Побудова класифікатора рослинних об’єктів за допомогою нейронних мереж

Abstract

UK: Розглянуто можливість побудови класифікатора рослинних об’єктів за допомогою двошарової та трьохшарової нейронної мережі. EN: Possibility for construction of a vegetative objects’ classifier is considered by means of a double-layer and three-layer neural network. The neural networks are built for classification of plants on three classes. Probabilities of discrimination are compared for a double-layer and three-layer neural network. The construction of neural networks was made on the basis of experimental data obtained in the field with the help of a spectrometer. When constructing the neural network in the MatLab program, as a type of the neural networks, the neural network of direct distribution of signals was selected, it is being studied by the Levenberg-Marquardt method. Decrease of features’ amount has been proposed to raise an operation speed of the vegetative object classifier.