<![CDATA[Banana fruit quality is an important factor in meeting consumer demand and maintaining product quality in the supply chain. The development of automatic methods for classifying the quality of bananas is becoming increasingly important as the worldwide consumption of bananas grows. In this study, we propose a classification method for banana fruit quality using the Stochastic Gradient Descent (SGD) algorithm. This study aims to evaluate the performance of SGD in classifying the quality of bananas and to analyze the effect of selecting hyperparameters on the classification results. The dataset collected is a dataset containing pictures of bananas with various levels of ripeness and conditions. This dataset is used to train and test a classification model using SGD. During the experiment, hyperparameter tuning processes such as learning rate, momentum, and batch size were carried out to understand how these parameters affect the performance of SGD in classification. We report the results of evaluating the classification based on accuracy and analyze changes in performance with variations in hyperparameters. The results of this study indicate that SGD has the potential to classify the quality of bananas, where the optimal SGD model obtained a classification accuracy of 99.9%, compared to the standard SGD model which only obtained a classification accuracy of 94.7%.]]>
Copyrights © 2023
