CLASSIFICATION OF MONKEY CHARACTERS IN BALINESE RAMAYANA SHADOW PUPPET USING CONVOLUTIONAL NEURAL NETWORK METHOD WITH RESNET-50 AND VGG-16 ARCHITECTURES

Cahyana, Wayan Restu (2026) CLASSIFICATION OF MONKEY CHARACTERS IN BALINESE RAMAYANA SHADOW PUPPET USING CONVOLUTIONAL NEURAL NETWORK METHOD WITH RESNET-50 AND VGG-16 ARCHITECTURES. Undergraduate thesis, Universitas Pendidikan Ganesha.

[img] Text (ABSTRACT)
2215101035-ABSTRAK.pdf
Download (9MB)
[img] Text (BAB 1 PENDAHULUAN)
2215101035-BAB 1 PENDAHULUAN.pdf
Download (9MB)
[img] Text (BAB 2 KAJIAN TEORI)
2215101035-BAB 2 KAJIAN TEORI.pdf
Restricted to Repository staff only
Download (9MB) | Request a copy
[img] Text (BAB 3 METODELOGI PENELITIAN)
2215101035-BAB 3 METODELOGI PENELITIAN.pdf
Restricted to Repository staff only
Download (9MB) | Request a copy
[img] Text (BAB 4 HASIL DAN PEMBAHASAN)
2215101035-BAB 4 HASIL DAN PEMBAHASAN.pdf
Restricted to Repository staff only
Download (9MB) | Request a copy
[img] Text (BAB 5 PENUTUP)
2215101035-BAB 5 PENUTUP.pdf
Restricted to Repository staff only
Download (9MB) | Request a copy
[img] Text (DAFTAR PUSTAKA)
2215101035-DAFTAR PUSTAKA.pdf
Download (9MB)
[img] Text (LAMPIRAN)
2215101035-LAMPIRAN.pdf
Download (395kB)
[img] Text (COVER)
2215101035-COVER.pdf
Download (10MB)

Abstract

Shadow puppet is an original Indonesian cultural heritage rich in artistic value and moral messages. However, in Balinese Ramayana shadow puppet shows, the general public often finds it difficult to recognize individual monkey characters due to their highly similar shapes and appearances. To solve this problem, this study designs an artificial intelligence-based image classification system, specifically utilizing the Convolutional Neural Network (CNN) method with transfer learning applied to the VGG-16 and ResNet-50 architectures. The dataset for this research uses 270 original images from Nagasepaha Village, categorized into 15 distinct classes consisting of 14 types of monkey characters and 1 non-monkey puppet class. To find the optimal performance, testing was conducted through various combinations of training data splits and hyperparameter settings. Additionally, image augmentation techniques were applied to expand the variation of the training data to prevent the models from overfitting. The experimental results show that both models are highly capable of recognizing the visual characteristics of the shadow puppets. The optimal performance was achieved by the VGG-16 architecture with a testing accuracy of 98.67%, followed closely by ResNet-50 with an accuracy of 98.33%. These results prove that both deep learning architectures are excellent at distinguishing detailed features of visually similar cultural objects, with VGG-16 showing a marginal advantage under limited dataset conditions. For future development, it is recommended to increase the volume and variation of images for each character. Future studies could also explore more modern deep learning architectures, perform advanced hyperparameter optimization, and develop this system to detect puppet characters in real-time video formats during live performances.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: convolutional neural network, ResNet-50, VGG-16, Balinese Ramayana shadow puppet, monkey characters
Subjects: T Technology > T Technology (General)
Divisions: Fakultas Teknik dan Kejuruan > Jurusan Teknik Informatika > Program Studi Ilmu Komputer (S1)
Depositing User: WAYAN RESTU CAHYANA
Date Deposited: 13 Jun 2026 06:01
Last Modified: 13 Jun 2026 06:01
URI: http://repo.undiksha.ac.id/id/eprint/29745

Actions (login required)

View Item View Item