Dietary Nutritional Information Autonomous Perception Method Based on Machine Vision in Smart Homes.

Area of Science:

• Computer Science
• Artificial Intelligence
• Nutrition Science

Background:

• Smart homes require advanced methods for automatic user monitoring.
• Accurate dietary tracking is crucial for health management and personalized nutrition.

Purpose of the Study:

• To develop an autonomous perception method for dietary nutritional information in smart homes.
• To enhance user health monitoring through machine vision-based food recognition and nutritional analysis.

Main Methods:

• Utilized a YOLOv5-based food recognition algorithm for dietary intake monitoring via social robots.
• Developed a method for calibrating food ingredient weight and calculating nutritional composition.
• Proposed the dietary nutritional information autonomous perception method (DNPM) for quantitative analysis.

Main Results:

• The YOLOv5 food recognition algorithm achieved an average accuracy of 89.7% on the CFNet-34 dataset.
• The DNPM system demonstrated an average nutritional composition perception accuracy of 90.1%.
• The system exhibited a fast response time (<6 ms) and high processing speed (>18 fps).

Conclusions:

• The proposed machine vision-based method provides accurate and robust dietary nutritional information perception in smart homes.
• The system's performance indicates its potential for real-time health monitoring and personalized dietary recommendations.