Vision-Based Approaches for Automatic Food Recognition and Dietary Assessment: A Survey

Abstract

Consuming the proper amount and right type of food have been the concern of many dieticians and healthcare conventions. In addition to physical activity and exercises, maintaining a healthy diet is necessary to avoid obesity and other health-related issues, such as diabetes, stroke, and many cardiovascular diseases. Recent advancements in machine learning applications and technologies have made it possible to develop automatic or semi-automatic dietary assessment solutions, which is a more convenient approach to monitor daily food intake and control eating habits. These solutions aim to address the issues found in the traditional dietary monitoring systems that suffer from imprecision, underreporting, time consumption, and low adherence. In this paper, the recent vision-based approaches and techniques have been widely explored to outline the current approaches and methodologies used for automatic dietary assessment, their performances, feasibility, and unaddressed challenges and issues.

Keywords

• Food recognition
• food classification
• food volume estimation
• food nutrient information
• food image datasets