Date Fruit Classification by Using Image Features Based on Machine Learning Algorithms

Abstract

The date fruit, scientifically known as Phoenix dactylifera, is a significant dietary component due to its high nutritional value and abundance of essential vitamins and minerals. The process of discerning the classification of this fruit, which exhibits a multitude of variations within its natural domain, needs a specialized skill set. The automated recognition of species based on images of agricultural goods has gained significant prevalence in recent times. In this objective, the present study employed machine learning algorithms to automatically identify seven types of date fruit. In the investigation, decision tree, K-nearest neighbor, artificial neural networks, and support vector machine through their different hyperparameters are employed for the purpose of classifying date fruit. The dataset was divided into ratios of 80% and 20% for training and testing, respectively, and the training process employed the five-fold cross-validation technique to avoid overfitting. In summary, the results indicate that the best algorithm is neural network with a layer size of 25. In this study, this proposed algorithm achieved a test accuracy rate of 93.85%. Given the absence of computational complexity in the investigation, it can be effortlessly incorporated into diverse tools, thereby facilitating the identification of the types of date fruit.

Keywords

• otomatik tespit
• sınıflandırma
• hurma meyveleri
• makine öğrenimi algoritmaları
• sinir ağları
• automatic detection
• classification
• date fruit
• machine learning algorithms
• neural networks