Lipi Toolkit is a generic toolkit for online Handwriting Recognition (HWR), and contains tools and algorithms for HWR. The current versions focus on isolated shapes and characters.
Supported platforms:
Lipi Toolkit can be installed on any of the following platforms:
· Windows XP Professional edition
· Redhat Enterprise Linux Edition 4.0
· Ubuntu Gutsy Gibbon 7.10
Product's homepage
What's New in This Release: [ read full changelog ]
Bug fixes:
· The size of euclidean filter should be specified as % and not a absolute value
· The parameter trainSet pointer to function trainFromListFile should be removed
· Give More understandable names to the command line options for the eval tool
· eval.pl shows the number of classes in the result file as 157 when there are only 156
· Evaluation scritpt should have a mode where it just computes the accuracy with out generating html files
· Rename "NNPreprocSequence" in nn.cfg to just "PreprocSequence"
· Calling runshaperec with invalid infiletype still runs
Improvements:
· Use getline operation for reading files
· Use STL map as a data structure to store the error code and the corresponding message
· Cleaning up LTKPreprocDefaults
· Overload recognize method in LTKShapeRecognizer class to take LTKShapeFeaturePtr as input