Evaluasi Kinerja U-Net ResNet34 dan MDSBN: Studi Komparatif untuk Segmentasi Naskah Kuno Indonesia
DOI:
https://doi.org/10.47065/bulletincsr.v6i5.1236Keywords:
Ancient Document Segmentation; Deep Learning; MDSBN; U-Net ResNet34; Indonesian Manuscript; Image BinarizationAbstract
The digitization of ancient documents is an important step in preserving historical and cultural information. However, the resulting images often suffer from degradation, such as stains, uneven background textures, faded ink, and low contrast, making text-background separation difficult. This study compares two deep learning architectures, namely U-Net ResNet34 and the Modified Deep Semantic Binarization Network (MDSBN), for the segmentation of Indonesian ancient documents. The dataset consists of Balinese palm-leaf manuscripts, Sundanese manuscripts, and additional ancient document images obtained from Wikimedia Commons. The experiments were conducted through a learning rate search and batch size sensitivity analysis, and the models were evaluated using the Dice Coefficient, Intersection over Union (IoU), Precision, Recall, and Root Mean Squared Error (RMSE). This study contributes through a controlled evaluation of both architectures using a consistent dataset, preprocessing pipeline, loss function, evaluation metrics, and computational environment, enabling performance differences to be analyzed more objectively. The results show that U-Net ResNet34 achieved its best performance using a learning rate of 5e-5 and a batch size of 16, with a test Dice score of 0.79338 and a test IoU score of 0.65752. It outperformed MDSBN, which achieved its best performance using a learning rate of 1e-6 and a batch size of 32, with a test Dice score of 0.75338 and a test IoU score of 0.60433. The functional advantage of U-Net ResNet34 is associated with the ability of its residual encoder to extract hierarchical features from complex textures and degradation patterns, while its skip connections help preserve the spatial details of thin text strokes. These characteristics make U-Net ResNet34 more adaptive to variations in degradation within the Indonesian ancient document dataset than the more compact MDSBN architecture.
Downloads
References
Z. Ziran, M. Mecella, and S. Marinai, “AI-Driven Enhancement Of Historical Documents,” Int. J. Digit. Libr., vol. 26, p. 22, 2025, doi: 10.1007/s00799-025-00430-y.
S. Münster et al., “Artificial Intelligence For Digital Heritage Innovation: Setting Up An R&D Agenda For Europe,” Heritage, vol. 7, no. 2, pp. 794–816, 2024, doi: 10.3390/heritage7020038.
Y. Zhou, S. Zuo, Z. Yang, J. He, J. Shi, and R. Zhang, “A Review Of Document Image Enhancement Based On Document Degradation Problem,” Appl. Sci., vol. 13, no. 13, p. 7855, 2023, doi: 10.3390/app13137855.
S. Khamekhem Jemni, M. A. Souibgui, Y. Kessentini, and A. Fornés, “Enhance To Read Better: A Multi-Task Adversarial Network For Handwritten Document Image Enhancement,” Pattern Recognit., vol. 123, p. 108370, 2022, doi: 10.1016/j.patcog.2021.108370.
W. Xiong, L. Zhou, L. Yue, L. Li, and S. Wang, “An Enhanced Binarization Framework for Degraded Historical Document Images,” EURASIP Journal on Image and Video Processing, vol. 2021, no. 1, art. no. 13, 2021, doi: 10.1186/s13640-021-00556-4.
Z. Yang, S. Zuo, Y. Zhou, J. He, and J. Shi, “A Review Of Document Binarization: Main Techniques, New Challenges, And Trends,” Electronics, vol. 13, no. 7, p. 1394, 2024, doi: 10.3390/electronics13071394.
B. Bataineh et al., “A Comprehensive Review On Document Image Binarization,” J. Imaging, vol. 11, no. 5, p. 133, 2025, doi: 10.3390/jimaging11050133.
R. D. Lins, R. Bernardino, R. da Silva Barboza, and R. C. De Oliveira, “Using Paper Texture For Choosing A Suitable Algorithm For Scanned Document Image Binarization,” J. Imaging, vol. 8, no. 10, p. 272, 2022, doi: 10.3390/jimaging8100272.
Q.-V. Dang and G.-S. Lee, “Document Image Binarization With Stroke Boundary Feature Guided Network,” IEEE Access, vol. 9, pp. 36924–36936, 2021, doi: 10.1109/ACCESS.2021.3062904.
D. Li, Y. Wu, and Y. Zhou, “SauvolaNet: Learning Adaptive Sauvola Network for Degraded Document Binarization,” Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), vol. 12824 LNCS, no. 189, pp. 538–553, 2021, doi: 10.1007/978-3-030-86337-1_36.
C. Tensmeyer and T. Martinez, “Document Image Binarization with Fully Convolutional Neural Networks,” Proc. Int. Conf. Doc. Anal. Recognition, ICDAR, vol. 1, pp. 99–104, 2017, doi: 10.1109/ICDAR.2017.25.
S. Suh, J. Kim, P. Lukowicz, and Y. O. Lee, “Two-Stage Generative Adversarial Networks For Binarization Of Color Document Images,” Pattern Recognit., vol. 130, p. 108810, 2022, doi: 10.1016/j.patcog.2022.108810.
