<![CDATA[This study aims to analyze the performance of Convolutional Neural Network (CNN) algorithm in classifying the nominal of Rupiah banknotes issued in 2022. Three test models are developed, namely two CNN architectures with different optimizers (Adam and RMSprop), and one transfer learning model using VGG16. The dataset used consists of 1,848 banknote images of seven denominations: Rp1,000, Rp2,000, Rp5,000, Rp10,000, Rp20,000, Rp50,000, and Rp100,000. The data was collected using a smartphone camera and processed through augmentation, normalization, and classification stages. The model was evaluated using accuracy, precision, recall, and F1-score metrics. The results show that CNN with Adam's optimizer achieves a validation accuracy of 98.97%, while CNN with RMSprop reaches 99.59%. Meanwhile, the VGG16 model achieved perfect validation accuracy of 100%, with precision, recall, and F1-score values of 1.00 each. These results show that the transfer learning approach provides the best performance compared to conventional CNN models. This research supports the development of an accurate and efficient banknote recognition automation system for digital finance applications.]]>
