A New Non-Destructive Multidimensional Yield Determination Method Approach for Walnut Crop

Authors : Remzi Gürfidan, Enes Açıkgözoğlu, Mevlüt Ersoy

Pages : 246-255

Doi:10.31202/ecjse.1435709

View : 47 | Download : 21

Publication Date : 2024-09-17

Article Type : Research Paper

Abstract :Walnut has an important place in agricultural production and research on it covers various fields. In this study, machine learning algorithms were used for non-destructive estimation of walnut productivity. The researchers developed a setup using audio recordings and images to determine the fullness and void status of walnuts. These data were processed with various machine learning algorithms and the results were evaluated. The algorithms used in the study include RESNET50, DenseNET121, VGG16 and CNN. However, when the results obtained are analyzed, it is seen that the VGG16 algorithm gives the most successful results with 99.79% accuracy and 91.42% val_accuracy values using imagenet weights. These results were found to be quite successful compared to similar studies in the literature. In future studies, it is aimed to expand the obtained dataset and increase the val_accuracy value even more. In addition, similar methods are planned to be applied on other nuts such as hazelnuts and almonds. This could be an important step to increase productivity in agricultural production. In conclusion, this study on walnut yield estimation using non-destructive methods offers a new and effective approach in agricultural applications. The use of machine learning algorithms offers potential in various areas such as increasing productivity in walnut production and detecting diseases.
Keywords : walnut yield determination, non destructive yield determination, machine learning, nut non destructive yield calculation

ORIGINAL ARTICLE URL