The increasing availability and variety of foods gives people more choice, but this has also lead to poor diets high in calories, carbohydrates and fats, consequently leading to chronic disease, such as diabetes and obesity. Many mobile apps have been proposed to help users monitor and control their diets. However, many apps are essentially cold calorie calculators or assume that users will robotically adhere to diet instructions. We aim to develop a human-centric, AI-driven mobile app to support food logging and recommendation that celebrates the love of food and gently encourages behavior change. In collaboration with dietitians and experts in public health, we are developing deep learning models to automatically classify food images to reduce the burden of food tracking, to recommend healthy foods more accurately, and investigating the user experience and context of food logging.

Our work is integrated into a mobile app, TasteHealthy, to recommend tasty and healthy food, and to enable easy food logging. More information at: http://tastehealthy.comp.nus.edu.sg/

Relevant papers: