Skip to content

Dataset · gksriharsha

Large Telugu OCR dataset

gksriharsha/chitralekha

Tens of millions of text image and transcript pairs for training Telugu optical character recognition systems.

What's inside

Contains 10 to 100 million image-text pairs for image-to-text (OCR) tasks in Telugu (te). The author manually cleaned the set to remove broken font rendering and incorrect ligatures.

What you can do with it

  • Training and fine-tuning OCR models for Telugu script recognition
  • Evaluating multimodal model quality when reading text in low-resource languages
Tasks
Image → text
Languages
Telugu
Size
10M–100M rows
Downloads this month
21.5K
Likes
11

How to load it

from datasets import load_dataset

ds = load_dataset("gksriharsha/chitralekha")