Classification of Food Nutrients Composition using Deep Learning

Abstract

Deep Learning is the technique that uses multiple layers of a neural network to automatically distinguish patterns to learn to make predictions. A problem that humans face on a daily basis is how to make a conscious decision regarding our daily food consumption that is nutritious and healthy. By having a tool that helps facilitate the decision making process of what type of food to eat by showing useful nutritional information to us immediately would greatly improve our lives. By critically analysing prominent research papers that relate to deep learning techniques to classify food and their nutrients composition, we decided upon the suitable Deep Learning algorithm to classify food nutrients composition as well as the appropriate image dataset to be used. Therefore in this paper we propose the classification of food nutrients composition utilizing deep learning techniques. The proposed framework uses convolutional neural networks (CNN) as a basis of recognising images of food and classifying the food into their corresponding nutrients composition such as fats, carbohydrates, proteins and more. As part of our future work, we shall use the proposed framework to conduct the training and implementation of the deep learning model to make predictions on food nutrients. The chosen dataset shall be used to train the model where patterns and characteristics of the food images are distinguished over multiple passes of the neural network. Once the model has been trained, then new food images may be introduced to make a prediction from the context that have been learned from before.