Awesome OCR

This list contains links to great software tools and libraries and literature related to Optical Character Recognition (OCR).

Contributions are welcome, as is feedback.

• Software
  • OCR engines
  • Older and possibly abandoned OCR engines
  • OCR file formats
    • hOCR
    • ALTO XML
    • TEI
    • PAGE XML
  • OCR CLI
  • OCR GUI
  • OCR Preprocessing
  • OCR as a Service
  • OCR evaluation
  • OCR libraries by programming language
    • Crystal
    • Elixir
    • Go
    • Java
    • .Net
    • Object Pascal
    • PHP
    • Python
    • Javascript
    • Ruby
    • Swift
    • Rust
    • R
  • OCR training tools
• Datasets
  • Ground Truth
• Literature
  • OCR-related publication and link lists
  • Blog Posts and Tutorials
  • OCR Showcases
  • Academic articles
    • 2011 and before
    • 2012
    • 2013
    • 2014
    • 2015
    • 2016
    • 2017
    • 2018

Software

OCR engines

• tesseract - The definitive Open Source OCR engine Apache 2.0
• EasyOCR - OCR engine built on PyTorch by JaidedAI, Apache 2.0
• ocropus - OCR engine based on LSTM, Apache 2.0
• ocropus 0.4 - Older v0.4 state of Ocropus, with tesseract 2.04 and iulib, C++
• kraken - Ocropus fork with sane defaults
• gocr - OCR engine under the GNU Public License led by Joerg Schulenburg.
• Ocrad - The GNU OCR. GPL
• ocular - Machine-learning OCR for historic documents
• SwiftOCR - fast and simple OCR library written in Swift
• attention-ocr - OCR engine using visual attention mechanisms
• RWTH-OCR - The RWTH Aachen University Optical Character Recognition System
• simple-ocr-opencv and its fork - A simple pythonic OCR engine using opencv and numpy
• Calamari - OCR Engine based on OCRopy and Kraken
• doctr - A seamless & high-performing OCR library powered by Deep Learning

Older and possibly abandoned OCR engines

• Clara OCR - Open source OCR in C GPL
• Cuneiform - CuneiForm OCR was developed by Cognitive Technologies
• Eye - an experimental Java OCR (image-to-text) application
• kognition - An omnifont OCR software for KDE
• OCRchie - Modular Optical Character Recognition Software
• ocre - o.c.r. easy
• xplab - A GTK 2 tool for pattern matching
• hebOCR - Hebrew character recognition library (previously named hocr, see Wikipedia article) GPL

OCR file formats

• abby2hocr.xslt XSLT script
• ocr-conversion-scripts

hOCR

• hocr-tools - Tools for doing various useful things with hOCR files, Apache 2.0
• hocr-spec - hOCR 1.2 specification
• ocr-transform - CLI tool to convert between hOCR and ALTO, MIT
• hocr-parser - hOCR Specification Python Parser
• hOCRTools - hOCR to ALTO conversion XSLT

ALTO XML

• ALTO XML Schema - XML Schema and development of the ALTO XML format
• ALTO XML Documentation - Documentation and use cases for ALTO
• alto-tools - Various tools to work with ALTO files, Python
• AbbyyToAlto - PHP script converting from Abbyy 6 to ALTO XML

TEI

• TEI-OCR - TEI customization for OCR generated layout and content information
• TEI SIG on Libraries - Best Practices for TEI in Libraries
• GDZ - METS/TEI-based GDZ document format

PAGE XML

• PAGE-XML Schema - XML schema of the PAGE XML format along with documentation and examples
• omni:us Pages Format (OPF) - XML schema very similar to PAGE XML that has some additional features.
• py-pagexml - Python library for handling PAGE XML and OPF files.

OCR CLI

• OCRmyPDF - OCRmyPDF adds an OCR text layer to scanned PDF files, allowing them to be searched
• Pdf2PdfOCR - A tool to OCR a PDF (or supported images) and add a text "layer" (a "pdf sandwich") in the original file making it a searchable PDF. GUI included. Tesseract and cuneiform supported.
• Ocrocis - Project manager interface for Ocropy, see also external project homepage
• tesseract-recognize - Tesseract-based tool that outputs result in Page XML format (docker image).

