sfidan42/turkish-ocr-noise-corpus-v2

Turkish OCR Noise Corpus, version 2 This corpus contains rendered Turkish document pages under one clean condition and five synthetic degradations, together with their PP-OCRv5 transcriptions. It supports the study of how OCR errors affect dense retrieval of Turkish question-answer text (Chapter 6 of the thesis cited below). Version 2 keeps the pages, the clean renders and the blur and low-resolution images of version 1. It makes rotation, noise and JPEG compression harsher… See the full description on the dataset page: https://huggingface.co/datasets/sfidan42/turkish-ocr-noise-corpus-v2.

<iframe src="https://silytics.si/embed/item/a311908b-e902-4e20-9d20-cb5aae94a9e6" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 18 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.