Search

The objective of this research was to classify the geographical origin of durians (cv. Monthong) based on geographical identification (GI) and regions (R) using near infrared (NIR). The samples were scanned with an FT-NIR spectrometer (12,500 to 4000 cm−1). The NIR absorbance differences among samples that were collected from different parts of the fruit, including intact peel with thorns (I-form), cut-thorn peel (C-form), stem (S-form), and the applied synthetic minority over-sampling technique (SMOTE), were also investigated. Models were developed across several classification algorithms by the classification learner app in MATLAB. The models were optimized using a featured wavenumber selected by a genetic algorithm (GA). An effective model based on GI was developed using SMOTE-I-spectra with a neural network; accuracy was provided as 95.60% and 95.00% in cross-validation and training sets. The test model was provided with a testing set value of %accuracy, and 94.70% by the testing set was obtained. Likewise, the model based on the regions was developed from SMOTE-ICS-form spectra, with the ensemble classifier showing the best result. The best result, 88.00FF% accuracy by cross validation, 86.50% by training set, and 64.90% by testing set, indicates the classification model of East (E-region), Northeast (NE-region), and South (S-region) regions could be applied for rough screening. In summary, NIR spectroscopy could be used as a rapid and nondestructive method for the accurate GI classification of durians. [ABSTRACT FROM AUTHOR]

The aim of this study was to evaluate and compare the performance of multivariate classification algorithms, specifically Partial Least Squares Discriminant Analysis (PLS-DA) and machine learning algorithms, in the classification of Monthong durian pulp based on its dry matter content (DMC) and soluble solid content (SSC), using the inline acquisition of near-infrared (NIR) spectra. A total of 415 durian pulp samples were collected and analyzed. Raw spectra were preprocessed using five different combinations of spectral preprocessing techniques: Moving Average with Standard Normal Variate (MA+SNV), Savitzky–Golay Smoothing with Standard Normal Variate (SG+SNV), Mean Normalization (SG+MN), Baseline Correction (SG+BC), and Multiplicative Scatter Correction (SG+MSC). The results revealed that the SG+SNV preprocessing technique produced the best performance with both the PLS-DA and machine learning algorithms. The optimized wide neural network algorithm of machine learning achieved the highest overall classification accuracy of 85.3%, outperforming the PLS-DA model, with overall classification accuracy of 81.4%. Additionally, evaluation metrics such as recall, precision, specificity, F1-score, AUC ROC, and kappa were calculated and compared between the two models. The findings of this study demonstrate the potential of machine learning algorithms to provide similar or better performance compared to PLS-DA in classifying Monthong durian pulp based on DMC and SSC using NIR spectroscopy, and they can be applied in the quality control and management of durian pulp production and storage. [ABSTRACT FROM AUTHOR]

The DM and TSS of durian pulp moving on a conveyor belt were measured for their rapid and non-destructive qualities based on quantitative and qualitative measurements. The calibration set and prediction set equaled 209 and 69 pulps, respectively. The quantitative test compared the performance of PLS regression for DM and TSS prediction developed from full wavelength (860–1754 nm) and a few significant variables using SPA, GA, and VIP methods. The qualitative test identified the possibility of maturity stage classification by comparing three supervised machine learning classifiers, namely SVM, random forest (RF), and LDA. Effective models for DM and TSS prediction were developed from second derivatives spectra combined with the GA method, exhibiting r2, SEP, and RPD of 0.85, 4.50%, and 2.64, respectively for DM, and 0.66, 5.15%, and 1.60, respectively, for TSS. The model classifying samples into two distinct groups, namely "reject" and "pass," utilizing the LDA algorithm, exhibited an impressive accuracy rate of 94.20%, making it a suitable choice for quality assurance purposes. This result indicates that the few effective variables were more efficient than full wavelength and improved model accuracy with greater model stability. Enhancing the classification model could involve data sample balancing in each group, leading to further improvements. • NIR applied for predicting DM and TTS in durian pulp. • NIR applied for durian pulp maturity classification based on DM and TTS content. • PLSR coupled with variable selection of SPA, GA, and VIP methods were developed. • Classification ability of SVM, LDA and random forest (RF) was compared. • NIR could classify the quality stage with on line measurement. [ABSTRACT FROM AUTHOR]

• Rapid classification of levels of N, P, and K in durian leaves via NIR spectroscopy was investigated. • Models developed regarding fresh and dried ground leaf spectra were compared. • The classification model based on ANN was optimum compared with those involving PLS and kNN. • The SMOTE to balance data could increase classification accuracy. For durian growth to produce high-quality fruit, plants should receive sufficient nutrients. Currently, farmers apply various fertilisers to produce a large quantity and quality of durian fruit, irrespective of the actual nutrients that the plant requires. Accordingly, the production cost is high and non-renewable resources. Therefore, this study focused on rapid classification primary macronutrient levels in durian (Durio zibethinus Murray CV. Mon Thong) leaves using Fourier transform near-infrared (FT-NIR) spectroscopy and investigated the effect of imbalanced data on efficient classification models. Contents of N, P, and K in durian leaves were measured via NIR with the wavelength range of 800–2,500 nm. Classification models were developed using partial least squares, k-nearest neighbour, and artificial neural networks (ANNs) with imbalanced and balanced data. The imbalanced data were balanced using a synthetic minority oversampling technique (SMOTE). In this study, the model regarding the fresh leaf sample performed better than that for the dried ground leaf sample. Moreover, the ANN was the best algorithm, exhibiting validation accuracies of classified levels corresponding to N = 0.99 and P = 0.97 when the data were analysed with SMOTE and K = 1.00 from the original balanced data. The imbalanced data affected biased classification when the models could increase the classification accuracy by applying balanced data for modelling. [ABSTRACT FROM AUTHOR]

• DMC and SSC analysis of durian pulp using in-line conveying system. • Scanning in different orientation of durian affects the prediction model. • PLSR model was suitable for the prediction of DMC and SSC. The prediction of dry matter content (DMC) and soluble solids content (SSC) in durian pulp were performed using a small laboratory scale in-line visible and short wave near infrared (Vis-SWNIR) spectroscopic system. The fiber optic diode array spectrometer with a charged coupled device (CCD) detector in a wavelength range of 450−1000 nm was used for spectral data acquisition. The spectra of the sample were acquired on the moving conveyor belt in two different orientations, including scanning in the upright position of pulps collected in 2018 and the stable position by scanning on the side of the pulps collected in 2019. Partial least squares regression (PLSR) was used to establish the relationship between the spectra and observed DMC and SSC values using the different wavelength ranges, including 450−1000, 700−1000, and 800−1000 nm for the comparison. The results showed that the durian pulp should be scanned in the upright position at the center of the pulp. Moving average smoothing preprocessing combined with the standard normal variate (SNV) for DMC and multiple scatter correction (MSC) for SSC gave the best result. The suitable wavelength range for model development to predict the DMC and SSC was 700−1000 nm and 800−1000 nm, respectively. After comparing the results, the optimum model showed the coefficient of determination of calibration (R C 2), and prediction (R P 2), root mean square error of prediction (RMSEP), bias, and the ratio of performance to interquartile distance (RPIQ) of 0.88, 0.83, 4.32 %, 1.25 %, and 3.52 for DMC and 0.70, 0.70, 4.0 %, 0.4 %, and 2.2 for SSC prediction. [ABSTRACT FROM AUTHOR]