[Corpora-List] Handwritten recognition without LM

remi lavalley remi.lavalley at gmail.com
Tue Oct 14 09:01:19 CEST 2014


Dear Corpora members,

I plan to collect a corpus of handwritten texts in German, by using electronic pens + handwritten recognition software. The problem is that I would like to obtain the original texts, including spelling errors (ideally, both of the corrected text and the text that includes the original misspellings).

But all the softwares I have found until now use some kind of language model to perform the recognition (i.e., if I write "cotpora list", it will be transcribed as "corpora list"), which makes sense to improve the efficiency of the recognition, but is problematic in my case.

Is anyone aware of a solution that would fit with my needs, by doing (efficient) recognition, but without language modelling?

Thank you,

Remi -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: text/html Size: 828 bytes Desc: not available URL: <https://mailman.uib.no/public/corpora/attachments/20141014/59d4487d/attachment.txt>



More information about the Corpora mailing list