Firmness is a key indicator of apple quality. Building a predictive model for apple firmness based on hyperspectral technology and regression algorithms can achieve rapid, non-destructive, and high-throughput detection of apple firmness. This paper adopts an Adaptive Window Length Savitzky-Golay Smoothing (AWL-SG smoothing) algorithm based on the Savitzky-Golay Smoothing (SG smoothing) algorithm, which can adaptively adjust the window length according to the change rate of spectral data at different wavelengths. SG smoothing, AWL-SG smoothing, Standard Normal Variate (SNV), and Multiplicative Scatter Correction (MSC) algorithms were used to preprocess the original spectral data, and Partial Least Squares (PLS), Ridge Regression (Ridge), and Kernel Ridge Regression (Kernel Ridge) predictive models were constructed to analyze the impact of different preprocessing methods on model prediction accuracy. The prediction models established with spectral data preprocessed by SG smoothing and AWL-SG smoothing algorithms showed significant improvement in predictive performance on the basis of the original spectral data, among which the AWL-SG smoothing algorithm performed the best. The Ridge model established with spectra data preprocessed by AWL-SG smoothing achieved an R2 of 0.8914 in the test set. Successive Projection Algorithm (SPA), Principal Component Analysis (PCA), and Independent Component Analysis (ICA) dimensionality reduction algorithms were used to reduce the dimensions of the full-band spectral data preprocessed by SG smoothing and AWL-SG smoothing algorithms, and Ridge and Kernel Ridge prediction models were constructed. The results showed that both SPA and PCA algorithms could improve the predictive performance of the models, with the PCA performing the best. The combination of AWL-SG + PCA + Ridge achieved the best predictive effect, with an R2 of 0.9146 in the test set.
摘要中文：
硬度是衡量苹果品质的关键指标。基于高光谱技术和回归算法构建苹果硬度预测模型，可实现对苹果硬度的快速、无损和高通量检测。本文在Savitzky-Golay平滑（SG平滑）算法的基础上，采用了一种自适应窗口长度的Savitzky-Golay平滑算法（AWL-SG平滑），该算法可根据不同波长处光谱数据的变化率自适应地调整窗口长度。研究中使用SG平滑、AWL-SG平滑、标准正态变量变换（SNV）和多元散射校正（MSC）等算法对原始光谱数据进行预处理，并构建了偏最小二乘回归（PLS）、岭回归（Ridge）和核岭回归（Kernel Ridge）预测模型，以分析不同预处理方法对模型预测精度的影响。结果表明，基于SG平滑和AWL-SG平滑预处理后的光谱数据所建立的预测模型，其性能相较于原始光谱数据有显著提升，其中AWL-SG平滑算法效果最佳。采用AWL-SG平滑预处理光谱数据构建的Ridge模型在测试集上取得了0.8914的决定系数（R²）。此外，研究还采用连续投影算法（SPA）、主成分分析（PCA）和独立成分分析（ICA）等降维算法，对经SG平滑和AWL-SG平滑预处理后的全波段光谱数据进行降维，并在此基础上构建Ridge和核岭回归预测模型。结果表明，SPA和PCA均能进一步提升模型的预测性能，其中PCA效果最优。最终，AWL-SG + PCA + Ridge组合取得了最佳预测效果，在测试集上的R²达到0.9146。

The processed food sector is primarily concerned with food quality, nutritional content, and processing technique as customers are expecting food products that are consistent in terms of texture, flavor, shelf life, and sensory attributes. Due to the increasing need for food in tandem with the growing global population, artificial intelligence has become modern technology in the food business during the past few decades. The demand for artificial intelligence-based technology in the processed food industry has increased owing to its increased capacity to perform a variety of tasks, including food quality assessment, control tools, food classification, and prediction, etc. This paper reviews the potential Artificial Intelligence (AI) based technologies that can be used in the food sector and the potential application of AI in processed food sectors like milk and milk products, baked goods, fruits and vegetables, etc. A future remark concerning the research gap and the research scope has also been covered in the present review.
摘

英文摘要 (English Abstract):
One of the major barriers to the mass industrial utilisation of brown seaweeds as food sources stem from the food safety risks associated with their iodine and arsenic concentrations, which typically exceed regulatory limits. Hydrothermal treatments might effectively reduce the high iodine and arsenic concentration of Phyllospora comosa below the Australian maximum residual limits (iodine = 1 mg/g and arsenic = 0.00667 mg/g; dry weight) set for brown seaweeds. The experimental hydrothermal treatments dictated that the 82 °C–250 s treatment reduced the iodine concentration from 2.76 mg/g to 0.88 mg/g (68% reduction) and arsenic concentration from 0.01693 mg/g to 0.00965 mg/g (43% reduction). Machine learning models predicted that blanching at 100 °C for ~4 minutes will reduce the arsenic concentration below its maximum residual limit. Additive log-ratio transformations showed that around 50% (dw) of the hydrothermally treated Phyllospora comosa samples were leached out during the highest treatment intensity, 82 °C–250 s. Even though, a half of the biomass is lost, hydrothermally treated Phyllospora comosa products are safer for human consumption and thus may permit the expansion of seaweed production and consumption in Australia.
摘要翻译 (中文)：
褐藻作为食品资源在大规模工业化应用中面临的主要障碍之一，是其碘和砷含量通常超出食品安全监管限值所带来的风险。水热处理可有效降低澳大利亚褐藻物种 Phyllospora comosa 中过高的碘和砷含量，使其低于澳大利亚针对褐藻设定的最大残留限量（碘：1 mg/g，砷：0.00667 mg/g，以干重计）。实验结果表明，在82 °C、250秒的水热处理条件下，碘含量从2.76 mg/g降至0.88 mg/g（降幅达68%），砷含量从0.01693 mg/g降至0.00965 mg/g（降幅达43%）。机器学习模型进一步预测，若在100 °C下焯水约4分钟，即可使砷含量降至最大残留限量以下。通过对数据进行加性对数比（additive log-ratio）转换分析发现，在最高强度处理条件（82 °C–250 s）下，约有50%（干重）的 Phyllospora comosa 样品成分被浸出流失。尽管如此，经水热处理后的 Phyllospora comosa 产品安全性显著提高，更适合人类食用，从而有望推动海藻在澳大利亚的生产与消费规模扩大。

