Aziz, Muhammad Fauzi Taj (2025) Prediksi Motif-Motif Batik Khas Solo dan Jogja Menggunakan Metode MixConv. Undergraduate thesis, UPN Veteran Jawa Timur.
![[img]](https://repository.upnjatim.ac.id/style/images/fileicons/text.png) |
Text (Cover)
21082010234-cover.pdf Download (787kB) |
|
Text (Bab 1)
21082010234-bab1.pdf Download (270kB) |
|
Text (Bab 2)
21082010234-bab2.pdf Restricted to Repository staff only until 12 September 2027. Download (555kB) |
|
Text (Bab 3)
21082010234-bab3.pdf Restricted to Repository staff only until 12 September 2027. Download (655kB) |
|
Text (Bab 4)
21082010234-bab4.pdf Restricted to Repository staff only until 12 September 2027. Download (5MB) |
|
Text (Bab 5)
21082010234-bab5.pdf Download (213kB) |
|
Text (Daftar Pustaka)
21082010234-daftarpustaka.pdf Download (153kB) |
|
Text (Lampiran)
21082010234-lampiran.pdf Restricted to Repository staff only Download (465kB) |
Abstract
This study discusses the application of the MixConv method (Mixed Depthwise Convolutional Kernels) as the core algorithm of MixNet for classifying batik pattern images. MixConv was chosen due to its superior ability to handle multi-scale feature maps compared to standard convolution methods. The findings indicate that using MixConv in deep learning models holds promising potential. The study classifies eight classes using 531 images collected from the internet. To address the issue of imbalanced datasets, Stratified K-Fold Cross-Validation is employed. In this research, 30 model training scenarios using MixNet are conducted on Google Colaboratory to determine the best-performing MixNet variant for the dataset. The results show that the pretrained MixNet-S model on Fold Dua achieved a validation accuracy of 0.9234 and a validation loss of 0.4322. As an additional contribution, the system has been successfully deployed via a cloud-based API using Google Cloud Service, enabling users to access the model directly through a web interface deployed with Vercel, without the need to install any applications. It is hoped that this approach can be utilized by other researchers to enhance public accessibility to AI-based classification systems and to further expand the application of MixConv in related fields.
| Item Type: | Thesis (Undergraduate) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Contributors: |
|
||||||||||||
| Subjects: | N Fine Arts > N Visual arts Q Science > QA Mathematics > QA75 Electronic computers. Computer science Q Science > QA Mathematics > QA76 Computer software Q Science > QA Mathematics > QA76.6 Computer Programming Q Science > QA Mathematics > QA76.625 Internet Programming |
||||||||||||
| Divisions: | Faculty of Computer Science > Departemen of Information Systems | ||||||||||||
| Depositing User: | Muhammad Fauzi Taj Aziz | ||||||||||||
| Date Deposited: | 16 Sep 2025 02:17 | ||||||||||||
| Last Modified: | 16 Sep 2025 02:17 | ||||||||||||
| URI: | https://repository.upnjatim.ac.id/id/eprint/43119 |
Actions (login required)
 |
View Item |