Skip to content. | Skip to navigation

Web-Based Offline Handwritten Text Line Recognizer
Not logged in. Login.
 

New Text Line Image

Upload your own text line image and recognize it with one of the provided recognition systems. Please consider the following caveats for a successful recognition of your images
  • A single text line must be on the image, multi-line recognition is not supported
  • The image must show enough contrast
  • Handwriting must be dark while background should be bright
  • The resolution of the image should be approx. 300 dpi
  • Most of the available recognition systems are trained on the IAM database. The more the uploaded handwriting resembles images of the IAM database, the higher will be the recognition performance

Back
Text Line Recognizer
Concept and Implementation
by Roman Bertolami
Thanks to
Peppo Brambilla
Horst Bunke
Andreas Fischer
Volkmar Frinken
Alex Graves
Simon Günter
Emanuel Indermühle
Marcus Liwicki
Urs-Victor Marti
Tamas Varga
Matthias Zimmermann
Included Tools/Frameworks
Hidden Markov Model Toolkit HTK
Image Magick
Ruby on Rails
SRI Language Modeling Toolkit
Licence
Web-Based Offline Handwritten Text Line Recognizer is released under Creative Commons Attribution-Noncommercial 2.5 Switzerland.
Creative Commons License
 
© FKI: Research Group on Computer Vision and Artificial Intelligence IAM, University of Bern