Klasifikasi Uang Rupiah Kertas Tidak Layak Edar Menggunakan CNN Xception Transfer Learning Berbasis Website

Abstract

- Rupiah banknotes are the main means of payment used by the public, but the lack of public knowledge regarding their maintenance and appropriateness characteristics causes damage to the Rupiah currency. Bank Indonesia is trying to overcome this problem with the "Love Proudly Understand the Rupiah" campaign, but it will be difficult to reach the entire community with this education alone. Therefore, a system was developed "Classification of Rupiah Currency Unfit for Circulation using Website-based CNN time which has high accuracy for image classification, producing an accurate model with a short training time. Using a dataset of 14 classes of 2016 emission Rupiah currency, including 7 eligible and non-eligible denominations. Final results show 99.22% accuracy for training, 96.19 % for validation, and 93.57% for testing, in addition to developing deep learning methods, this model will be implemented on the website, which aims to make it easier and help the public to find out the suitability of the Rupiah banknotes they have.