Loading...
4 results
Search Results
Now showing 1 - 4 of 4
- Estimation of soluble solids content and fruit temperature in 'rocha' pear using Vis-NIR spectroscopy and the spectraNet–32 deep learning architecturePublication . Martins, J. A; Rodrigues, Daniela; Cavaco, A. M.; Antunes, Maria Dulce; Guerra, Rui Manuel Farinha das NevesSpectra-based methods are becoming increasingly important in Precision Agriculture as they offer non-destructive, quick tools for measuring the quality of produce. This study introduces a novel approach for esti-mating the soluble solids content (SSC) of 'Rocha' pears using the SpectraNet-32 deep learning architecture, which operates on 1D fruit spectra in the visible to near-infrared region (Vis-NIRS). This method was also able to estimate fruit temperatures, which improved the SSC prediction performance. The dataset consisted of 3300 spectra from 1650 'Rocha' pears collected from local markets over several weeks during the 2010 and 2011 seasons, which had varying edaphoclimatic conditions. Two types of partial least squares (PLS) feature selection methods, under various configurations, were applied to the input spectra to identify the most significant wavelengths for training SpectraNet-32. The model's robustness was also compared to a similar state-of-the-art deep learning architecture, DeepSpectra, as well as four other classical machine learning algorithms: PLS, multiple linear regression (MLR), support vector machine (SVM), and multi-layer perceptron (MLP). In total, 23 different experimental method configurations were assessed, with 150 neural networks each. SpectraNet-32 consistently outperformed other methods in several metrics. On average, it was 6.1% better than PLS in terms of the root mean square error of prediction (RMSEP, 1.08 vs. 1.15%), 7.7% better in prediction gain (PG, 1.67 vs. 1.55), 3.6% better in the coefficient of determination (R2, 0.58 vs. 0.56) and 5.8% better in the coefficient of variation (CV%, 8.35 vs. 8.86).
- A TSS classification study of 'Rocha' pear (Pyrus communis L.) based on non-invasive visible/near infra-red reflectance spectraPublication . Bexiga, Florentino; Rodrigues, Daniela; Guerra, Rui Manuel Farinha das Neves; Brazio, António; Balegas, Tiago; Cavaco, A. M.; Antunes, Maria Dulce; Valente de Oliveira, JOSÉThe study focuses on the application of machine learning techniques for classifying the internal quality of 'Rocha' Pear (Pyrus communis L.), i.e., the total soluble solids (TSS), using the non-invasive technique of visible/near infra-red reflectance spectroscopy. Six representative classifiers were evaluated under realistic experimental conditions. The classifiers include representatives of classic parametric (logistic and multiple linear regression), non-parametric distance based methods (K-nearest neighbors), correlation-based (partial least squares), ensemble methods (random forests) and maximum margin classifiers (support vector machines). The classifiers were assessed against metrics such as accuracy, Cohen's Kappa, F-Measure, and the area under the precision recall curve (AUC) in a 10 x 10-fold cross-validation plan. For result analysis non-parametric statistical test of hypotheses were employed. A total of 4880 fruit samples from different origins, maturation states, and harvest years were considered. The main conclusion is that the maximum margin classifier outperforms all the others studied ones, including the commonly used partial least squares. The conclusion holds for both a reflectance spectrum with 1024 features and for a 128 subsample of these. An estimate of the out-of-sample performance for the best classifier is also provided.
- Identification of asymptomatic plants infected with Citrus tristeza virus from a time series of leaf spectral characteristicsPublication . Afonso, Andreia; Guerra, Rui Manuel Farinha das Neves; Cavaco, A. M.; Pinto, Patricia IS; Andrade, André; Duarte, Amílcar; Power, Deborah; Marques, N T.Citrus tristeza virus (CTV) affects citrus crops with differing severity, depending on the viral strain, the citrus cultivar and the scion/rootstock combinations. In this study we address the problem of identifying asymptomatic infected plants using reflectance spectra of the leaves in the visible/near infrared region. Sixteen young citrus plants (8 Citrus x clementina hort. ex Tanaka ‘Fina’ and 8 Citrus sinensis (L.) Osbeck ‘Valencia Late’) were split into control and T318A isolate infected groups. Measurements of reflectance in the 400-1100 nm range, in two leaves per plant, were performed monthly over 6 months and the presence of the virus was confirmed by IC/RT-PCR and real-time PCR. The spectra acquired in a single day of measurements was inconsistent for inoculated and control plants. However, by monitoring the same leaves over 6 months it was possible to identify infected plants on the basis of the spectra time evolution. In order to achieve this a simple unfolding implementation of 3-way PCA was applied such that group separation in the scores plot was spontaneous and not forced by any a priori assumption. The model was tested through leave-one-out cross validation with a good rate of correct classification for the left out sample. A real situation was simulated by applying the NPCA algorithm to healthy plants only and checking if the infected ones would be projected on the model scores plot as outliers. Again, a good rate of classification was obtained. Finally, we discuss the spectral features that may be associated with the clustering obtained through NPCA and their physiological significance. Reflectance measurements between infected and healthy samples of two citrus cultivars and their correlation with real-time PCR results for the presence of CTV suggest reflectance spectra of the leaves in the visible/near infrared region is a promising tool for plant stress monitoring linked to the presence of CTV infection prior to symptom expression.
- Determination of the botanical origin of honey by sensor fusion of impedance e-tongue and optical spectroscopyPublication . Ulloa, P. A.; Guerra, Rui Manuel Farinha das Neves; Cavaco, A. M.; Costa, Ana M. Rosa da; Figueira, A.C.; Fernandes, A.The aim of this study was to discriminate four commercial brands of Portuguese honeys according to their botanical origin by sensor fusion of impedance electronic tongue (e-tongue) and optical spectroscopy (UV–Vis–NIR) assisted by Principal Component Analysis (PCA) and Cluster Analysis (CA). We have also introduced a new technique for variable selection through one-dimensional clustering which proved very useful for data fusion. The results were referenced against standard sample identification by classical melissopalynology analysis. Individual analysis of each technique showed that the e-tongue clearly outperformed the optical techniques. The electronic and optical spectra were fitted to analytical models and the model coefficients were used as new variables for PCA and CA. This approach has improved honey classification by the e-tongue but not by the optical methods. Data from the three techniques was then considered simultaneously. Simple concatenation of all matrices did not improve the classification results. Multi-way PCA (MPCA) proved to be a good option for data fusion yielding 100% classification success. Finally, a variable selection method based on one-dimensional clustering was used to define two new approaches to sensor fusion, and both yielded sample clusters even better defined than using MPCA. In this work we demonstrate for the first time the feasibility of sensor fusion of electronic and optical spectroscopy data and propose a new variable selection method that improved significantly the classification of the samples through multivariate statistical analysis.