Analysis of neural network architectures for computer vision tasks and prospects for integrating interval computing into them

This paper provides an overview of the development of neural networks in the field of computer vision. The history of the first prototypes of neural networks for image processing, convolutional neural networks, is presented. The typical structure of convolutional neural networks is described and the analysis of the main modern neural networks of this type is carried out. The main transformers for image processing are described. The prospect of developing neural networks using interval analysis is considered, and the basic elements of convolutional neural networks for interval computing are modified. A computational experiment of the implemented interval convolutional network model is carried out.