Synthetic OCR benchmark for Arabic, Cyrillic, and Latin script Kazakh OCR. Labels are in metadata.csv.
Curated by: Henry Gagnier, Sophie Gagnier, Ashwin Kirubakaran
Language(s) (NLP): Kazakh (kk)
License: MIT
@inproceedings{gagnier-etal-2026-kazakhocr,
title = "{K}azakh{OCR}: A Synthetic Benchmark for Evaluating Multimodal Models in Low-Resource {K}azakh… See the full description on the dataset page:
https://huggingface.co/datasets/henrygagnier/kazakh-ocr.
