Dear folks,
I'm an open source developer, I recently play with deep learning and
I'm hoping to contribute a music sheet OCR model to musescore project.
Our recent result is a formula OCR model, which you can view here:
http://namepredict.com:8888/The main contribution are done by the Harvard NLP team, I only
done a little bit simple work, help them create a formula-matrix-70k dataset.
With 70,000 pairs of formulas and images extracted from
math.stackexchange.com, our model successfully recognize new formula
which hasn't been seen.
We hope to build a similar dataset, called musescore-20k or
musescore-100k, depending on how much data we can get. In order to
benefit the whole machine learning community, we hope to open the
musescore-20k dataset to everyone (every music sheets licensed as "To
share").
I asked for a consumer API key according to
http://developers.musescore.com/ , but haven't received reply yet. Do
you know if MuseScore provide any open database dump like
https://archive.org/details/stackexchange? Or is there any simple way to enumerate all sheets licensed as "To
share" so we can download in a batch?
Thank you!
--
Regards,
Qian Hong
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, SlashDot.org!
http://sdm.link/slashdot_______________________________________________
Mscore-developer mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/mscore-developer