Indonesian Local Script Handwritten Recognition
A dual-pipeline AI system for automated handwritten character recognition of nine Indonesian indigenous writing systems, combining deep learning research with a publicly accessible machine learning demonstration. Notes! Because the paper hasn't been published yet, to protect my privacy, I've made the GitHub repository private for this project.
Indonesia's nine local scripts are at risk of being forgotten with no comprehensive digital recognition system to preserve them.
Designed two parallel pipelines a Deep Learning + ML pipeline (MobileNetV2/EfficientNet-B0 + XGBoost/LightGBM/CatBoost ensemble) for academic research, and a CV + ML pipeline for public demonstration deployed on Hugging Face across 228 character classes and 55,713 images. Research paper currently being processed for IEEE conference publication.