riotu-lab/SARD

SARD: Synthetic Arabic Recognition Dataset Overview SARD (Synthetic Arabic Recognition Dataset) is a large-scale, synthetically generated dataset designed for training and evaluating Optical Character Recognition (OCR) models for Arabic text. This dataset addresses the critical need for comprehensive Arabic text recognition resources by providing controlled, diverse, and scalable training data that simulates real-world book layouts. Key Features… See the full description on the dataset page: https://huggingface.co/datasets/riotu-lab/SARD.