英文摘要 (English Abstract):
For quality evaluation of Gastrodia elata f. glauca (GEFG), it’s crucial to develop an analytical method to determine the geographical origin. Herein, 371 GEFGs are collected from five provinces, focusing on analysis of dry matter content (DMC), origin identification, geographical indication (GI) production area discrimination by using a combination of Fourier transform infrared (FTIR) spectroscopy and deep learning, data driven version of soft independent modeling of class analogy (DD-SIMCA). A significant difference in DMC of GEFG between Yunnan and other origins, which may be related to precipitation, altitude, temperature, and soil. The residual neural network (ResNet) model based on synchronous two-dimensional correlation spectroscopy (2DCOS) images has stable performances, its accuracy is 100%. The DD-SIMCA model can differentiate GI production areas of GEFG, while for non-GI areas, the model specificity is 71.38%. This study provides a promising approach for GEFG geographical traceability and GI production area differentiation.
摘要翻译 (中文)：
在对乌天麻（Gastrodia elata f. glauca，简称GEFG）进行品质评价时，建立一种能够准确判定其地理来源的分析方法至关重要。本研究采集了来自中国五个省份的371份GEFG样品，结合傅里叶变换红外光谱（FTIR）与深度学习技术，以及基于数据驱动的类类比软独立建模方法（DD-SIMCA），重点开展了干物质含量（DMC）分析、产地溯源及地理标志（GI）产区判别研究。
结果表明，云南产GEFG与其他产地样品在干物质含量上存在显著差异，该差异可能与降水量、海拔、温度及土壤条件等环境因素相关。基于同步二维相关光谱（2DCOS）图像构建的残差神经网络（ResNet）模型表现出优异且稳定的性能，产地识别准确率达到100%。此外，DD-SIMCA模型能够有效区分GEFG的地理标志产区；对于非GI产区样品，该模型的特异性为71.38%。本研究为乌天麻的地理溯源及地理标志产区鉴别提供了一种具有应用前景的新方法。

英文摘要 (English Abstract):
Fruits have become essential in people’s daily lives, with both their external defects and internal quality receiving close attention from sellers and consumers alike. This study employs hyperspectral imaging technology combined with deep learning to rapidly and non-destructively detect external defects and soluble solids content (SSC) in Nanfeng mandarins. Hyperspectral data (380-1030 nm) were collected from Nanfeng mandarins with four types of defects (anthracnose, black spot, decay, and scarring) and sound fruits. Firstly, an end-to-end convolutional neural network (CNN) model for qualitative analysis was proposed, and its classification performance was compared with traditional classification models. Three preprocessing methods and three feature selection techniques were applied. The results showed that the CNN model based on competitive adaptive reweighting sampling (CARS) achieved the highest overall accuracy for defect discrimination (97.27%). Additionally, using 150 sound Nanfeng mandarins as subjects, quantitative predictive models for SSC were developed using full spectrum and feature wavelength-based partial least squares regression (PLSR), least squares support vector machine (LSSVM), and CNN. Among these, the best predictive model for the SSC of Nanfeng mandarins was the CNN, with R², RMSEP, and RPD values of 0.9290, 0.3772, and 3.7655, respectively. Overall, this study has demonstrated the feasibility of using hyperspectral imaging combined with deep learning for defect identification and SSC prediction in Nanfeng mandarins, providing a new method for the internal and external quality assessment of other fruits.
摘要翻译 (中文)：
水果已成为人们日常生活中不可或缺的食品，其外部缺陷和内部品质均受到销售者与消费者的高度重视。本研究结合高光谱成像技术与深度学习方法，对南丰蜜桔的外部缺陷及可溶性固形物含量（SSC）进行快速、无损检测。实验采集了具有四类缺陷（炭疽病、黑斑病、腐烂和疤痕）以及完好无损的南丰蜜桔样本在380–1030 nm波段范围内的高光谱数据。首先，提出了一种用于定性分析的端到端卷积神经网络（CNN）模型，并将其分类性能与传统分类模型进行了对比。研究中应用了三种预处理方法和三种特征选择技术。结果表明，基于竞争性自适应重加权采样（CARS）特征选择的CNN模型在缺陷识别任务中取得了最高的总体准确率，达到97.27%。此外，以150个完好南丰蜜桔为对象，分别基于全光谱和特征波长构建了偏最小二乘回归（PLSR）、最小二乘支持向量机（LSSVM）和CNN三种定量预测模型用于SSC预测。其中，CNN模型表现最优，其预测决定系数（R²）、预测均方根误差（RMSEP）和残差预测偏差（RPD）分别为0.9290、0.3772和3.7655。综上所述，本研究证实了高光谱成像结合深度学习在南丰蜜桔外部缺陷识别与SSC预测中的可行性，为其他水果内外品质的综合评估提供了一种新方法。