(Archived) feedback for handwritting and emailed scans to trans.mbox?

[evermullah 8]

hi en-staff,

concerning the community translation project.

i recently typed the words on the screenshot on the searchfield and got a totally wrong result 8-)

(its ok with me, because my handwritting is kinda mess, i know...)

that brought up the question: is there any feedback to the mail we send, or is the process silently done?

i let some friends write the text and emailed the images to your translations mailbox, but never get any information, if the mail arrived or proceeded or is on hold or anything else to let me know, that the job is done. (on my side, and on yours)

so whats the status here?

[engberg 89]

The search results for images and PDF documents don't have anything to do with our community translation. The translation system is used to translate the English words on Evernote's interfaces. The image processing/OCR is an automated system that's not related.

If you log on to the web and open a note there, you can view the note properties to see a link you can use to submit feedback about an incorrect image recognition. This can be used by our R&D team to improve our text processing.

On the other hand ... if you take a blurry photo of some handwriting, you may not get great results when you search.

[evermullah 8]

thanks for pointing that out. i corrected the thread-subject.

the webpage, i refered to, has the heading "Write to us in your language" but the community translation project is mentioned in the first lines, so i mixed it up, somehow.

so i ask my initial question again (in other words):

is there any feedback that the user who send you an email with his/hers handwritting scanned image gets?

i.e. "thanks, your image arrived and will be processed by our student apprentice if shes back from the supermarket where she buys us some drinks..." or "your handwritting is too illegible, so we had to dump it. please try again later..." ?

[engberg 89]

No, there's no easy way for us to tell the difference between those cases in any automated way.

[evermullah 8]

No, there's no easy way for us to tell the difference between those cases in any automated way.

hmm. my guess was, that a real person has to train the OCR-engine by hand with all the handwritting sample scans/photos sent to the translations mailbox.

if that is not the case, i really like to know how then you improve the OCR engine and what the handwritting sample images/photos are good for.

but maybe, i did not made myself clear enough and we're talking about different things :=)