Marathi Handwritten Text Dataset

Marathi Handwritten Text Dataset

MH Specific

2,500+ images of full handwritten Marathi text (sentences and paragraphs, not isolated characters). Native speakers wrote pre-designed text covering nearly all Marathi characters, words, and diacritical marks. Fills the gap between character-level datasets (like Devanagari HWR) and real-world handwritten text recognition (HTR).

Build an end-to-end handwritten Marathi text recognition system for digitizing handwritten documents and forms.
HomepageDownload

Quick Start

# Marathi Handwritten Text Dataset
import cv2
print('Marathi Handwritten Text Dataset')
print('For handwritten text line recognition in Devanagari')
Modality
image
Size
2,500+ images; ~14 MB; full handwritten text passages
License
Format
Image files (PNG/JPG)
Language
mr
Update Frequency
static
Organization
Independent researcher (Kaggle)
Category
vision-ocr-multimodal

Schema

FieldTypeDescription
imageimageHandwritten Marathi text line image
textstringGround truth transcription

Build With This

Create a handwritten Marathi prescription reader for pharmacies to reduce medication errors
Develop a historical Marathi manuscript OCR system for digitizing Modi script and old Devanagari documents
Build a handwriting-to-digital converter for Maharashtra government offices processing handwritten applications

AI Use Cases

Marathi handwritten text recognition (HTR)Document digitization for Marathi manuscriptsHandwriting analysis and writer identificationHistorical document OCR pipeline development
Last verified: 2026-03-09