Deep Learning Approaches for Batak Script Recognition: A Literature Review

Abstract

The Batak script, one of Indonesia’s ancient writing systems, has recently garnered increasing attention in the fields of digital preservation and pattern recognition. However, the complexity of its character structures and the limited availability of annotated datasets pose significant challenges to automated recognition. This paper presents a comprehensive literature review on deep learning approaches for Batak script recognition. We analyze existing studies that apply convolutional neural networks (CNNs), recurrent neural networks (RNNs), hybrid models, and transfer learning to character recognition tasks. Furthermore, we highlight the strengths and limitations of these methods in addressing challenges such as character similarity, dataset scarcity, and noise in historical manuscripts. The review also discusses research gaps and potential future directions, including the integration of attention mechanisms, data augmentation strategies, and multimodal approaches. By synthesizing recent developments, this study provides valuable insights for researchers aiming to advance Batak script recognition and contributes to the broader effort of preserving Indonesia’s cultural heritage through deep learning technologies.