OCR GUI

• moz-hocr-editor - Firefox Addon for editing hOCR files Discontinued
• qt-box-editor - QT4 editor of tesseract-ocr box files.
• ocr-gt-tools - Client-Server application for editing OCR ground truth.
• Paperwork - Using scanners and OCR to grep paper documents the easy way.
• Paperless - Scan, index, and archive all of your paper documents.
• gImageReader - gImageReader is a simple Gtk/Qt front-end to tesseract-ocr.
• VietOCR - A Java/.NET GUI frontend for Tesseract OCR engine, including jTessBoxEditor a graphical Tesseract box data editor
• PoCoTo - Fast interactive batch corrections of complete OCR error series in OCR'ed historical documents.
• OCRFeeder - GTK graphical user interface that allows the users to correct characters or bounding boxes, ODT export and more.
• PRImA PAGE Viewer - Java based viewer for PAGE XML files (layout + text content). Also supports ALTO XML, FineReader XML, and HOCR.
• LAREX - A semi-automatic open-source tool for Layout Analysis and Region EXtraction on early printed books.
• archiscribe - Web application for transcribing OCR ground truth from Archive.org. Deployed instance available at https://archiscribe.jbaiter.de/, results are available in @jbaiter/archiscribe-corpus.
• nw-page-editor - Simple app for visual editing of Page XML files. Provides desktop and server docker-based versions.

OCR Preprocessing

• NoiseRemove.java in MathOCR - Java implementation of Adaptive degraded document image binarization by B. Gatos , I. Pratikakis, S.J. Perantonis
• binarize.c in ZBar - C implementations of two binarization algorithms, based on Sauvola
• typeface-corpus - A repository for typefaces to train Tesseract and OCRopus for natural history collections and digital humanities.
• binarizewolfjolion - Comparison of binarization algorithms. Blog post
• crop_morphology.py in oldnyc - Cropping a page to just the text block
• Whiteboard Picture Cleaner - Shell one-liner/script to clean up and beautify photos of whiteboards
• Fred's ImageMagick script textcleaner - Processes a scanned document of text to clean the text background
• localcontrast - Fast O(1) local contrast optimization

OCR as a Service

• Open OCR - Run Tesseract in Docker containers
• tesseract-web-service - An implementation of RESTful web service for tesseract-OCR using tornado.
• docker-ocropy - A Docker container for running the ocropy OCR system.
• ABBYY Cloud OCR SDK Code samples - Code samples for using the proprietary commercial ABBYY OCR API.
• nidaba - An expandable and scalable OCR pipeline
• gamera - A meta-framework for building document processing applications, e.g. OCR
• ocr-tools - Project to provide CLI and web service interfaces to common OCR engines
• ocrad-docker - Run the ocrad OCR engine in a docker container
• kraken-docker - Run the kraken OCR engine in a docker container
• Konfuzio - Free Online OCR up to 2.000 pages per month and OCR API by [@atraining], see https://youtu.be/NZKUrKyFVA8 (code is not open)
• ocr.space - Free Online OCR and OCR API by @a9t9 based on Tesseract (code is not open)
• OCR4all - Provides OCR services through web applications. Included Projects: LAREX, OCRopus, calamari and nashi.

OCR evaluation

• ISRI OCR Evaluation Tools with a User Guide from 1996 :!:
  • isri-ocr-evaluation-tools - further development by @eddieantonio (2015, 2016)
  • ancientgreekocr-evaluation-tools - further development by @nickjwhite (2013, 2014)
• ocrevalUAtion - Cross-format evaluation, CLI and GUI
• ngram-ocr-eval - Brute and simple OCR evaluation using ngrams
• quack - Quality-Assurance-tool for scans with corresponding ALTO-files

OCR libraries by programming language

Crystal

• tesseract-ocr - A Crystal wrapper for tesseract-ocr.

