EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Modeling the Physical Characteristics of Ultrasound Pretreated Dialium guineense Whole Fruits Using Neural Network Pattern Recognition Model

Mfrekemfon Akpan, Uduak George, Okon Esua

AgriFood: Journal of Agricultural Products for Food · 2026

DOI: 10.1002/agf2.70012 ↗

Kurz gefasst

Fachartikel in der Zeitschrift AgriFood: Journal of Agricultural Products for Food, 2026. Verfasst von Mfrekemfon Akpan und 2 weiteren Autor:innen. Die Arbeit nennt 31 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Dialium guineense is a wild fruit of African descent with rich nutritional content. The main aim of this work was therefore to model some physical characteristics of ultrasound pretreated whole wild edible fruits of Dialium guineense (DG) was modeled using Neural Network pattern recognition toolbox of Matlab 2015a. The physical characteristics of DG measured for 300 randomly selected DG fruits were: Length, width, thickness, sphericity, unit mass, and aspect ratio. These were measured using standard procedures for each physical characteristic. These 300 DG fruit was made into three groups consisting of three groups of prelabeled data of ultrasound treated DG whole fruits at 40% and 50% ultrasound power, and untreated DG whole fruits. The data was reshuffled to obtain a dataset with the three patterns evenly‐distributed, and then randomly split in the ratio of 70:15:15 for training, validation, and testing respectively. The result indicates the best validation performance at epoch 8 with the error value 0.122. The Receiver Operating Characteristic (ROC) clearly shows a model that perfectly recognizes the third pattern. Moreover, the confusion matrix distinctively reveals that the model performs at the accuracy, precision, and sensitivity of 100%. Accuracy of 84.44% is recorded for both pattern 1 and pattern 2. Precision for pattern 1 and pattern 2 is 82.34% and 78.95% respectively while a sensitivity of 83.33% and 77.78% for patterns 1 and 2 respectively is obtained by the model. The model performance can be enhanced by increasing the amount of data supplied to the neural network, as well as adjusting the number of neurons at the hidden layer of the network. The research has successfully modeled the physical characteristics of ultrasound treated wild DG whole fruits which provides an enablement for automation of its processing.

Zitieren

APA

Akpan, M., George, U., & Esua, O. (2026). Modeling the Physical Characteristics of Ultrasound Pretreated Dialium guineense Whole Fruits Using Neural Network Pattern Recognition Model. AgriFood: Journal of Agricultural Products for Food. https://doi.org/10.1002/agf2.70012

Vancouver

Akpan M, George U, Esua O. Modeling the Physical Characteristics of Ultrasound Pretreated Dialium guineense Whole Fruits Using Neural Network Pattern Recognition Model. AgriFood: Journal of Agricultural Products for Food. 2026. doi:10.1002/agf2.70012

BibTeX
@article{Akpan2026,
  author = {Akpan, Mfrekemfon and George, Uduak and Esua, Okon},
  title = {Modeling the Physical Characteristics of Ultrasound Pretreated Dialium guineense Whole Fruits Using Neural Network Pattern Recognition Model},
  journal = {AgriFood: Journal of Agricultural Products for Food},
  year = {2026},
  doi = {10.1002/agf2.70012}
}
Kostenlos starten →