MonOCR (Universal SDK)

MonOCR is a high-performance, cross-platform Optical Character Recognition (OCR) engine for the Mon language (mnw). Powered by ONNX Runtime, it provides a unified API for image and PDF text recognition across multiple programming environments.

This repository is the central hub for the MonOCR SDKs, maintained by the MonDevHub organization.

Supported Platforms

SDK	Directory	Registry	Status
JavaScript/Node.js	js/	npm: monocr	✅ Production
Python	python/	PyPI: monocr-onnx	✅ Production
Go	go/	github.com/MonDevHub/monocr-onnx/go	✅ Production
Rust	rust/	-	🚧 Planned

Core Features

• Unified API: Identical read_image and read_pdf patterns across all languages.
• Zero-Config Model Management: Automatically fetches and caches the 56MB ONNX model from Hugging Face on first use.
• Layout Awareness: Built-in line segmentation for full-page document processing.
• Optimized Performance: Leverages modern SIMD/GPU acceleration via ONNX Runtime.

Quick Installation

JavaScript (npm)

npm install monocr || pnpm add monocr

Python (pip or uv)

pip install monocr-onnx || uv add monocr-onnx

Go (modules)

go get github.com/MonDevHub/monocr-onnx/go

Quick Start Example (Python)

from monocr_onnx import read_image

# Model is automatically downloaded and cached at ~/.monocr/models/
text = read_image("scanned_document.jpg")
print(text)

Documentation

Each SDK contains its own detailed documentation and examples:

• JavaScript Documentation
• Python Documentation
• Go Documentation
• Publishing & Deployment Guide

Model Hub

The underlying weights and multi-format exports (ONNX, TFLite, PyTorch) are hosted on Hugging Face: MonDevHub/monocr

License

MIT License. See LICENSE for details.