Transliterasi Aksara Jawa – Latin Berbasis Model Deteksi Menggunakan Convolutional Neural Network

Asti, Dwi Sripamuji (2023) Transliterasi Aksara Jawa – Latin Berbasis Model Deteksi Menggunakan Convolutional Neural Network. Undergraduate Thesis thesis, Institut Teknologi Telkom Purwokerto.

	Text COVER_Laporan TA Asti_API_IQA_PERFECT.pdf Download (2MB)
	Text ABSTRACT_Laporan TA Asti - API - IQA.pdf Download (266kB)
	Text ABSTRAK_Laporan TA Asti - API - IQA.pdf Download (268kB)
	Text BAB 1_Laporan TA Asti - API - IQA.pdf Download (174kB)
	Text BAB 2_Laporan TA Asti - API - IQA.pdf Download (282kB)
	Text BAB 3_Laporan TA Asti - API - IQA.pdf Download (610kB)
	Text BAB 4_Laporan TA Asti - API - IQA.pdf Restricted to Registered users only Download (513kB) | Request a copy
	Text BAB 5_Laporan TA Asti - API - IQA.pdf Download (165kB)
	Text DAFTAR PUSTAKA_Laporan TA Asti - API - IQA.pdf Download (293kB)
	Text LAMPIRAN_Laporan TA Asti - API - IQA.pdf Restricted to Registered users only Download (2MB) | Request a copy

Abstract

Language is a tool for conveying ideas and thoughts. The writing and representation of language is indicated by certain letters/characters. The script from one region to another in Indonesia has its own characteristics. Javanese script is one of the regional scripts that has a complex way of writing. An automatic Javanese script detection model is needed so that the introduction of Javanese characters can be done easily. The detection model was built in this study where the model can change the image of the Javanese script into the Latin text. The literature review stage led the researcher to find a Javanese script detection model which was still limited to a single character. In addition, the researchers also found that the regional character detection model gave a small accuracy score. This study intends to build a detection model for Javanese script that is equipped with sandhangan automatically using one of the deep learning algorithms, namely the Convolutional Neural Network. The research results obtained, the Conv2D model (64, 32, 16) has superior performance as indicated by an training accuracy of 99.87% and testing accuracy of 89%. Keywords: Javanese Script, Detection, Deep Learning, Convolutional Neural Network

Actions (login required)

View Item