CHIPS - Corpus of Handwritten Indic Scripts (Page-Level OCR)

MH Subset Needed

Page-level handwritten OCR dataset for Indic scripts with both text detection and recognition annotations. Part of the PLATTER (Page-Level Handwritten Text Recognition) project. Unlike word-level datasets, CHIPS provides full-page handwritten document images with bounding box annotations for text regions plus Unicode transcriptions, enabling training end-to-end page-level OCR systems that handle detection and recognition jointly. Covers multiple Indic scripts including Devanagari.

Build a page-level Marathi handwritten OCR system that processes entire document pages without manual line segmentation.

Maharashtra subset not yet extracted. This is a global dataset that contains data covering Maharashtra. A regional subset can be extracted by filtering on geographic coordinates or administrative boundaries.

Homepage Paper

Quick Start

# PLATTER project - page-level handwritten OCR
# Paper: https://arxiv.org/abs/2502.06172
# Contact authors for dataset access

print("CHIPS: Page-level handwritten Indic OCR dataset")
print("Supports detection + recognition jointly")
print("Filter Devanagari script pages for Marathi OCR")

Modality

Image (full-page handwritten documents with detection + recognition annotations)

Size

1,458+ page-level document images; 11 Indic scripts; 463 writers

License

Research use

Format

PNG/JPEG with COCO-format annotations

Language

mr, hi, bn, ta, te, kn, ml, gu, pa, or, as

Update Frequency

static

Organization

Multi-institutional (PLATTER project)

Schema

Field	Type	Description
image	image	Full-page handwritten document scan
text_regions	json	Bounding box coordinates for text line regions
transcriptions	array	Unicode transcriptions for each detected text region
script	string	Script identifier
writer_id	string	Writer identifier

Build With This

Create an automatic Marathi handwritten exam paper grader that detects and reads handwritten answers from full pages

Develop a historical Marathi letter digitizer that handles full pages of cursive Devanagari handwriting

Build a handwritten form processor for Maharashtra government offices that extracts field values from handwritten forms

AI Use Cases

Page-level handwritten Marathi text detection and recognitionEnd-to-end handwritten document OCRText line detection in handwritten Indic documentsHandwriting segmentation model training

Related Datasets

AIKOSH IIT Bombay Indic Datasets (IndiaAI)

multimodal

Bharat Scene Text Dataset (BSTD)

Image (scene text)

CMATERdb - Devanagari-Roman Mixed-Script Handwritten Documents

Image (handwritten mixed-script document pages with word-level annotations)

COCO Captions Marathi

Text (caption pairs)

Last verified: 2026-03-12