor, on the experiment results. An increase in the time required for counting butterflies by the improved method relative to the visual method by an average of 30% was noted. The presented method of quantitative evaluation of the results of pheromone monitoring in vineyards is a transitional stage to the full automation of this information collection and analysis. For this purpose, to date, photosets have been collected containing more than 1000 photos of glue inserts of pheromone traps with captured butterflies and training of a convolutional neural network has begun designed to detect and quantify such grape lepidopteran phytophages as Lobesia botrana, Helicoverpa armigera Hbn., Theresimima ampelophaga Bayl. (Insecta: Lepidoptera).

[学科分类] 材料科学（综合）