goFOOD^TM: An Artificial Intelligence System for Dietary Assessment

Area of Science:

• Computer Science
• Nutrition Science
• Biomedical Engineering

Background:

• Accurate nutritional information is crucial for promoting healthier diets and improving clinical outcomes.
• Traditional dietary assessment methods can be time-consuming and prone to inaccuracies.

Purpose of the Study:

• To introduce goFOOD™, an AI-powered system for estimating meal calorie and macronutrient content using smartphone images.
• To evaluate the performance of goFOOD™ compared to experienced dietitians.

Main Methods:

• goFOOD™ utilizes deep neural networks for food detection, segmentation, and recognition from two meal images or a short video.
• A 3D reconstruction algorithm estimates food volume, combined with a nutrient database for calculations.
• The system supports 319 fine-grained food categories.

Main Results:

• goFOOD™ demonstrated superior performance compared to experienced dietitians on a non-standardized meal database.
• The system's accuracy was comparable to dietitians on a fast-food meal database.
• Validation was performed on two diverse multimedia databases.

Conclusions:

• goFOOD™ offers a simple, efficient, and accurate solution for end-users to perform dietary assessments.
• AI-driven image analysis presents a promising approach for automated nutritional estimation.
• This technology has the potential to support personalized nutrition and public health initiatives.