API for Food Recognition in Images using Convolutional Neural Networks and YOLO

Resumo

DOI: https://zenodo.org/doi/10.5281/zenodo.10711765

This project aims to create a framework that enables the development of an application to assist individuals who have begun nutritional counseling and need to monitor the macronutrients in their meals. This is achieved through the identification of food items in meal images and the retrieval of nutritional information for each item using a convolutional neural network, specifically the YOLO architecture. Images of dishes containing common foods in the Brazilian diet were used to train the model responsible for identifying the items in the photo, and nutritional information for each item was extracted from the Brazilian Food Composition Table (TACO). This innovative approach has the potential to enhance the accuracy and convenience of dietary tracking for those under the guidance of nutritionists.