Aerospace Instrument-Making Annotation << Back EXPERIMENTAL INVESTIGATION WITH «DARKNET» NEURAL NETWORK CAPABILITIES ON THE TASK OF REMOTE SENSING IMAGES PROCESSING V.P. Fralenko The results of the practical application of the neural network of the «Darknet» type to solve the problem of searching and classifying targets in images obtained from unmanned aerial vehicles are presented. The developed subsidiary tools for preparation of training samples, optimal configuration and operation of a neural network are described. A multiclass and one-class problem of searching and recognizing targets was solved. A comparative analysis of the results is performed. Key words: artificial neural network, search, recognition, Earth remote sensing, experimental research, unmanned aerial vehicle. Contacts: Е-mail: alarmod@pereslavl.ru Pp. 44-52.