Implementasi Transfer Learning Pada Algoritma Convolutional Neural Network untuk Mengklasifikasikan Image Objek Wisata
-
Mira Mira
*
Universitas Kristen Satya Wacana, Salatiga,
Indonesia
- Irwan Sembiring Universitas Kristen Satya Wacana, Salatiga, Indonesia
- Hindriyanto Dwi Purnomo Universitas Kristen Satya Wacana, Salatiga, Indonesia (*) Corresponding Author
Abstract
This study classifies the image of a tourist attraction with 9 labels sky, tree, mountain, water, street, temple, garden, stone and ricefield. The results of multi-label labeling can be used to see the frequency and recommendations of tourist attractions in Central Java, and build a transfer learning model to determine the accuracy value. Classification with multi-label images has its own complexity in the labeling process and few people use it. Testing and evaluating the model uses the equation of accuracy and f-1 score. Several previous researchers also stated that the higher the amount of training data and the number of epochs per step, the higher the accuracy produced. Based on the results of training and evaluation of the four training processes, that 210 data using bs 8, lr 1e-3 and epoch 50 showed an accuracy of 0.8598 with a loss of 0.3245, while 290 data with bs 16, lr 1e-3 and epoch 50 showed an accuracy of 0.8685. with a loss of 0.2903. Then 594 data with bs 32, lr 1e-3 and epoch 50 showed an accuracy of 0.8852 with a loss of 0.2756, and 1000 data with bs 46, lr 1e-3 and epoch 50 showed an accuracy of 0.8833 with a loss of 0.2863. This can answer the statement that the greater the number of datas, the higher the accuracy produced, so that the transfer learning model on the ResNet-50 architecture with multi-label image datas can be applied by showing accuracy results close to the accuracy value on ResNet-50 in the imagenet project. In addition, the contribution of this research is to provide recommendations for potential tourist objects in Central Java, namely tourism objects with the theme of nature, then tourism processed by human hands such as historical places, cultural heritage and family recreation areas.
Downloads
References
D. Vaya and T. Hadpawat, “Classification in Data Mining : A Survey,” vol. 29, no. 3, pp. 13072–13078, 2020.
S. A. Shahriyar, K. M. Rokibul Alam, S. S. Roy, and Y. Morimoto, “An Approach for Multi Label Image Classification Using Single Label Convolutional Neural Network,” 2018 21st Int. Conf. Comput. Inf. Technol. ICCIT 2018, no. December, 2019, doi: 10.1109/ICCITECHN.2018.8631970.
T. Bariyah, M. A. Rasyidi, and N. Ngatini, “Convolutional Neural Network untuk Metode Klasifikasi Multi-Label pada Motif Batik,” Techno.Com, vol. 20, no. 1, pp. 155–165, 2021, doi: 10.33633/tc.v20i1.4224.
C. U. K. Enie Yuliani1, Afifah Nur Aini2, “Perbandingan Jumlah Epoch Dan Steps Per Epoch Pada Convolutional Neural Network Untuk Meningkatkan Akurasi Dalam Klasifikasi Gambar,” J. Inf. Politek. Indonusa Surakarta ISSN, vol. 5, no. 3, pp. 23–27, 2019.
F. Ratnawati, “Implementasi Algoritma Naive Bayes Terhadap Analisis Sentimen Opini Film Pada Twitter,” INOVTEK Polbeng - Seri Inform., vol. 3, no. 1, p. 50, 2018, doi: 10.35314/isi.v3i1.335.
S. N. Kane, A. Mishra, and A. K. Dutta, “Preface: International Conference on Recent Trends in Physics (ICRTP 2016),” J. Phys. Conf. Ser., vol. 755, no. 1, 2016, doi: 10.1088/1742-6596/755/1/011001.
J. Rozaqi, A. Sunyoto, and R. Arief, “Implementasi Transfer Learning pada Algoritma Convolutional Neural Network Untuk Identifikasi Penyakit Daun Kentang Implementation of Transfer Learning in the Convolutional Neural Network Algorithm for Identification of Potato Leaf Disease,” Procedia Eng. Life Sci., vol. 1, no. 1, 2021, [Online]. Available: https://press.umsida.ac.id/index.php/PELS/article/view/820/478.
M. A. H. Abas, N. Ismail, A. I. M. Yassin, and M. N. Taib, “VGG16 for plant image classification with transfer learning and data augmentation,” Int. J. Eng. Technol., vol. 7, no. 4, pp. 90–94, 2018, doi: 10.14419/ijet.v7i4.11.20781.
F. Nashrullah, S. Adhi, and G. Budiman, “Investigasi Parameter Epoch Pada Arsitektur ResNet- 50 Untuk Klasifikasi Pornografi,” J. Comput. Electron. Telecommun., vol. 1, no. 1, pp. 1–8, 2020.
B. Zhao, “Encyclopedia of Big Data,” Encycl. Big Data, no. December, 2020, doi: 10.1007/978-3-319-32001-4.
R. Wightman, H. Touvron, and H. Jégou, “ResNet strikes back: An improved training procedure in timm,” Cornel Univ., vol. 1, no. october, pp. 1–22, 2021, [Online]. Available: http://arxiv.org/abs/2110.00476.
P. M. Radiuk, “Impact of Training Set Batch Size on the Performance of Convolutional Neural Networks for Diverse Datasets,” Inf. Technol. Manag. Sci., vol. 20, no. 1, pp. 20–24, 2018, doi: 10.1515/itms-2017-0003.
T. Bezdan and N. Bačanin Džakula, “Convolutional Neural Network Layers and Architectures,” Int. Sci. Conf. Inf. Technol. DATA Relat. Res., no. January, pp. 445–451, 2019, doi: 10.15308/sinteza-2019-445-451.
V. Kumar and M. L., “Deep Learning as a Frontier of Machine Learning: A Review,” Int. J. Comput. Appl., vol. 182, no. 1, pp. 22–30, 2018, doi: 10.5120/ijca2018917433.
S. Jadon, “A survey of loss functions for semantic segmentation,” IEEE Access, vol. 2, no. 11, 2020.
G. Tika and Adiwijaya, “Klasifikasi Topik Berita Berbahasa Indonesia Menggunakan Multilayer Perceptron,” e-Proceeding Eng., vol. 6, no. 2, p. 2137, 2019.
Elgendy, M.(2019). Deep Learning for Vision System. 475.
Bila bermanfaat silahkan share artikel ini
Berikan Komentar Anda terhadap artikel Implementasi Transfer Learning Pada Algoritma Convolutional Neural Network untuk Mengklasifikasikan Image Objek Wisata
Pages: 209−216
Copyright (c) 2022 Mira, Irwan Sembiring, Hindriyanto Dwi Purnomo
This work is licensed under a Creative Commons Attribution 4.0 International License.
Authors who publish with this journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under Creative Commons Attribution 4.0 International License that allows others to share the work with an acknowledgment of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgment of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (Refer to The Effect of Open Access).
