Publications

LSTM

2018

• Improving OCR Accuracy on Early Printed Books by combining Pretraining, Voting, and Active Learning (2018)
  Christian Reul, Uwe Springmann, Christoph Wick, Frank Puppe

2017

• Transfer Learning for OCRopus Model Training on Early Printed Books (2017)
  Christian Reul, Christoph Wick, Uwe Springmann, Frank Puppe [arXiv version]

• Improving OCR Accuracy on Early Printed Books by utilizing Cross Fold Training and Voting (2017)
  Christian Reul, Uwe Springmann, Christoph Wick, Frank Puppe

• OCR and post-correction of historical Finnish texts (2017)
  Senka Drobac, Pekka Kauppinen, Krister Lindén

2016

• (kraken) Important New Developments in Arabographic Optical Character Recognition (OCR) (2016)
  Maxim Romanov, Matthew Thomas Miller, Sarah Bowen Savant, Benjamin Kiessling
• OCR of historical printings with an application to building diachronic corpora: A case study using the RIDGES herbal corpus (2016)
  Uwe Springmann, Anke Lüdeling
• Automatic quality evaluation and (semi-) automatic improvement of mixed models for OCR on historical documents (2016)
  Uwe Springmann, Florian Fink, Klaus U. Schulz
• Generic Text Recognition using Long Short-Term Memory Networks (2016)
  Ph.D Thesis by Adnan Ul-Hasan
  Thesis Reviewers: Andreas Dengel, Faisal Shafait, Marcus Liwicki
• OCRoRACT: A Sequence Learning OCR System Trained on Isolated Characters (2016)
  Andreas Dengel, Adnan Ul-Hasan, Syed Saqib Bukhari

2015

• Benchmarking of LSTM Networks (2015)
  Thomas M. Breuel
• Recognition of Historical Greek Polytonic Scripts Using LSTM (2015)
  Fotini Simistira, Adnan Ul-Hassan, Vassilis Papavassiliou, Basilis Gatos, Vassilis Katsouros, Marcus Liwicki
• A Segmentation-Free Approach for Printed Devanagari Script Recognition (2015)
  Tushar Karayil, Adnan Ul-Hasan, Thomas M. Breuel
• A Sequence Learning Approach for Multiple Script Identification (2015)
  Adnan Ul-Hasan, Muhammad Zeshan Afzal, Faisal Shfait, Marcus Liwicki, Thomas M. Breuel
• Statistical Language Modeling for Historical Documents using Weighted Finite-State Transducers and Long Short-Term Memory (2015)
  Ph.D Thesis by Mayce Al Azawi
  Thesis supervisors: Thomas M Breuel, Andreas Dengel, Marcus Liwicki

2014

• OCR of historical printings of Latin texts: Problems, Prospects, Progress (2014)
  Uwe Springmann, Dietmar Najock, Hermann Morgenroth, Helmut Schmid, Annette Gotscharek, Florian Fink

2013

• High Performance OCR for Printed English and Fraktur using LSTM Networks (2013)
  Thomas M. Breuel, Adnan Ul-Hasan, Mayce Al Azawi. Faisal Shafait
• Can we build language-independent OCR using LSTM networks? (2013)
  Adnan Ul-Hasan, Thomas M. Breuel
• Offline Printed Urdu Nastaleeq Script Recognition with Bidirectional LSTM Networks (2013)
  Adnan Ul-Hasan, Saad Bin Ahmed, Sheikh Faisal Rashid, Faisal Shafait, Thomas M. Breuel
• ICDAR 2013 LSTM Tutorial - Building Fast High-Performance Recognition Systems with Recurrent Neural Networks and LSTM
  Thomas Breuel, Volkmar Frinken, Marcus Liwicki

Binarization

A paper describing the algorithm used in 'ocropus-nlbin':

• Robust Binarization of Stereo and Monocular Document Images Using Percentile Filter (2013)
  Muhammad Zeshan Afzal, Martin Krämer, Syed Saqib Bukhari, Mohammad Reza Yousefi, Faisal Shafait, Thomas M. Breuel

Version 0.6

This old version used a different OCR engine than the current LSTM-based version.

• Local Logistic Classifiers for Large Scale Learning (2012)
  Mohammad Reza Yousefi, Thomas M. Breuel

Older versions

• The OCRopus open source OCR system (2008) Thomas M. Breuel

• Recent progress on the OCRopus OCR system (2009) Thomas Breuel