The paper substantiates the need to obtain an assessment of the harm of food products for consumers with chronic diseases or allergies, which is important to prevent a possible worsening of the course of the disease or eliminate an acute allergic reaction of the human body to hazardous ingredients present in the product. It is proposed to use food labels and packaging as the primary sources of information about the food product that is available to the consumer. It is shown that, the printed information on the packages of Ukrainian food products meets the requirements of the law "On Consumer Information on Food Products" and the labeling on food labels is presented in the text-graphic form. In this work it is used convolution neural networks for text-graphic information processing. It is proposed and substantiated the system structure for detecting and analyzing the text-graphical information of product composition. It is developed mobile software solution.