Deep learning-based regression of food quality attributes using near-infrared spectroscopy and hyperspectral imaging: A review

Yuxin Xiao, Lei Zhou, Yiying Zhao, Hengnian Qi, Yuanyuan Pu, Chu Zhang,
Deep learning-based regression of food quality attributes using near-infrared spectroscopy and hyperspectral imaging: A review,
Food Chemistry,
Volume 493, Part 4,
2025,
145932,
ISSN 0308-8146,
https://doi.org/10.1016/j.foodchem.2025.145932.
(https://www.sciencedirect.com/science/article/pii/S0308814625031838)
Abstract: Near-infrared (NIR) spectroscopy and hyperspectral imaging (HSI) are two popular non-destructive tools for food quality and safety inspection. For food quality attributes quantification, the key is to develop regression models to link the features (spectral, spatial and their fusion) and the quality attributes. In addition to conventional machine learning methods, deep learning-based regression has proved to be a promising and advantageous approach to quantify the quality attributes. This review presents a comprehensive summary of recent advances in applying deep learning algorithms for quantifying food quality attributes using NIR spectroscopy and HSI. Deep learning regression algorithms are briefly introduced and compared with conventional data analysis strategies for regression. Furthermore, the strategies that help to fully reveal the advantages of deep learning are highlighted. The challenges and future perspectives are also discussed. This review provides a comprehensive understanding of the application of deep learning in food quality attribute quantification.
Keywords: Deep learning for spectral data analysis; Non-destructive food quality prediction; Near-infrared spectroscopy and hyperspectral imaging regression models; Spectral feature extraction; Transfer learning; Data augmentation