datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
RASAM-2
RASAM 2 — Line-Level HTR Ground-Truth for Arabic historical manuscripts (Maghrebi)
Dataset Description
RASAM 2 extends the initial initiative RASAM 1 by integrating the outcomes of the second manuscript transcription hackathon organized by the Research Consortium Middle-East and Muslim Worlds (GIS MOMM) and Calfa at BULAC (Paris) between December 2021 and March 2022.
This HuggingFace dataset provides cropped line-level images paired with their… See the full description on the dataset page: https://huggingface.co/datasets/calfa-ai/RASAM-2.RASAM-1
RASAM 1 — Line-Level HTR Ground-Truth for Arabic historical manuscripts (Maghrebi)
Dataset Description
RASAM 1 is a specialized dataset for Handwritten Text Recognition (HTR) focusing on Arabic historical manuscripts in Maghrebi script.
This HuggingFace dataset provides cropped line-level images paired with their transcriptions and rich metadata for 3 Arabic Maghrebi manuscripts from the BULAC Library. It is designed as a ready-to-use resource for… See the full description on the dataset page: https://huggingface.co/datasets/calfa-ai/RASAM-1.RASAM
RASAM dataset
An Open Dataset for the Recognition and Analysis of Scripts in Arabic Maghrebi
How to cite
The paper has been presented during the ICDAR 2021 conference (ASAR workshop). To cite this work and this dataset, please use the following informations:
@InProceedings{2021rasam-dataset,
author="Vidal-Gorène, Chahan and Lucas, Noëmie and Salah, Clément and Decours-Perez, Aliénor and Dupin, Boris",
editor="Barney Smith, Elisa H. and Pal, Umapada",
title="RASAM -- A… See the full description on the dataset page: https://huggingface.co/datasets/johnlockejrr/RASAM.rasam_datasetautotrain-data-rasa