Version 1.0 adds staff symbols (stafflines, staffspaces, and the container symbol staff) and relationships of various objects to these symbols, such as clef-staff, measure separator-staff, or notehead-staffline/staffspace. This enables inferring pitches, durations and onsets with relatively minimal heuristics, as implemented in the muscima package.
The staff objects and their relationships are added automatically, and have not been manually checked so far, although automated consistency checks have been implemented. The manual-only annotations are kept in a separate subdirectory, data/cropobjects_manual, and the files with staff objects and relationships added are kept in data/cropobjects_withstaff.
A few more errors have been fixed (see ERRATA).
Alexander Pacha (https://www.ims.tuwien.ac.at/people/alexander-pacha) went through all the symbols and provided a list of mistakes; all of them (about 120) have been fixed. The list of mistakes and what was done with them is given in the new ERRATA file. Big thanks to Alexander!
The MUSCIMA++ dataset has been released! We call the version 0.9, because it still lacks a few ingredients to enable fully inferring pitch and duration information, especially staff objects (stafflines, staffspaces, ...) and the relationship of other symbols to these objects