Hidayah, Aisyah Nurul (2020) Deteksi Huruf Pada Tulisan Tangan Latin Menggunakan Metode You Only Look Once (YOLO). Other thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
07211640000053-Undergraduate_Thesis.pdf Download (10MB) | Preview |
Abstract
Tulisan tangan saat ini masih terlibat dalam berbagai bidang, khususnya tulisan tangan latin. Tulisan tangan latin memiliki beberapa tantangan. Pertama, tulisan tangan memiliki banyak variasi. Variasi yang dimaksud adalah ukuran huruf, kemiringan huruf, baseline atau garis dasar tulisan, tekanan penulisan, bahkan jarak antarhuruf dan antarkata tulisan [1]. Kedua, setiap huruf pada tulisan tangan latin menyambung menjadi satu. Tantangan ini menyebabkan proses pengenalan tulisan latin memakan waktu yang lebih lama. You Only Look Once (YOLO) digunakan untuk mendeteksi huruf pada tulisan tangan latin sebagai cara agar proses pengenalan tulisan menjadi lebih cepat. Percobaan ini dilakukan dengan mengumpulkan sampel citra dari 4 jenis tulisan tangan yang berbeda. Hasil percobaan menggunakan GPU NVIDIA GeForce RTX 1050 menunjukkan bahwa processing time per image untuk deteksi huruf pada citra satu kata ialah 0.0776 detik dengan nilai threshold confidence score 0.3.
==================================================================================================================================
Handwriting is currently still involved in various fields, specifically Latin handwriting. Latin handwriting has several challenges. First, handwriting has many variations. Variations discussed are letter size, the slope of the letter, baseline or the baseline of writing, even the different distance between letters and the words [1]. Second, each letter in Latin handwriting links together. This challenge led to a longer time for the recognition of Latin handwriting. You Only Look Once (YOLO) is used to detect letters in Latin handwriting as a way to make the recognition process of Latin handwriting faster. This experiment carried out by collecting sample images of 4 different handwritten types. The experimental result using GPU NVIDIA GeForce RTX 1050 shows processing time per image for the detection system in one-word images is 0.0776 seconds using a threshold score of 0.3.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | tulisan, deteksi, yolo, handwriting, detection. |
Subjects: | Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Computer Engineering > 90243-(S1) Undergraduate Thesis |
Depositing User: | Aisyah Nurul Hidayah |
Date Deposited: | 23 Aug 2020 13:45 |
Last Modified: | 04 Jul 2023 16:30 |
URI: | http://repository.its.ac.id/id/eprint/79263 |
Actions (login required)
![]() |
View Item |