Nutrition plays a key role in avoiding different health problems such as diseases, and it also linked with mental and physical wellbeing. Nutrition monitoring consists firstly of identifying food types and secondly, recommending the appropriate food for better nutrition. This project addresses the aspect of automatically identifying and classifying local dishes, from an image, using Artificial Intelligence (AI). Automatically defining the structure and composition of a dish, from an image, is very complex from a traditional image processing perspective. However, Deep Learning (DL) techniques are effective in identifying complex structures from images. Our approach is based on identifying and assessing the most appropriate Deep Learning models for Mauritian food classification, which entails the creation of a dataset for local food images. Once built, trained and tested for an accurate prediction, the model will be accessible through a mobile application with basic recommendation capabilities. This project couples the accessibility of mobile phone cameras with the innovative possibilities of AI, to automate the screening and assessment of different nutritional elements. This system will contribute to maintain a healthy diet while avoiding or controlling the spread of noncommunicable diseases (NCD).
