Prediction of Strawberries’ Quality Parameters Using Artificial Neural Networks

Abstract

Strawberry is a very popular fruit, appreciated for its unique flavor and many beneficial traits such as antioxidants and useful amino acids, which strongly contribute to the overall quality of the product. Indeed, the quality of fresh fruit is a fundamental aspect for consumers, and it is crucial for the success of breeding activities as well as for enhancing the competitiveness and profitability of the fruit industry. Nowadays, the entire supply chain requires simple and fast systems for quality evaluation. In this context, the pomological and chemical traits (i.e., soluble solids, firmness, titratable acidity, dry matter) as well as nutritional ones such as total phenols, total anthocyanins and antioxidant potential were evaluated and compared for seven strawberry cultivars and three harvest times. The prediction of the qualitative traits was carried out using color space coordinates (L*, a* and b*) and two statistical techniques, i.e., the multiple linear regression models (MLR) and artificial neural networks (ANNs). Unsatisfactory prediction performances were obtained for all parameters when MLR was applied. On the contrary, the good prediction of the internal quality attributes, using ANN, was observed, especially for both antioxidant activity and the total monomeric anthocyanin (R2 = 0.906, and R2 = 0.943, respectively). This study highlighted that color coordinates coupled with ANN can be successfully used to evaluate the quality of strawberry.

Physico-Chemical Properties Prediction of Flame Seedless Grape Berries Using an Artificial Neural Network Model

“…The signal from the hidden layer to the output layer was also subject to the same procedure. The hidden ( x i ) and output ( y i ) neuron activities are defined as follows [ 28 ]: where the activation function used in the output or hidden layers is f ( v i ). Due to its improved performance, feed-forward neural network models typically use entire topologies with hidden layers and an active transfer function, which is the sigmoid in this study [ 38 ].…”

“…When compared to traditional sample-based measurements, image processing has the ability to quickly capture a fruit’s dimension and color. Additionally, machine learning methods, such as artificial neural networks (ANN), have been used to predict fruit quality [ 28 ]. The development of methods for evaluating the quality of fruits and vegetables has received significant attention recently, and tools of quality assessment have gained momentum [ 29 ].…”

“…Moreover, different studies examined ANN for determining the quality of different fruit and vegetables with the aid of some features that include fruit color, size, etc. [ 28 , 30 ]. Although determining the quality of any food material is a challenging task, ANN technologies and their intelligent capabilities make them both highly accurate and economically advantageous [ 29 ].…”

“…An ANN’s powerful computing power lets it treat more data and solve many identical multifaceted nonlinear problems [ 30 ]. As an example, Amoriello et al [ 28 ] collected data for seven strawberry varieties and created two statistical procedures, i.e., multiple linear regression (MLR) and ANN models, to predict titratable acidity, dry matter, soluble solids, and firmness as well as nutritional attributes, such as antioxidant potential, total anthocyanins, and total phenols. The inputs to the models were color space coordinates, which were denoted by L*, a*, and b* of strawberries.…”

“…Moreover, the quality attributes of Flame Seedless grape berries can be analyzed qualitatively using laboratory analysis. However, many previous research papers [ 28 , 30 ] have specified that the ANN technique is a very operative and consistent predicting tool, which has been extensively used in fruit quality field and has a very high prediction precision. The present research aimed to notice the dependability of the ANN model to predict the berry weight, berry length, and berry diameter as well as berry adherence strength and chemical properties (i.e., anthocyanin, total soluble solids (TSS), TSS/titratable acidity, total sugars, titratable acidity, reducing sugars, and non-reducing sugars) of Flame Seedless grape berries and to discover the effects of CIELab colorimeter parameters of berries L*, a*, and b* on the key fruit quality attributes of them.…”