ICDAR MLT-2019

ICDAR MLT-2019

MH Specific

ICDAR 2019 Robust Reading Challenge on Multi-lingual Scene Text Detection and Recognition. Covers 10 languages across 7 scripts including Devanagari, applicable to Marathi scene text

Build a multilingual scene text detector that handles Devanagari alongside other scripts in Indian street scenes.
Homepage

Quick Start

# ICDAR MLT 2019
import json
# Download from https://rrc.cvc.uab.es/
print('ICDAR MLT 2019: https://rrc.cvc.uab.es/')
print('Multi-lingual scene text detection and recognition')
Modality
Image (scene text)
Size
20K real images + 277K synthetic images
License
Format
PNG/JPEG
Language
mr
Update Frequency
static
Organization
ICDAR MLT Organizers
Category
vision-ocr-multimodal

Schema

FieldTypeDescription
imageimageScene image with multilingual text
annotationslist[object]Text bounding boxes with language and script labels

Build With This

Create an automated road sign reader for Maharashtra that handles Marathi, Hindi, and English text simultaneously
Develop a document language identifier from images that classifies Devanagari text as Marathi vs Hindi vs Sanskrit
Build a mixed-script OCR pipeline for Indian documents containing multiple languages and scripts

AI Use Cases

Multilingual scene text detection and recognition
Last verified: 2026-03-07