The use of neural networks for image analysis allows for fast and high-quality recognition in conditions of insufficient data. Analyzed a neural network based on the architecture of CNN to determine the recipe of the dish in the image. The analysis of images of culinary dishes using neural networks potentially makes it possible to determine their approximate composition and possible preparation recipe. The concept of a mobile application for carrying out such an analysis by transferring an image to a server that performs analysis using a neural network has been proposed. The possibility of loading educational images and recipes is considered.