1.32 GB
7 files
Updated about 1 month ago
Name
Size
data
.gitattributes2.42 kB
xet
README.md2.35 kB
xet
README.md

Muharaf-public line images

This dataset contains 24,495 line images of Arabic handwriting and the corresponding text. The images are Manuscripts from the 19th to 21st century. See the official code, paper, and zenodo archive below. Our work has been accepted to NeurIPS 2024.

Code Paper Dataset

How to use

from datasets import load_dataset

ds = load_dataset("aamijar/muharaf-public")

# remember to index into the row first to avoid decoding all images at once (https://huggingface.co/docs/datasets/en/image_load#load-image-data)
print(ds["train"][0]["image"])

# instead of
print(ds["train"]["image"][0])

Citation

If you are using this dataset in your work, please cite our paper below.

@inproceedings{
  saeed2024muharaf,
  title={Muharaf: Manuscripts of Handwritten Arabic Dataset for Cursive Text Recognition},
  author={Mehreen Saeed and Adrian Chan and Anupam Mijar and Joseph Moukarzel and Gerges Habchi and Carlos Younes and Amin Elias and Chau-Wai Wong and Akram Khater},
  booktitle={The Thirty-eight Conference on Neural Information Processing Systems Datasets and Benchmarks Track},
  year={2024},
  url={https://openreview.net/forum?id=1s8l1tnTXW}
}
Total size
1.32 GB
Files
7
Last updated
Oct 22
Pre-warmed CDN
US EU US EU

Contributors