README.md

metadata

language:
  - hi
  - bn
  - kn
license: apache-2.0
task_categories:
  - image-to-text
tags:
  - ocr
  - handwritten
  - indic
  - paddleocr
size_categories:
  - 1M<n<10M

Indic OCR Dataset (Hindi, Bengali, Kannada)

Comprehensive dataset containing handwritten and printed word images for Indic OCR training and evaluation.

Dataset Structure

• engali/: Bengali handwritten word images and annotations.
• hindi/: Hindi handwritten word images and annotations.
• kannada/: Kannada handwritten word images and annotations.
• data/: Printed split datasets (Train, Val, Test) for Hindi, Bengali, Kannada.

Usage with PaddleOCR

Prepared for fine-tuning PaddleOCR models on DGX systems.