The aim of this work was to investigate how sampling of vegetal material affects the collection of NIR spectra for building a multivariate discriminant model for Prunus dulcis varietal classification. ConclusionsThe results indicated that variety was the most important factor for classification. The spectral pre-treatment that provided the best results was a combination of standard normal variate (SNV), Savitzky-Golay first derivative, and mean-centering methods. With regard to the type of processed sample, the highest percentages of correct classifications were obtained with fresh and dried powdered leaves at both the training set and test set validation levels. Points of studySources of variation ❖ Position of the leaves in the trees (young/adult leaves) ❖ Differences among trees of the same variety ❖ Differences at varietal level Processed samples: ❖ Fresh leaves (direct measurement) ❖ Dried leaves (65 o C for 48h) ❖ Dried-powdered leaves (65 o C for 48h + grinding) Experimental Results Raw spectra Material and Methods ❖ Three varieties of Prunus dulcis (Avijor, Guara, and Pentacebas) ❖ Principal component analysis (PCA) ❖ Partial least-squares discriminant analysis (PLS-DA) ❖ ANOVA simultaneous component analysis (ASCA) ❖ Antaris II FT-NIR analyzer (Thermo Scientific, USA)

The emergence of new almond tree (Prunus dulcis) varieties with agricultural interest is forcing the nursery plant industry to establish quality systems to keep varietal purity in the production stage. The aim of this study is to assess the capability of near-infrared spectroscopy (NIRS) to classify different Prunus dulcis varieties as an alternative to more expensive methods. Fresh and dried-powdered leaves of six different varieties of almond trees of commercial interest (Avijor, Guara, Isabelona, Marta, Pentacebas and Soleta) were used. The most important variables to discriminate between these varieties were studied through of three scientifically accepted indicators (Variable importance in projection¸ selectivity ratio and vector of the regression coefficients). The results showed that the 7000 to 4000 cm−1 range contains the most useful variables, which allowed to decrease the complexity of the data set. Concerning to the classification models, a high percentage of correct classifications (90–100%) was obtained, where dried-powdered leaves showed better results than fresh leaves. However, the classification rate of both kinds of leaves evidences the capacity of the near-infrared spectroscopy to discriminate Prunus dulcis varieties. We demonstrate with these results the capability of the NIRS technology as a quality control tool in nursery plant industry.

Computer vision coupled to deep learning is a promising technique with multiple applications in the industry. In this work, the potential of this technique has been assessed in the classification of two varieties of almond trees (Prunus dulcis), Soleta and Pentacebas. For that, a convolutional neural network named VGG16 was used. The most appropriate configuration for model training was studied, which included the comparison between two different filling modes (reflect and nearest) in the data augmentation step, the evaluation of the batch size and the analysis of the image sizes. The robustness of the model was also checked, and information was obtained about how the model extracts the information from the images. The results showed that the reflect fill mode was more effective than the nearest one. The best results were obtained using batches with 30 and 40 images, with an image size of (224 × 224) pixels. The verification of the robustness proved the capability of the technique as a promising tool for plant varietal identification.

Varietal control to avoid unwanted varietal mixtures is an important objective for the nursery plant industry. In this study, we have developed and analyzed the capabilities of a computer vision system based on deep learning for the control of plant varieties in the nursery plant industry and for evaluating its capabilities. For this purpose, three datasets of nursery plant images were compared. The datasets came from two varieties of almond trees (Prunus dulcis) named Soleta and Pentacebas. Each dataset contained images with three different scales: whole plant, leaf, and venation. The Gradient-weighted Class Activation Mapping (Grad-CAM) technique was used to unveil the most important features to discriminate between both varieties. The three datasets provided classification accuracies above 97% in the test set, being the leaf dataset, with a 98.8% accuracy, the one providing the best results. Concerning the most important features of the plants, the Grad-CAM showed that they are located in the center of the leaf, that is, the venation. In conclusion, we have shown that computer vision is a promising technique for the control of plant varietal mixtures.

The Contributor assigns to Wiley-Blackwell, during the full term of copyright and any extensions or renewals, all copyright in and to the Contribution, and all rights therein, including but not limited to the right to publish, republish, transmit, sell, distribute and otherwise use the Contribution in whole or in part in electronic and print editions of the Journal and in derivative works throughout the world, in all languages and in all media of expression now known or later developed, and to license or permit others to do so.