Projects‎ > ‎

Neuroph OCR - Handwriting Recognition

This application is part of Neuroph project, and it demonstrates how neural networks can be applied for handwritting recognition. It is an A.I. research open source student project. Project was started in October 2009 and it's first version was developed by students of FON - School of Business Administration, University of Belgrade, Serbia and mentored by Zoran Ševarac,teaching assistent at Department of Software Engineering. This alpha version of application is for demonstration purpose.

jHRT application uses artificial neural network to recognize handwritten letters and transforms them into editable document (as MS Word .doc or Notepad and Wordpad .txt file). It is based on neural network that can learn to recognize more characters.​

You can download Neuroph OCR jHRT alpha 0.2 at jHRT



Requirements


How does it work

  • Step 1
    First step would be to draw the actual letter that you want to be recognized. If the letter is drawn out of boundaries then it will be not recognized, in this case press 'clear' and try again


  • Step 2

When the 'recognized' button is pressed, draw letter is croped by the edges of character and resized to 20x20 pixels. This size is used in this application as a standard letter size. After processing, every image is sent to neural network for recognizing. Neural network recognizes every image using it's allready learned hand wrote letters, and as a result returns a list of letters with the matching percents.


  • Step 3
In this version of application for demo is used TIMES NEW ROMAN font, capital letters only.The characters are assembled as editable text in the text editing panel, and there they can be deleted, enlarged, bolded, italic or underlined.


  • In development
Images & training set allows the user to create letters for his own database which he has to train with

Performance issues


Since this project is still in it's alpha state, the success rate of recognition is not, yet, at the satisfactory level. Due to high level of different handwriting styles some letters may not be recognized successfully, however in all of the tested cases the letter in question was in top three results.

Plans


Current plans for the future for this project are:
  • To train new neural network with greater base of knowledge (Capital and small letters, numbers, punctuation characters, various fonts), which would help with the performance of our application.
  • To put the application on Netbeans platform

Related projects


Other info

  • This application was developed in Java in Netbeans 6.8 IDE

Developers


Students of FON - School of Business Administration, University of Belgrade, Serbia:

Mentor

Comments


Affiliations