Audiveris / omr-dataset-tools

Reference of OMR data
GNU Affero General Public License v3.0
18 stars 5 forks source link

Repeat dots are not yet annotated #24

Closed hbitteur closed 6 years ago

hbitteur commented 7 years ago

We need examples for all the various meanings of dots.

nasehim7 commented 6 years ago
screen shot 2018-05-19 at 2 58 11 pm

Is this the kind of output we want after resolving this issue? :)

hbitteur commented 6 years ago

@nasehim7 Perhaps I don't really understand your question

My position is that reference data should cover as much configurations as possible. This include the repeat signs. In this example, there are 2 repeat signs, one left and one right. For each of them, try to provide the outer compound (such as left repeat sign) and its inner members (thick barline, thin barline, top repeat dot, bottom repeat dot). Sorry, I'm writing this away from my computer, I don't have the exact symbol names at hand.

nasehim7 commented 6 years ago

@hbitteur No problem, I got it. We want a nested approach in this one too. Working on it, currently, I'm done with getting the outer element which can be seen in the above control image. Will get back soon. :)

nasehim7 commented 6 years ago

I did the nesting part. For Reference, Below are the control images:

screen shot 2018-05-20 at 1 59 19 am screen shot 2018-05-20 at 1 59 31 am

Hope it is in sync with what we want for the OMR. @hbitteur @lasconic

hbitteur commented 6 years ago

This looks OK, thanks

nasehim7 commented 6 years ago

@hbitteur what exactly are the names of these thick and thin barlines? :)

lasconic commented 6 years ago

You can use the SMuFL names : http://www.smufl.org/version/latest/range/barlines/

nasehim7 commented 6 years ago

@hbitteur @lasconic I think now the implementation is more refined than the previous one. Below are the control images:

screen shot 2018-05-22 at 12 33 41 am screen shot 2018-05-22 at 12 06 40 am

I have added the commit to the existing PR. Hope it suits well. :)

nasehim7 commented 6 years ago

@lasconic Should we close this too? It will be good in the next dataset. :)