Klasifikasi Citra Nilai Nominal Pada Uang Kertas Tahun Emisi 2022 Untuk Membantu Tunanetra Melalui Metode Convolutional Neural Network

Muhammad, Syeikhan Al Faridizi (2024) Klasifikasi Citra Nilai Nominal Pada Uang Kertas Tahun Emisi 2022 Untuk Membantu Tunanetra Melalui Metode Convolutional Neural Network. Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text Cover_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (765kB)
	Text Abstract_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (6kB)
	Text Abstrak_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (6kB)
	Text Bab 1_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (88kB)
	Text Bab 2_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (415kB)
	Text Bab 3_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (500kB)
	Text Bab 4_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Restricted to Registered users only Download (386kB) | Request a copy
	Text Bab 5_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (9kB)
	Text Dafpus_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Download (143kB)
	Text Lampiran_20107013_Muhammad Syeikhan Al Faridzi_Skripsi_(ACC).pdf Restricted to Registered users only Download (31kB) | Request a copy

Abstract

Using the Convulutional Neural Network (CNN) technique, this study attempts to create a classification system for nominal value images on banknotes from the 2022 edition that will aid the blind. Recognizing the importance of enabling the visually impaired to independently identify banknote values, this study employs three CNN models: EfficientNetV2M, VGG16, and EfficientNetB0. 1,400 banknote photos make up the dataset 60%:40% for testing. EfficientNetV2M model had the highest accuracy at 98.04% with a Loss value of 0.0632, followed by the VGG16 model at 94.64% with a Loss value of 0.1388, and the EfficientNetB0 model at 99.82% with a Loss value of 0.0137. In summary, parameter adjustment greatly affects Accuracy and Loss numbers. The EfficientNetB0 model is the best at classifying images of banknotes with nominal values. The creation of a mobile application to assist the blind in independently identifying banknote values is one of the study's implications. Keywords: Image classification, nominal value, banknotes, visually impaired, Convolutional Neural Network, EfficientNetV2M, VGG16, EfficientNetB0.

Actions (login required)

View Item