Rapid Non-Destructive Analysis of Food Nutrient Content Using Swin-Nutrition

Abstract

Food non-destructive detection technology (NDDT) is a powerful impetus to the development of food safety and quality. One of the essential tasks of food quality regulation is the non-destructive detection of the food’s nutrient content. However, existing food nutrient NDDT performs poorly in terms of efficiency and accuracy, which hinders their widespread application in daily meals. Therefore, this paper proposed an end-to-end food nutrition non-destructive detection method, named Swin-Nutrition, which combined deep learning and NDDT to evaluate the nutrient content of food. The method aimed to fully capture the feature information from the food images and thus accurately estimate the nutrient content. Swin-Nutrition resorted to Swin Transformer, the feature fusion module (FFM), and the nutrient prediction module to evaluate nutrient content. In particular, Swin Transformer acted as the backbone network for feature extraction of food images, and FFM was used to obtain the discriminative feature representation to improve the accuracy of prediction. The experimental results on the Nutrition5k dataset demonstrated the effectiveness and efficiency of our proposed method. Specifically, the mean value of the percentage mean absolute error (PMAE) for calories, mass, fat, carbohydrate, and protein were only 15.3%, 12.5%, 22.1%, 20.8%, and 15.4%, respectively. We hope that our simple and effective method will provide a solid foundation for the research of food NDDT.

Artificial Intelligence in Food Safety: A Decade Review and Bibliometric Analysis

“…Dietary monitoring systems based on ML methods even could automatically assess dietary intake [ 44 , 113 ]. Shao et al (2022) introduced a non-destructive detection method, combining Swin Transformer with a feature fusion module, which was applied to evaluate the nutrient content of food [ 153 ]. Chen et al (2021) implemented a proprietary deep-learning model that provided a nutrition assessment of restaurant food [ 154 ].…”

“…Computer vision systems based on ML or DL represented great potentiality in food processing and distribution [ 72 , 84 , 86 , 102 , 103 , 115 , 158 , 160 , 184 ]. In addition, non-destructive detection methods based on ML or DL shed light on nutrient estimation and dietary assessment [ 116 , 153 , 154 , 164 , 165 ].…”

“…Tay et al (2020) discussed current dietary assessment methods, including DL applications in food volume estimation [ 170 ]. To evaluate nutrient content quickly and accurately, Shao et al (2022) introduced a non-destructive detection method based on Swin Transformer as the backbone network for image feature extraction [ 153 ].…”

DPF-Nutrition: Food Nutrition Estimation via Depth Prediction and Fusion

“…We referred to the methods proposed by Thames et al [10] and Shao et al [15] as Google-Nutrition and RGB-D Nutrition, respectively. We compared the methods including: Google-Nutrition, RGB-D Nutrition, Swin-Nutrition [11] and our proposed DPF-Nutrition.…”

“…Similarly, Thames et al [10] demonstrated that the performance of the vison-based method outperformed the professional human nutritionists where they used a multi-task convolutional neural network to estimate multiple nutrients. Shao et al [11] employed a combination of non-destructive detection technology and deep learning to analyze the nutritional content of food. They improved the detection of small target foods to improve the nutrient estimation accuracy.…”