People
Projects
Software
Datasets
        
 

DOCLIB Sample Driver
Researchers
Overview
DOCLIBDemo is a combined driver of several DOCLIB add-ons designed for use on the Windows Command Prompt. It is intended to provide a sense of what the add-ons can do (note that the add-ons themselves are cross-platform). As such, some of the add-ons in the demo may have limited functionality as compared to the full versions. The add-ons included in the demo are as follows:

Docstrum - an implementation of the Docstrum image segmenation algorithm

Jargon - used to determine the language of the provided image file, based on the provided template file (several example template files are included)

LogoDetectDOCLIB - the DOCLIB implementation of a logo detection algorithm

LogoDetectUMD - the University of Maryland's implementation of a logo detection algorithm

MachineHandwriting - used to determine if the provided image file is handwritten or machine print

SignatureDetect - the University of Maryland's implementation of a signature detection algorithm

StampDetect - the University of Maryland's implementation of a stamp detection algorithm

TextLineDetection - used to detect the number and location of text lines in the provided image file

ZoneSegmentation - an implementation of the Voronoi++ algorithm

To use DOCLIBDemo, just follow the provided help text:

C:/>DOCLIBDemo.exe -h
DOCLIBDemo toolName [options]

     To see the options of a particular tool, use DOCLIBDemo toolName -h

The following tools are enabled in this demo:
          Docstrum
          Jargon
          LogoDetectDOCLIB
          LogoDetectUMD
          MachineHandwriting
          SignatureDetect
          StampDetect
          TextLineDetection
          ZoneSegmentation

Results can be viewed with the GEDI tool, also available from this site.

DISCLAIMER: Users interested in the package should contact the UMD researchers below. DocLib is an SDK tool kit that is intended to be integrated with the users system. It includes not only the add-ons but also the underlying document analysis tools that support them. In addition, the add-ons my have options that are not exposed in this Sample Driver. The performance using this driver, on an arbitrary data set should not be used as the only indicator of the capabilities of the tools.
Contact
David Doermann: doermann@umiacs.umd.edu
Software
Publications


Last Updated: Wednesday 19 August, 2009



home | language group | media group | sponsors & partners | publications | seminars | contact us | staff only
© Copyright 2001, Language and Media Processing Laboratory, University of Maryland, All rights reserved.