The main machine learning algorithms that can be used in the process of modeling equipment and parts are presented. The scientific novelty lies in determining ways to increase the efficiency of product modeling using machine learning methods. Neural networks are considered one of the most popular algorithms in the field of big data processing. Two basic directions of using neural networks in image recognition are described: classification and regression. The method of step-by-step modeling of 3-D objects used in Image recognition (classification) is presented. Examples of neural networks that generate images based on the Latin and Cyrillic alphabets are given. It is proven that the presented approach allows companies to optimize the process of generating parts. These approaches are at the stage of initial implementation in practical activities since it is not always possible to accurately create visualizations based on a text description.