Nutrition information estimation from food photos using machine learning based on multiple datasets

Abstract

Bodyweight, blood pressure, and cholesterol are all risk variables that can aid people in making educated decisions regarding their health promotion activities. Food choices are among the most effective methods for preventing chronic illnesses, including heart disease, diabetes, stroke, and some malignancies. Because various meals give varying amounts of energy and minerals, good eating necessitates keeping track of the nutrients we ingest. Furthermore, there is a paucity of information on whether understanding food constituents might aid in more accurate nutrition calculations. Therefore, this research suggests processing food images on social media to anticipate the contents of each food and extracting nutrition information for each food image to serve as healthy implicit feedback to take advantage of the rapid accumulation of rich photos on social media. The proposed methodology is a framework based on a machine-learning model for predicting food ingredients. We also compute critical health metrics for each ingredient and combine them to obtain nutrition data for the food. The result revealed a promising way of extracting food components and nutrition information. Compared with other researchs, our proposed prediction and attribute extraction strategy achieves a remarkable accuracy of 85%.

Topics & Concepts