Mayona, Dela Ayu Putri (2026) Javanese Character Recognition Using MobileNetV3 Feature Extraction and Support Vector Machine (SVM). Undergraduate thesis, UPN Veteran Jawa Timur.
|
Text (cover)
22081010008.-cover.pdf Download (4MB) | Preview |
|
|
Text (bab 1)
22081010008.-bab1.pdf Download (985kB) | Preview |
|
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (bab 2)
22081010008.-bab2.pdf Restricted to Repository staff only until 10 June 2029. Download (5MB) |
|
|
Text (bab 3)
22081010008.-bab3.pdf Restricted to Repository staff only until 10 June 2029. Download (7MB) |
|
|
Text (bab 4)
22081010008.-bab4.pdf Restricted to Repository staff only until 10 June 2029. Download (46MB) |
|
|
Text (bab 5)
22081010008.-bab5.pdf Download (516kB) | Preview |
|
|
Text (daftar pustaka)
22081010008.-daftarpustaka.pdf Download (928kB) | Preview |
|
|
Text (lampiran)
22081010008.-lampiran.pdf Restricted to Repository staff only until 10 June 2029. Download (220kB) |
Abstract
Javanese script is one of Indonesia’s cultural heritages that holds significant historical and educational value. One of the main challenges in preserving Javanese script lies in the complexity of its characters, many of which have visually similar shapes. This study aims to develop an image-based recognition system for Javanese script by utilizing MobileNetV3 for feature extraction and SVM as the classification algorithm. The dataset used consists of 2,500 images of Javanese characters, with 125 images for each class. The preprocessing stage includes the application of Contrast Limited Adaptive Histogram Equalization (CLAHE) to enhance image contrast, resizing images to 224×224 pixels, and normalizing pixel values. Feature extraction is performed using the MobileNetV3 architecture, which is capable of producing efficient and informative feature representations. The SVM model is then trained using various kernel types with different combinations of parameters C and gamma, along with multiple data splitting scenarios to obtain the best performance. The experimental results show that the best model is achieved using a Linear kernel with C = 0.1 and no gamma parameter under an 80:20 data split, resulting in an accuracy of 98.4%, precision of 98.49%, recall of 98.4%, and F1-score of 98.40%. These results indicate that the combination of MobileNetV3 and SVM is effective in recognizing Javanese script. The best-performing model is further implemented into a web-based application using Flask to enable interactive user testing. This study is expected to serve as a foundation for the development of technology-based learning systems for Javanese script.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science | ||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Informatics | ||||||||||||
| Depositing User: | Dela Ayu Putri Mayona Dela | ||||||||||||
| Date Deposited: | 10 Jun 2026 08:28 | ||||||||||||
| Last Modified: | 10 Jun 2026 08:28 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/53825 |
Actions (login required)
 |
View Item |