This research consists of the analysis of the methods of image recognition, focusing on the problem of food classification, aiming to use the methods in a mobile application for the assistance in food monitoring and control. Thus, the development of the work contemplates the use of the deep learning method, focused on the recognition of food in images, with the use of neural convolution networks (CNN). For this purpose, a data set consisting of more than 1000 images and 5 food classes was constructed in order to simulate the SimpleNet, MiniVGGNet and Small Xception models, and thus define a learning model for food classification.