Determination of the Requirements for the Neural Network for Recognition Algorithm UHRSI

The Earth remote sensing is becoming a new and quickly developing multiscience area of a practical importance and different business applications. The IT industry now in accordance with the academic sector is developing new competences in the area which includes the development of new algorithms and calibration of existing and already working program complexes and libraries. Investigation about using neural networks for detection geo-objects on the satellite images are held by different research organizations. In this paper we present the new way of determination of the initial data, the architecture type of the neural network, the accuracy characteristics of the recognized objects for the algorithm of recognition of soil use areas of open type based on the data of ultrahigh resolution space imagery (UHRSI) which was made according to the practice on the newest satellite images. The results could be used in the various application of remote data analysis including satellite application in different branches of military and economic needs. © 2019 Published under licence by IOP Publishing Ltd.