METHODS AND PRACTICES OF TRAINING AND APPLICATION OF CONVOLUTIONAL NEURAL NETWORKS FOR DETECTING VARIOUS DEFECTS ON THE SURFACES OF BUILDING STRUCTURES

The subject of research in this article is the practice of implementing and applying types of artificial intelligence – convolutional neural networks and data science - to detect various defects on the surfaces of building structures. The purpose of the study is to identify the advantages and disadvantages of implementing and using convolutional neural network technologies together with other software and hardware components to automate traditional methods of monitoring the technical condition of the external surfaces of buildings and structures. The article solves the following tasks: to substantiate the effectiveness of implementation of convolutional neural network technologies and data science methods; to apply them in practice with software and hardware technologies to automate traditional methods of performing work on detecting various defects on the surfaces of building structures in the construction and operation of buildings and structures; to show the problems and shortcomings of these methods and technologies. To solve the tasks, an integrated approach was used with the use of general scientific and special research methods (analysis, explanation, generalisation, comparison). The following results have been obtained: the features that affect the accuracy of the analysis of the collected data used by neural convolutional network technologies to detect various defects on the surfaces of building structures have been identified; practices and methods for more efficient and accurate application of this technology have been reflected. Conclusions: the study allowed to identify the practical opportunities and problems that exist in this technology; recommendations for the effective use of this technology were developed; factors influencing the more efficient use of this technology in industry were identified.