<![CDATA[Batak script is a cultural heritage of the Batak people in North Sumatra that requires preservation through digitization. This study proposes the FastViT-SA12 architecture for handwritten Batak script recognition using the Batak Char 20 dataset, which consists of 117 character classes. To address data imbalance, data augmentation was performed using ImgAug, resulting in 650 images per class. The dataset was divided into training, validation, and testing sets with a ratio of 70:20:10, and all images were resized to 64 × 64 pixels to reduce computational overhead. The model was trained for 30 epochs using the Adam optimizer with a learning rate of 0.001 and a batch size of 64. The proposed model achieved a training accuracy of 99.97%, a validation accuracy of 97.96%, and a test accuracy of 89.31%, with weighted precision, recall, and F1-score of 91.84%, 89.31%, and 88.99%, respectively. FastViT-SA12 demonstrates promising potential for recognizing 117 Batak script classes while maintaining computational efficiency with a compact parameter footprint, making it a viable candidate for mobile-first Optical Character Recognition (OCR) systems in manuscript preservation.]]>
Copyrights © 2026
