Intelligent Agent for Food Recognition in a Smart Fridge

Respecting an adequate diet is essential for a healthy lifestyle. However, keeping track of the daily consumed food and of what one has in the home fridge may be tedious and time consuming if not backed up by user-friendly applications. In this context, the paper presents some of the main components of an intelligent agent and system aimed to create a user-oriented application for recommending personalized food recipes based on the available ingredients and dietary restrictions in a domestic environment. The developed application includes a component to accurately detect the ingredients from an image taken from inside a fridge equipped with mounted cameras, and allows the user to get information about each ingredient, such as the ingredient type, name and associated calories. To facilitate the development of this application, our own data set of food ingredients was collected and annotated, and an ontology of recipes was developed.