Elixir

• tesseract_ocr - Elixir library wrapping the tesseract executable.

Go

• gosseract - Golang OCR library, wrapping Tesseract-ocr.

Java

• Tess4J - Java Native Access bindings to Tesseract.
• tess-two - Tools for compiling Tesseract on Android and Java API.

.Net

• tesseract for .net - A .Net wrapper for tesseract-ocr.

Object Pascal

• TTesseractOCR4 - Object Pascal binding for tesseract-ocr 4.x.

PHP

• Tesseract OCR for PHP - Tesseract PHP bindings.

Python

• pytesseract - A Python wrapper for Google Tesseract.
• pyocr - A Python wrapper for Tesseract and Cuneiform.
• ocrodjvu - A library and standalone tool for doing OCR on DjVu documents, wrapping Cuneiform, gocr, ocrad, ocropus and tesseract
• tesserocr - A Python wrapper for the tesseract-ocr API

Javascript

• ocracy - pure javascript lstm rnn implementation based on ocropus
• gocr.js - Javascript port (emscripten) of gocr
• ocrad.js - Javascript port (emscripten) of ocrad
• tesseract.js - Javascript port (emscripten) of Tesseract
• node-tesseract-ocr - A simple wrapper for the Tesseract OCR package.
• node-tesseract-native - C++ module for node providing OCR with tesseract and leptonica.

Ruby

• rtesseract - Ruby library wrapping the tesseract and imagemagick executables.
• ruby-tesseract - Native Tesseract bindings for Ruby MRI and JRuby
• ocr_space - API wrapper for free ocr service ocr.space. Includes CLI

Rust

• tesseract.rs - Rust bindings for tesseract OCR.
• leptess - Productive and safe Rust bindings/wrappers for tesseract and leptonica.

R

• tesseract - R bindings for tesseract OCR.

Swift

• Tesseract OCR iOS - Swift and Objective-C wrapper for Tesseract OCR.
• SwiftOCR - Fast and simple OCR library written in Swift. Optimized for recognizing short, one line long alphanumeric codes.

OCR training tools

• glyph-miner - A system for extracting glyphs from early typeset prints
• ocrodeg - Document image degradation for OCR data augmentation

