Today all branches of industrial production are being actively automated. This determines the relevance of the development of tools for creating automated systems. The article proposes the use of AutoML approach to automate the creation of machine learning models that are used for automatic monitoring. The model of object recognition in the image VGG19 was chosen as an example. On its example the automation of structural-parametric synthesis of models and optimization of hyperparameters was demonstrated. The presented system was realized as a part of the AutoGenNet software platform. This platform realizes the concept of No-Code development, which allows to hide from users the complexity of the processes of model creation and training. The use of the No-Code development concept allows to reduce the entry threshold for working with the program. Also, on the basis of AutoGenNet platform the mechanism of generation of program shells for operation of trained models was realized. All this allowed to implement AutoML approach for automation of VGG19 model generation and training processes, which in turn simplified and accelerated the process of solving automatic monitoring problems with deep learning models. The created system can be scaled and used for automated generation of other object recognition models for solving a wide variety of applied problems.