The research is devoted to determining the prospects of developing a model of segmentation of images of damage to critical infrastructure objects using machine learning tools. The development of legislation on artificial intelligence as a machine learning tool is analyzed. It was determined that the important elements of the critical infrastructure are objects with a cement concrete coating. It was determined that damage analysis is important for assessing and forecasting the state of critical infrastructure objects. Available software for processing information on damage to infrastructure objects by the photofixation method is considered. The shortcomings of the existing data processing application software are indicated. The need for model development for segmentation analysis of infrastructure objects damage images using machine learning tools is identified. In today's conditions of uncertainty and limited funding, there is a separate problem of developing a system for operational management of critical infrastructure objects, including one focused on solving the problems of restoring and building infrastructure damaged by military actions. It was determined that such models and systems should be based on a combination of advanced solutions in the field of collecting and processing data on the quality condition of objects.