Currently using identifiers is enough to ensure that each object is provided a bounding box. The idea would be to incorporate this information in the same file as the transcription using a separate tree or list of elements. Moreover
Verovio generates bounding boxes of objects by default. A proof of concept dataset can be produced from there.
Tokens can have duplicate IDs or be hard copies of the same object. Therefore, when parsing these objects, there needs to be some form of mapping object that keeps track of all unique objects present.
Currently using identifiers is enough to ensure that each object is provided a bounding box. The idea would be to incorporate this information in the same file as the transcription using a separate tree or list of elements. Moreover