Datasets

Ground Truth

• archiscribe-corpus - >4,200 lines transcribed from 19th Century German prints via archiscribe CC-BY 4.0
• CIS OCR Test Set - 2 example documents each in German/Latin/Greek with ground truth for PoCoTo
• Rescribe - Transcriptions of Caroline Minuscule Manuscripts PDM 1.0
• CLTK - Corpora from Classical Language Toolkit PDM 1.0
• DIVA-HisDB - 150 pages^PAGE-XML of three medieval manuscripts CC-BY-NC 3.0
• EarlyPrintedBooks - ~8,800 lines from several early printed books CC-BY-NC-SA 4.0
• EEBO-TCP - 25,363 EEBO documents transcribed by TCP PDM 1.0
• ECCO-TCP - 2,188 ECCO documents transcribed by TCP PDM 1.0
• eMOP-TCP - 2,188 ECCO-TCP documents, cleaned up by eMOP PDM 1.0
• Evans-TCP - 4,977 Evans documents transcribed by TCP
• FDHN - Finnish Digitised Historical Newspapers, Paper, (free) registration required, Terms of Use
• FROC-MSS - 4 Old French Medieval Manuscripts CC-BY 4.0
• GERMANA - 764 Spanish manuscript pages, (free) registration required non-commercial use only
• GT4HistOCR - Ground Truth for German Fraktur and Early Modern Latin CC-BY 4.0
• imagessan - Sanskrit images & ground truth (Devanagari script)
• IMPACT-BHL - 2,418 pages^PAGE-XML from the Biodiversity Heritage Library, XML@GitHub CC-BY 3.0
• IMPACT-BL - 294 pages^PAGE-XML from the British Library, (free) registration required PDM 1.0
• IMPACT-BNE - 215 pages^PAGE-XML from the National Library of Spain, (free) registration required, XML@GitHub CC-BY-NC-SA 4.0
• IMPACT-BNF - 151 pages^PAGE-XML from the National Library of France, (free) registration required CC-BY-NC-SA 4.0
• IMPACT-KB - 142 pages^PAGE-XML from the National Library of the Netherlands CC-BY 4.0
• IMPACT-NKC - 187 pages^PAGE-XML from the Czech National Library, (free) registration required CC-BY-NC-SA 4.0
• IMPACT-NLB - 19 pages^PAGE-XML from the National Library of Bulgaria, (free) registration required CC-BY-NC-ND 4.0
• IMPACT-NUK - 209 pages^PAGE-XML from the National Library of Slovenia, (free) registration required CC-BY-NC-SA 4.0
• IMPACT-PSNC - 478 pages^PAGE-XML from four Polish digital libraries, XML@GitHub CC-BY 3.0
• LascivaRoma/lexical - Transcription of 19th century lexical resources for Latin learning
• MJSynth - 9m synthetic images covering 90k English words
• OCR19thSAC - 19,000 pages Swiss Alpine Club yearbooks transcribed via Text+Berg digital CC-BY 4.0
• OCR-D - 180 pages^PAGE-XML of German historical prints from OCR-D CC-BY-SA 4.0
• OCR_GS_Data - Double-checked Arabic Gold Standard from OpenITI
• old-books - 322 old books from Project Gutenberg GPL 3.0
• PRImA-ENP - 528 pages^PAGE-XML historic newspapers from Europeana Newspapers, (free) registration required PDM 1.0
• RODRIGO - 853 Spanish manuscript pages, (free) registration required non-commercial use only
• Toebler-OCR - (Kraken) Ground Truth transcription of few pages of the Tobler-Lommatzsch: Altfranzösisches Wörterbuch

Literature

OCR-related publication and link lists

• IMPACT: Tools for text digitisation - List of tools software projects related, some related to OCR
• OCR-D - List of OCR-related academic articles in the context of the OCR-D project. :de:
• Mendeley Group "OCR - Optical Character Recognition" - Collection of 34 papers on OCR
• eadh.org projects - List of Digital Humanities-related projects in Europe, some related to OCR
• Wikipedia: Comparison of optical character recognition software
• OCR [and Deep Learning] by @handong1587
• Ocropus Wiki: Publications

Blog Posts and Tutorials

• Tesseract Blends Old and New OCR Technology (2016) @theraysmith
  • Tutorial@DAS2016, Updated "What You Always Wanted to Know" slides
• What You Always Wanted To Know About Tesseract (2014) @theraysmith
  • Tutorial@DAS2014, includes demos
• Extracting text from an image using Ocropus (2015)
• Training an Ocropus OCR model (2015) @danvk
• Ocropus Wiki: Compute errors and confusions (2016) @zuphilip
• Ocropus Wiki: Working with Ground Truth (2016) @zuphilip
• OCRopus (2016) @jze
  • mostly on column separation in ocropus
• 10 Tips for making your OCR project succeed (2013) @cneud
  • general things to consider for OCR projects
• Overview of LEADTOOLS Image Cleanup and Pre-processing SDK Technology -
  • feature list for a commercial image pre-processing library; has nice before-after samples for pre-processing steps related to OCR
• Extracting Text from PDFs; Doing OCR; all within R @shawngraham
  • How to work with OCR from PDFs in the R programming environment
• Tutorial: Command-line OCR on a Mac @bmschmidt
  • Tutorial on how to run tesseract in Mac OSX
• Practical Expercience with OCRopus Model Training (2016) @jze
• Homemade Manuscript OCR (1): OCRopy (2017) @Jean-Baptiste-Camps
  • Tutorial on applying OCR to medieval manuscripts with OCRopy
• Optimizing Binarization for OCRopus (2017) @jze
• Prototype demo for OCR postfix in Danish Newspapers (2016) @thomasegense
• How Can I OCR My Dictionary? (2016) @JessedeDoes
• "Needlessly complex" blog (2016) @mzucker. Several image processing how-tos (Python based), particularly:
  • Page dewarping (code)
  • Compressing and enhancing hand-written notes (code)
  • Unprojecting text with ellipses (code)
• (Open-Source-)OCR-Workflows (2017) @wrznr :de: overview of the state of the art in open source OCR and related technologies (binarisation, deskewing, layout recognition, etc.), lots of example images and information on the @OCR-D project.
• A gentle introduction to OCR (2018) @shgidi
• Worauf kann ich mich verlassen? Arbeiten mit digitalisierten Quellen, Teil 1: OCR (2019) @eliaskreyenbuehl :de: A reflection/criticism on OCR quality, OCR pitfalls in Fraktur fonts.

