Klasifikasi Kualitas Buah dengan Menggunakan Convolutional Neural Network (Studi Kasus: Dataset Fresh and Rotten Classification)

Abstract

This research aims to develop a deep learning model for fruit quality classification using Convolutional Neural Network (CNN) with the Fresh and Rotten Classification dataset. Two CNN models are compared, with the first model serving as the baseline and the second model resulting from parameter tuning based on the first model. The results indicate that increasing the number of epochs improves the model accuracy, as evidenced by the first model achieving 91% accuracy with 10 epochs and 93% accuracy with 15 epochs. Similar patterns are observed in the second model, with 87% accuracy at 10 epochs and 90% accuracy at 15 epochs. Despite the second model involving the addition of layers and parameters, its accuracy tends to be lower compared to the first model. The research emphasizes that increasing the number of epochs enhances model performance, while adding layers does not always lead to significant improvements, depending on the model's complexity and dataset characteristics. The first model, trained with 15 epochs, demonstrates the highest accuracy, approaching results from similar previous studies. This evaluation provides valuable insights for developing a CNN-based fruit classification model on the Fresh and Rotten Classification dataset.

Keywords: Fruit Classification, Rotten, Fresh, Convolutional Neural Network, Accuracy, Epochs