PENERAPAN MODEL HYBRID CNN-BILSTM UNTUK KONVERSI TULISAN TANGAN KE DOKUMEN DIGITAL
DOI:
https://doi.org/10.23969/jp.v11i03.66135Keywords:
Handwritten Text Recognition, Hybrid CNN-BiLSTM, Document Digitization, Deep Learning, CTC LossAbstract
Digital transformation through Handwritten Text Recognition (HTR) technology based on deep learning can optimize the efficiency of processing public administration documents. In regional government agencies, processing handwritten files manually causes severe bottlenecks and human error. This study aims to analyze the effect of hyperparameter configurations and evaluate the quantitative performance of a hybrid CNN-BiLSTM architecture with Connectionist Temporal Classification (CTC) Loss using real handwritten administrative document datasets from the Department of Cooperatives and SMEs in Makassar City. The CNN module acts as a spatial visual feature extractor, while BiLSTM processes sequence transitions bidirectionally across 128 time-steps. The dataset consists of 400 line images divided into an 80:20 ratio for training and testing. The experimental results demonstrate that the hybrid CNN-BiLSTM model achieved a character accuracy of 27.03% (CER 72.97%, WER 98.75%), Precision of 0.1919, Recall of 0.1781, and F1-Score of 0.1747, outperforming single baseline models (CNN 11.45% and BiLSTM 8.20%). The model achieved 100% exact match transliteration on clean background documents. The main factors limiting recognition include cursive handwriting, sharp slant angles, and local non-standard acronyms. In conclusion, the hybrid CNN-BiLSTM architecture effectively integrates spatial feature extraction and sequential contextual modeling, providing a reliable backend foundation for automating regional public document digitization.
Downloads
References
Alom, M. Z., Taha, T. M., Yakopcic, C., Westberg, S., Sidike, P., Nasrin, M. S., ... & Asari, V. K. (2023). A state-of-the-art survey on deep learning theory and architectures. Electronics, 8(3), 292-310.
El Khayati, R., Kich, I., & Taouil, A. (2024). CNN-based feature extraction and sequence modeling for handwritten document recognition. Journal of Information Systems and Emerging Technologies, 12(1), 45-58.
Goodfellow, I., Bengio, Y., & Courville, A. (2023). Deep learning. Cambridge: MIT Press.
Graves, A., Fernández, S., Gomez, F., & Schmidhuber, J. (2023). Connectionist temporal classification: labelling unsegmented sequence data with recurrent neural networks. Proceedings of the 23rd International Conference on Machine Learning, 369-376.
Hassaballah, M., Ribeiro, A. L., & Abdelmgeid, A. A. (2024). Deep learning in computer vision: Principles and applications. CRC Press.
Hidayat, R., Ramadhani, S., & Pratama, A. (2023). Analisis hambatan digitalisasi dokumen publik pada instansi pemerintahan daerah. Jurnal Administrasi Publik dan Sektor Komputasi, 5(2), 112-125.
Hochreiter, S. (2024). Long short-term memory networks and deep sequential learning. Neural Computation, 9(8), 1735-1780.
Kang, L., Lee, J., & Park, H. (2021). Hybrid CNN-BiLSTM architecture for handwritten text recognition on standard benchmarks. IEEE Access, 9, 45210-45221.
Kurniawan, A., & Ramadhani, F. (2024). Variabilitas morfologi tulisan tangan dan dampaknya pada performa model HTR. Jurnal Teknologi Informasi dan Ilmu Komputer, 11(1), 89-98.
Mahadevkar, S. V., Khemani, B., Patil, S., Kotecha, K., Vora, D. R., & Abraham, A. (2024). A review on machine learning styles in Handwritten Text Recognition. Archives of Computational Methods in Engineering, 31(2), 815-845.
Nugroho, B., & Pratama, A. (2023). Digitalisasi birokrasi pemerintahan menuju tata kelola berbasis elektronik. Jurnal Tata Kelola Publik, 8(1), 34-47.
Oktarino, D., Defit, S., & Yuhandri, Y. (2025). Analisis CNN-BiLSTM model hybrid untuk pengenalan tulisan tangan bahasa Indonesia. Jurnal Informasi dan Teknologi (JIDT), 7(1), 102-110.
Pratama, I., & Setiawan, B. (2023). Keterbatasan Optical Character Recognition (OCR) konvensional pada variasi teks non-cetak. Jurnal Sistem Informasi dan Ilmu Komputer, 4(3), 201-210.
Purohit, S., & Panwar, A. (2025). Hybrid CNN-BiLSTM for multilingual handwritten text recognition. Pattern Recognition Letters, 168, 115-124.
Ribeiro, M., Silva, J., & Santos, L. (2023). End-to-end optical character recognition using deep hybrid neural networks. Pattern Recognition, 134, 109-122.
Sharma, V., & Nagpal, S. (2024). Challenges in handwritten text recognition for non-standard local scripts and acronyms. International Journal of Document Analysis and Recognition, 27(2), 145-159.
Sokolova, M., & Lapalme, G. (2023). A systematic analysis of performance measures in classification tasks. Information Processing & Management, 45(4), 427-437.
Suryani, A., & Wijaya, C. (2023). Evaluasi model deep learning pada data riil dokumen birokrasi lokal. Jurnal Rekayasa Sistem dan Teknologi Informasi, 7(4), 512-520.
Tan, H., Zhang, Y., & Wang, L. (2025). Spatial feature extraction and temporal alignment in deep sequence modeling. IEEE Transactions on Pattern Analysis and Machine Intelligence, 47(3), 1890-1902.
Zhou, X., Chen, Y., & Huang, Z. (2023). Attention-based hybrid CNN-BiLSTM model for HTR. Applied Soft Computing, 132,109850.
Downloads
Published
Issue
Section
License
Copyright (c) 2026 Pendas : Jurnal Ilmiah Pendidikan Dasar
This work is licensed under a Creative Commons Attribution 4.0 International License.
