Abstract
This study explores offline handwritten word recognition for Central Kurdish (Sorani) script using a Convolutional Recurrent Neural Network (CRNN) framework with Connectionist Temporal Classification (CTC). The research presents an empirical multi-backbone evaluation across various CNN feature extraction architectures, analyzing recognition accuracy, character error rates, and sequence modeling capabilities for low-resource Kurdish handwriting.
Citation
Alsaqi, I. M., & Fattah, P. (2025). Kurdish handwritten word recognition with CRNN using multi-backbone evaluation. Academic Journal of International University of Erbil, 2(4), 439–449. https://doi.org/10.63841/iue24539
Extra info
- Type: Journal Article
- Link: Academic Journal of International University of Erbil