Y. Guo, C. Ji, X. Zheng, Q. Wang, and X. Luo, “Multi-Scale Multi-Attention Network For Moiré Document Image Binarization,” Signal Process. Image Commun., vol. 90, p. 116046, 2021, doi: 10.1016/j.image.2020.116046.
L. Zhang, K. Wang, and Y. Wan, “An Efficient Transformer--CNN Network For Document Image Binarization,” Electronics, vol. 13, no. 12, p. 2243, 2024, doi: 10.3390/electronics13122243.
Z. Du and C. He, “Nonlinear Diffusion System For Simultaneous Restoration And Binarization Of Degraded Document Images,” Comput. & Math. with Appl., vol. 153, pp. 237–248, 2024, doi: 10.1016/j.camwa.2023.11.033.
O. Ronneberger, P. Fischer, and T. Brox, “U-Net: Convolutional Networks for Biomedical Image Segmentation,” in Medical Image Computing and Computer-Assisted Intervention—MICCAI 2015, Lecture Notes in Computer Science, vol. 9351, pp. 234–241, 2015, doi: 10.1007/978-3-319-24574-4_28..
N. Siddique, S. Paheding, C. P. Elkin, and V. Devabhaktuni, “U-Net and Its Variants for Medical Image Segmentation: A Review of Theory and Applications,” IEEE Access, vol. 9, pp. 82031–82057, 2021, doi: 10.1109/ACCESS.2021.3086020..
N. Detsikas, N. Mitianoudis, and N. Papamarkos, “A Dilated MultiRes Visual Attention U-Net For Historical Document Image Binarization,” Signal Process. Image Commun., vol. 122, p. 117102, 2024, doi: 10.1016/j.image.2024.117102.
S. Kang, B. K. Iwana, and S. Uchida, “Complex Image Processing With Less Data—Document Image Binarization By Integrating Multiple Pre-Trained U-Net Modules,” Pattern Recognit., vol. 109, no. 109, p. 107577, 2021, doi: 10.1016/j.patcog.2020.107577.
I. Yuadi et al., “A Benchmark Study of Classical and U-Net ResNet34 Methods for Binarization of Balinese Palm Leaf Manuscripts,” Heritage, vol. 8, no. 8, art. no. 337, 2025, doi: 10.3390/heritage8080337.
B. Nair B. J. and N. Shobha Rani, “A Modified Deep Semantic Binarization Network for Degradation Removal in Palm Leaf Manuscripts,” Multimedia Tools and Applications, vol. 83, no. 23, pp. 62937–62969, 2024, doi: 10.1007/s11042-023-18020-y.
F. J. Castellanos, A.-J. Gallego, and J. Calvo-Zaragoza, “Unsupervised Neural Domain Adaptation For Document Image Binarization,” Pattern Recognit., vol. 119, p. 108099, 2021, doi: 10.1016/j.patcog.2021.108099.
R. Sukesh, M. Seuret, A. Nicolaou, M. Mayr, and V. Christlein, “A Fair Evaluation of Various Deep Learning-Based Document Image Binarization Approaches,” Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics), vol. 13237 LNCS, pp. 771–785, 2022, doi: 10.1007/978-3-031-06555-2_52.
M. W. A. Kesiman, J.-C. Burie, J.-M. Ogier, G. N. M. A. Wibawantara, and I. M. G. Sunarya, “AMADI_LontarSet: The First Handwritten Balinese Palm Leaf Manuscripts Dataset,” in 2016 15th International Conference on Frontiers in Handwriting Recognition (ICFHR), pp. 168–172, 2016, doi: 10.1109/ICFHR.2016.39..
M. W. A. Kesiman, D. Valy, J.-C. Burie, E. Paulus, M. Suryani, S. Hadi, M. Verleysen, S. Chhun, and J.-M. Ogier, “ICFHR 2018 Competition on Document Image Analysis Tasks for Southeast Asian Palm Leaf Manuscripts,” in 2018 16th International Conference on Frontiers in Handwriting Recognition (ICFHR), pp. 483–488, 2018, doi: 10.1109/ICFHR-2018.2018.00090.
M. Suryani, E. Paulus, S. Hadi, U. A. Darsa, and J.-C. Burie, “The Handwritten Sundanese Palm Leaf Manuscript Dataset from 15th Century,” in 2017 14th IAPR International Conference on Document Analysis and Recognition (ICDAR), vol. 1, pp. 796–800, 2017, doi: 10.1109/ICDAR.2017.135.
K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” Proc. IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit., vol. 2016-Decem, pp. 770–778, 2016, doi: 10.1109/CVPR.2016.90.
B. Bischl et al., “Hyperparameter Optimization: Foundations, Algorithms, Best Practices, And Open Challenges,” Wiley Interdiscip. Rev. Data Min. Knowl. Discov., vol. 13, no. 2, p. e1484, 2023, doi: 10.1002/widm.1484.
Bila bermanfaat silahkan share artikel ini
Berikan Komentar Anda terhadap artikel Evaluasi Kinerja U-Net ResNet34 dan MDSBN: Studi Komparatif untuk Segmentasi Naskah Kuno Indonesia
ARTICLE HISTORY
How to Cite
Issue
Section
Copyright (c) 2026 Rino Zakharia, Budi Nugroho, Eka Prakarsa Mandyartha
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).
