A portable NIR-system for mixture powdery food analysis using deep learning. (January 2022)

• Record Type: Journal Article
• Authors: Zhou, Lei; Tan, Lehao; Zhang, Chu; Zhao, Nan; He, Yong; Qiu, Zhengjun
• Is Part Of: Lebensmittel-Wissenschaft + Technologie =. Volume 153(2022)
• Publication Date: 2022-01
• Subjects: Powdery food -- NIR spectroscopy -- Chemometrics -- Convolutional neural network -- Feature selection; Food industry and trade -- Periodicals; Food -- Composition -- Periodicals; Microbiology -- Periodicals; Nutrition -- Periodicals; 664.005
• DOI: 10.1016/j.lwt.2021.112456
• Languages: English
• ISSNs: 0023-6438

Abstract

Abstract: The combination of near-infrared spectroscopy and machine intelligence has been an emerging nondestructive tool for powdery food evaluation. In this research, a novel portable system (defined as NIR-Spoon) was presented for simultaneously evaluating the mixing proportion of multi-mixture powdery food. Convolutional neural networks for multi-regression (CNN-MR) and that for feature selection (CNN-FS) were proposed for spectra processing. Multi-mixture powder samples, which contained one or more components including milk, rice, corn and wheat, were inspected by the NIR-Spoon. Results showed that the partial least squares regression (PLSR) model estimated the proportion of mixture with root mean square error (RMSE) of 0.059 and correlation coefficient (R 2 ) of 0.938. The proposed CNN-MR realized a further improvement comparing to the benchmark PLSR method, with 0.035 for RMSE and 0.976 for R 2 . The CNN-MR still kept R 2 of 0.970 based on 25 features selected by the CNN-FS algorithm. Moreover, the integrated load sensor could convert the proportion to the weight of each component. All hardware and software were integrated on the NIR-Spoon. Overall, the NIR-Spoon provided satisfactory accuracy and user-friendly mobile applications. It also has excellent potential to be extended for inspecting other kinds of food products in future research. Highlights: A portable system is proposed for mixtures powdery food analysis. NIR sensor, load sensor and deep learning methods are integrated on the system. Deep learning multi-regression is proposed for mixtures powder evaluation. CNN-based feature selection is proposed for model simplification.