Skip to content

Dataset · AlbertoChestnut

Telugu book scans and OCR text pairs

AlbertoChestnut/telugu-ocr

About 25.5 thousand pairs of scanned book pages and human-verified Telugu text from Wikisource.

What's inside

The corpus is compiled from 221 Telugu books from the Telugu Wikisource project, with a size of about 11 GB. It contains roughly 25,565 file pairs: JPG page images and corresponding human-annotated UTF-8 text transcriptions. Distributed under the CC BY-SA 4.0 license.

What you can do with it

  • Training and fine-tuning OCR models for Telugu script recognition
  • Evaluating the quality of existing book digitization systems for low-resource languages
Tasks
Image → text
Languages
Telugu
Size
100K–1M rows
Downloads this month
30.5K
Likes
1

How to load it

from datasets import load_dataset

ds = load_dataset("AlbertoChestnut/telugu-ocr")