PENGENALAN KARAKTER TULISAN TANGAN HANGEUL MENGGUNAKAN ALGORITMA HYBRID VISION TRANSFORMER (ViTs) DAN CONVOLUTIONAL NEURAL NETWORK (CNN)

Kezia, Kezia (2024) PENGENALAN KARAKTER TULISAN TANGAN HANGEUL MENGGUNAKAN ALGORITMA HYBRID VISION TRANSFORMER (ViTs) DAN CONVOLUTIONAL NEURAL NETWORK (CNN). Undergraduate thesis, Universitas Pembangunan Nasional Veteran Jawa Timur.

	Text (Cover) 20081010258.-cover.pdf Download (3MB)
	Text (bab1) 20081010258.-bab1.pdf Download (1MB)
	Text (bab2) 20081010258.-bab2.pdf Restricted to Repository staff only until 20 September 2026. Download (2MB)
	Text (bab3) 20081010258.-bab3.pdf Restricted to Repository staff only until 20 September 2026. Download (4MB)
	Text (bab4) 20081010258.-bab4.pdf Restricted to Repository staff only until 20 September 2026. Download (2MB)
	Text (bab5) 20081010258.-bab5.pdf Download (1MB)
	Text (daftarpustaka) 20081010258.-daftarpustaka.pdf Download (1MB)

Abstract

The increase in the number of Indonesian tourists visiting South Korea, which reached 110,000 visitors from January to June 2023, reflects the growing interest in Korean culture driven by the Korean Wave phenomenon. However, this increase in tourism also presents challenges, especially for those who do not understand Korean characters (Hangeul), making them vulnerable to fraudulent practices. According to data from South Korea's national police force, 67.8% of the 320 complaints filed were related to consumer issues, including fraud and discriminatory services against tourists. This research aims to develop a handwritten Hangeul character recognition system to assist tourists in understanding essential information. The system proposes a hybrid algorithm combining Vision Transformer (ViTs) and Convolutional Neural Network (CNN). While CNN demonstrates high accuracy on training data, it has limitations in capturing global relationships between image features. The integration of ViTs seeks to enhance the system's flexibility and adaptability by utilizing a self-attention mechanism that can grasp the global context within images. The testing results show that the hybrid ViTs and CNN model achieved 97% accuracy in classifying handwritten Hangeul characters. This high accuracy demonstrates the model's ability to accurately recognize and differentiate each character. Thus, the model effectively addresses the challenges of handwritten Hangeul recognition, which requires precise attention to the small details of each character, providing a significant solution to improving the experience of tourists in South Korea.

Item Type:

Thesis (Undergraduate)

Contributors:

Contribution	Contributors	NIDN/NIDK	Email
Thesis advisor	Sari, Anggraini Puspita	NIDN0716088605	anggraini.puspita.if@upnjatim.ac.id
Thesis advisor	Maulana, Hendra	NIDN1423128301	hendra.maulana.if@upnjatim.ac.id

Subjects:

T Technology > T Technology (General)

Divisions:

Faculty of Computer Science > Departemen of Informatics

Depositing User:

Kezia Kezia

Date Deposited:

20 Sep 2024 02:47

Last Modified:

20 Sep 2024 02:47

URI:

https://repository.upnjatim.ac.id/id/eprint/29197