Table 1.
Camera settings used to obtain spinach images.
Fig 1.
Spinach freshness level guidelines for sensory evaluation.
Fig 2.
Flow chart of the algorithm used to evaluate the color and local features of spinach samples extracted from smartphone images.
Fig 3.
Twelve accuracies were calculated for the individual panels. The red and blue numbers mean correct and incorrect answers. The true label is the rounded value of the average from all panel evaluations.
Fig 4.
Test set confusion matrices of 5-fold cross validated SVM classifier and ANN models.
True label is the mean value of the sensory evaluation: (a) a four-level classification, (b) a three-level classification, and (c) a two-level classification by SVM; (d) a four-level classification, (e) a three-level classification, and (f) a two-level classification by ANN.
Fig 5.
Accuracy sensory evaluation for test set by a panel and machine learning model.
The true label is the rounded value of the average from all panel evaluations. Boxplot is the results of 12 non-trained panel, where whiskers indicate the minimum and maximum of the accuracy. Red and blue points: accuracy by SVM and ANN, respectively.