Development of a prototype of an intelligent system for processing and visualization of neural network results when solving detection, classification and segmentation tasks

Currently, most of the known applications that use neural networks for video or image processing are designed for specific tasks, so they have their own features. Programmers for each new task of detection, classification and segmentation for a particular neural network architecture need to create additional software (software) taking into account all the dependencies and specifics of the project. The article substantiates the necessity of creating a common universal program for processing and visualization of neural network results, and formulates criteria for it. It describes the development of a prototype of such an intelligent system, which will allow to obtain the required result at once, without preliminary creation of a number of software modules, thus reducing the time for product development.