ANALYSIS OF RESEARCH IN THE APPLICATION OF ARTIFICIAL INTELLIGENCE SYSTEMS IN WEED CONTROL

DOI 10.32651/238-49

Issue № 8, 2023, article № 8, pages 49-54

Section: DIGITALIZATION IN THE AGRO-INDUSTRIAL COMPLEX

Language: Russian

Original language title: АНАЛИЗ ИССЛЕДОВАНИЙ В ОБЛАСТИ ПРИМЕНЕНИЯ СИСТЕМ ИСКУССТВЕННОГО ИНТЕЛЛЕКТА ПРИ БОРЬБЕ С СОРНЯКАМИ

Keywords: AGRICULTURE, ARTIFICIAL INTELLIGENCE, WEED CONTROL, WEED IDENTIFICATION, RESEARCH REVIEW

Abstract: Weed control is a major problem in agricultural production. Clogging of cultivated fields with weeds leads to a sharp decrease in crop yields. The purpose of the study is to analyze research in the field of the use of artificial intelligence systems in weed control. The results of the selection of publications from international citation databases on the research problem are presented. To identify weeds, convolutional neural networks are mainly used, in some cases in combination with methods of support vector machine, nearest neighbor, various classifiers, and also their ensembles (combinations). It has been established that artificial intelligence systems allow, with an accuracy of over 90%, to identify individual weeds and their groups in crops of cultivated plants. The effects of the use of artificial intelligence systems in weed control are highlighted, which consist in reducing the expenditure of plant protection products and crop yields during spot spraying of weeds. Specific recommendations are given on the use of artificial intelligence systems for mechanical weeding by robotics.

Authors: Semin Aleksandr Nikolaevich, Skvortsov Egor Artemovich, Skvortsova Ekaterina Gennadevna