OCR Showcases

• abbyy-finereader-ocr-senate - Using OCR to parse scanned Senate Financial Disclosure forms.
• cvOCR - An OCR system for recognizing resume or cv text, implemented in Python and C and based on tesseract
• MathOCR - A printed scientific document recognition system, pre-alpha

Academic articles

2011 and before

• High performance document layout analysis (2003) Breuel
• Adaptive degraded document image binarization (2006) Gatos, Pratikakis, Perantonis
• [Internship Report] (2007) Gupta
• OCRopus Addons (Internship Report) (2007) Dantrey

2012

• Local Logistic Classifiers for Large Scale Learning (2012) Yousefi, Breuel

2013

• High Performance OCR for Printed English and Fraktur using LSTM Networks (2013) Breuel, Ul-Hasan, Mayce Al Azawi. Shafait
• Can we build language-independent OCR using LSTM networks? (2013) Ul-Hasan, Breuel
• Offline Printed Urdu Nastaleeq Script Recognition with Bidirectional LSTM Networks (2013) Ul-Hasan, Ahmed, Rashid, Shafait, Breuel

2014

• OCR of historical printings of Latin texts: Problems, Prospects, Progress. (2014) Springmann, Najock, Morgenroth, Schmid, Gotscharek, Fink
• Correcting Noisy OCR: Context beats Confusion (2014) Evershed, Fitch

2015

• TypeWright: An Experiment in Participatory Curation (2015) Bilansky
  • On crowd-sourcing OCR postcorrection
• Benchmarking of LSTM Networks (2015) Breuel
• Recognition of Historical Greek Polytonic Scripts Using LSTM (2015) Simistira, Ul-Hassan, Papavassiliou, Basilis Gatos, Katsouros, Liwicki
• A Segmentation-Free Approach for Printed Devanagari Script Recognition (2015) Karayil, Ul-Hasan, Breuel
• A Sequence Learning Approach for Multiple Script Identification (2015) Ul-Hasan, Afzal, Shfait, Liwicki, Breuel

2016

• Important New Developments in Arabographic Optical Character Recognition (OCR) (2016) Romanov, Miller, Savant, Kiessling
  • on kraken
  • using OpenArabic/OCR_GS_Data for ground truth data
• OCR of historical printings with an application to building diachronic corpora: A case study using the RIDGES herbal corpus (2016) Springmann, Lüdeling
• Automatic quality evaluation and (semi-) automatic improvement of mixed models for OCR on historical documents (2016) Springmann, Fink, Schulz
• Generic Text Recognition using Long Short-Term Memory Networks (2016) Ul-Hasan -- Ph.D Thesis
• OCRoRACT: A Sequence Learning OCR System Trained on Isolated Characters (2016) Dengel, Ul-Hasan, Bukhari
• Recursive Recurrent Nets with Attention Modeling for OCR in the Wild (2016) Lee, Osindero

2017

• Telugu OCR Framework using Deep Learning (2015/2017) Achanta, Hastie
  • see also TeluguOCR, banti_telugu_ocr, chamanti_ocr, #49

2018

• A Two-Stage Method for Text Line Detection in Historical Documents (2018) Grüning, Leifert, Strauß, Labahn. Code available at https://github.com/TobiasGruening/ARU-Net