Updating database of repository of neural network models for analysis and prediction of spatial processes

The article describes the solution for the problem of updating the database of the repository of neural network models for the analysis and forecasting of spatial processes based on the project-oriented approach. The process of integrating deep neural networks in the repository is based on the project-oriented approach, on the basis of which each stored deep neural network is compared with the range of tasks within which it can be used and with the data it analyzes. Within the framework of the repository, models are integrated that allow solving the problems of classification, segmentation and detection of objects based on remote space monitoring data.