Cloud Development Kit

Answered

How can I recognize different languages in one request?

while using  ->  /api/v3.0/recognition/rest/text/doSimpleRecognition.json

pls give me some tips.


Best Answer

Dear 宇航 余,


thank you for your questions.


Dear Phil, Thank you for your help! You are exactly right, were are using lexicons (basically based on the words/cities/ first names/last names...) specific to each country. The goal is to constrain the number of words that can be recognized to have the best accuracy.


Nevertheless, for each language, most common english words can be recognized (basically, each language contains the lexicon for english words). 


For any other use-case, e.g. Italian words mixed with Chinese, as the words will not be supported, the default Chinese configuration will not return with a satisfying result. One solution consist in creating a "custom lexicon", which in this latter case would contain the Italian words. This should return with a proper recognition.


Creating cutom resources can be done as explained in the following post: https://developer-support.myscript.com/support/discussions/topics/16000022917


Let us know if this helps.


Best regards,


Olivier


Hi,

I don't work for MyScript, but thought to chip in. The documentation has an ISO language option within the textInput field:

https://developer.myscript.com/refguides/cdk/cloud/3.2/#api-Text_Recognition-doSimpleRecognition

I set that based on my users preferred language.

Regards

Phil

thanks Phil

But now I want to recognize users' handwritten text, which may contain different languages on the same page, so...

Hi,

Wow, that seems like a really hard problem to solve. You are basically asking MyScript to detect every possible language on any given stroke group. Given the fact that they are probably using a language dictionary to help determine the word, it would imagine there is going to be massive overlap with different languages. 'risk' in English looks similar to 'fisk' in Swedish. Without knowing the context of the language I would imagine you'd drop the accuracy of the results.

I'd be interested to hear from MyScript if they can handle that.

Regards

Phil


Answer

Dear 宇航 余,


thank you for your questions.


Dear Phil, Thank you for your help! You are exactly right, were are using lexicons (basically based on the words/cities/ first names/last names...) specific to each country. The goal is to constrain the number of words that can be recognized to have the best accuracy.


Nevertheless, for each language, most common english words can be recognized (basically, each language contains the lexicon for english words). 


For any other use-case, e.g. Italian words mixed with Chinese, as the words will not be supported, the default Chinese configuration will not return with a satisfying result. One solution consist in creating a "custom lexicon", which in this latter case would contain the Italian words. This should return with a proper recognition.


Creating cutom resources can be done as explained in the following post: https://developer-support.myscript.com/support/discussions/topics/16000022917


Let us know if this helps.


Best regards,


Olivier

Login or Signup to post a comment