Structure from motion-convolutional neural network model (SfM-CNN) achieved accurate portable Chinese dietary chemical composition estimation for dietary recall

IF 8.5 1区农林科学 Q1 CHEMISTRY, APPLIED

Food Chemistry Pub Date : 2025-05-27 DOI:10.1016/j.foodchem.2025.144908

Peihua Ma , Hsuan Chih Hong , Xiaoxue Jia , Cheng Jan Chi , Ning Xiao , Bei Fan , Fengzhong Wang , Cheng-I Wei

{"title":"Structure from motion-convolutional neural network model (SfM-CNN) achieved accurate portable Chinese dietary chemical composition estimation for dietary recall","authors":"Peihua Ma , Hsuan Chih Hong , Xiaoxue Jia , Cheng Jan Chi , Ning Xiao , Bei Fan , Fengzhong Wang , Cheng-I Wei","doi":"10.1016/j.foodchem.2025.144908","DOIUrl":null,"url":null,"abstract":"<div><div>Accurately estimating the chemical composition of dietary intake is essential for health and nutrition management, especially in regions with complex culinary diversity like China. This study introduces a novel AI-driven solution using a Structure from Motion-Convolutional Neural Network (SfM-CNN) model to automate chemical composition analysis of Chinese food. By integrating advanced 3D reconstruction techniques with deep learning, specifically the Scale-Invariant Feature Transform (SIFT) algorithm, we achieved superior feature extraction and food volume estimation with less than 4 % error. Our model, trained on the newly developed ChineseDish-100 dataset, demonstrated an R<sup>2</sup> of 0.949 for carbohydrate content estimation using the SIFT-ResNet50 architecture. The model's interpretability was enhanced through visualizations, facilitating parameter optimization and reliable chemical composition estimation. These results underscore the potential of AI-powered models in providing efficient, accurate, and culturally relevant dietary analysis tools, marking a significant advancement for nutritional science, food chemistry, and public health initiatives in culturally diverse regions.</div></div>","PeriodicalId":318,"journal":{"name":"Food Chemistry","volume":"489 ","pages":"Article 144908"},"PeriodicalIF":8.5000,"publicationDate":"2025-05-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Food Chemistry","FirstCategoryId":"97","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0308814625021594","RegionNum":1,"RegionCategory":"农林科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CHEMISTRY, APPLIED","Score":null,"Total":0}

引用次数: 0

Abstract

Accurately estimating the chemical composition of dietary intake is essential for health and nutrition management, especially in regions with complex culinary diversity like China. This study introduces a novel AI-driven solution using a Structure from Motion-Convolutional Neural Network (SfM-CNN) model to automate chemical composition analysis of Chinese food. By integrating advanced 3D reconstruction techniques with deep learning, specifically the Scale-Invariant Feature Transform (SIFT) algorithm, we achieved superior feature extraction and food volume estimation with less than 4 % error. Our model, trained on the newly developed ChineseDish-100 dataset, demonstrated an R² of 0.949 for carbohydrate content estimation using the SIFT-ResNet50 architecture. The model's interpretability was enhanced through visualizations, facilitating parameter optimization and reliable chemical composition estimation. These results underscore the potential of AI-powered models in providing efficient, accurate, and culturally relevant dietary analysis tools, marking a significant advancement for nutritional science, food chemistry, and public health initiatives in culturally diverse regions.

查看原文本刊更多论文

基于运动-卷积神经网络模型（SfM-CNN）的结构实现了精确的便携式中式膳食化学成分估计

准确估计膳食摄入的化学成分对于健康和营养管理至关重要，特别是在像中国这样烹饪多样性复杂的地区。本研究提出了一种新的人工智能驱动的解决方案，利用运动-卷积神经网络（SfM-CNN）模型来自动分析中国食品的化学成分。通过将先进的3D重建技术与深度学习相结合，特别是尺度不变特征变换（SIFT）算法，我们实现了卓越的特征提取和食物体积估计，误差小于4 %。我们的模型在新开发的ChineseDish-100数据集上进行训练，使用SIFT-ResNet50架构估算碳水化合物含量的R2为0.949。通过可视化增强了模型的可解释性，便于参数优化和可靠的化学成分估计。这些结果强调了人工智能模型在提供高效、准确和与文化相关的饮食分析工具方面的潜力，标志着文化多样性地区营养科学、食品化学和公共卫生倡议取得了重大进展。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Food Chemistry 工程技术-食品科技

CiteScore

16.30

自引率

10.20%

发文量

3130

审稿时长

122 days

期刊介绍： Food Chemistry publishes original research papers dealing with the advancement of the chemistry and biochemistry of foods or the analytical methods/ approach used. All papers should focus on the novelty of the research carried out.