Rancang Bangun Deteksi Nominal Uang Kertas Rupiah Berbasis Deep Learning Pada Perangkat Android

Permana, Amri Abyad (2023) Rancang Bangun Deteksi Nominal Uang Kertas Rupiah Berbasis Deep Learning Pada Perangkat Android. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 07111940000076-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (2MB) | Request a copy

Abstract

Tunanetra merupakan kondisi dimana seseorang mengalami gangguan atau keterbatasan pada indra penglihatannya. Berdasarkan estimasi Kementrian Kesehatan Republik Indonesia, sebanyak 3.750.000 atau sekitar 1% dari penduduk Indonesia adalah penyandang tunanetra. Dalam melakukan aktivitas sehari-hari manusia tidak jauh dari kegiatan jual beli, ada berbagai macam metode pembayaran yang dapat digunakan saat melakukan transaksi, salah satu yang paling umum dijumpai adalah menggunakan uang kertas. Bagi orang dengan penglihatan normal, uang kertas sangatlah mudah digunakan. Namun berbeda dengan penyandang tunanetra yang harus mengenali uang tersebut sesuai dengan nominal yang dibutuhkan. Pada tugas akhir ini dibuat suatu aplikasi pendeteksi nominal uang kertas Rupiah berbasis deep learning yang dapat dijalankan pada perangkat android dan dapat memberikan output berupa suara sehingga mudah digunakan oleh penyandang tunanetra. Sistem deteksi nominal uang dapat dilakukan dengan metode Deep learning menggunakan arsitektur Deep Neural network (DNN). Terdapat dua arsitektur yang digunakan yaitu arsitektur YOLOv5 dan arsitektur DNN yang dibuat pada tugas akhir ini, arsitektur ini tersusun dari beberapa layer seperti Convolutional2D, Separable Convolutional 2D, Maxpooling, Global Average Pooling dan dense layer serta layer aktivasi seperti ReLU dan Softmax. Hasil penelitian menunjukan tingkat akurasi model YOLOv5n yang telah dikonversi menjadi model Tensorflow lite sebesar 99% dengan waktu inferensi rata-rata 250.26 ms, kemudian tingkat akurasi model DNN setelah dikonversikan menjadi model tensorflow lite sebesar 96% dengan waktu inferensi rata-rata 215.86 ms. ================================================================================================================================
Vision impairment or blindness is a condition where someone has limitations on the sense of sight. Based on estimates from the Indonesian Ministry of Health, As many as 3.750.000 people, or around 1% of Indonesia's population are visually impaired. In carrying out daily activities, humans are not far from transactions. There are various kinds of payment methods, one of the most commonly found is using cash. People with normal vision can easily make transactions with cash, but it is not easy for visually impaired people who have to differentiate and recognize the cash amount first. Therefore, this undergraduate thesis designed a currency recognition system based on an android smartphone. In this final project, an application for detecting denominations of Indonesian Rupiah banknotes based on deep learning was developed. The application can be run on Android devices and provides audio output, making it easy to use for visually impaired individuals. The money detection system utilizes deep learning with the Deep Neural Network method. There are two architectures used in the application, namely YOLOv5 architecture and DNN architecture developed for this final project. The CNN architecture consists of several layers such as Convolutional2D, Separable Convolutional 2D, Maxpooling, Global Average Pooling, dense layer, and activation layers such as ReLU and Softmax. The research results show that the accuracy level of the YOLOv5n model, which has been converted into a TensorFlow Lite model, is 99%, with an average inference time of 250.26 ms. Furthermore, the accuracy level of the DNN model, after being converted into a TensorFlow Lite model, is 96%, with an average inference time of 215.86 ms.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Accessibility, Deep learning, Currency Recognition, Vision impairment, YOLOv5, Aksesibilitas, Pengenalan Uang, Tunanetra
Subjects:	Q Science > Q Science (General) > Q325.5 Machine learning. Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.774.A53 Android Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Electrical Engineering > 20201-(S1) Undergraduate Thesis
Depositing User:	Amri Abyad Permana
Date Deposited:	24 Jul 2023 06:32
Last Modified:	24 Jul 2023 06:32
URI:	http://repository.its.ac.id/id/eprint/99100

Actions (login required)

View Item