Based on my previous ticket, I wanted to understand if we can train MyScript Interactive SDK for making sure of correct recognition.
Currently for some user specially where single letter written with multiple strokes are 50% wrongly recognized.
for eg: [/\/\] for M is some time recognized as iN or Ni IVI
Is there a way where I can update myscript with a feedback about stroke recognision using JIIX or something?
Dear Irshad Chohan,
What probably happens in your case is that you are trying to recognize words that are not part of the engine lexicon, so the engine falls back to a mode where there is no lexical attraction.
As it is not possible to "live train" the MyScript engine, you could try creating a lexicon containing as many indian names as possible (that lexicon would be created the same way I described above).
PS : you will find extra information the configuration files format here.
Also any comment on whether we can live train myscript engine, saying these strokes represent 'M'?
Thanks for the input, We are using it to recognize Indian Names (Full Name: First Name Last Name).
But problem happens specially with first letter on the name.
Also on side note, can you elaborate on following properties.
SetTextListSize 1 SetWordListSize 5 SetCharListSize 1
Dear Irshad Chohan,
If you want to recognize only a single character in a ContentPart, you should create a lexicon in which you put single letters only. This will improve very much the recognition process, as the iink SDK engine will know that it must recognize only single letters.
1) To learn how to create a text lexicon, please refer to: http://doc.myscript.com/Common/DevTools/index.html (in your case, you should create a text file that lists all letters you want to recognize, one letter per line).
2) To build the associated resource file, go to: https://developer.myscript.com/support/recognition-assets (you will get a .res file).
3) To use this resource, you must change the .conf file associated with the language you target. In case you target en_US language, open `recognition-assets/conf/en_US.conf` and append a new configuration name and description, as follow (rename the MY_ISOLATED_RESOURCE with your resource name):
... existing content ... ... (leave empty line below) ... Name: isolated Type: Text Configuration-Script: AddResource en_US/en_US-ak-cur.res AddResource MY_ISOLATED_RESOURCE.res EnableAlienCharacters SetTextListSize 1 SetWordListSize 5 SetCharListSize 1
4) Change the text configuration name at runtime:
This will force the iink SDK engine to recognize only one character per ContentPart, which will improve the recognition accuracy.