The CollabScore Dataset -Towards Robust and Generalized OMR Evaluation
Résumé
We present in this paper the CollabScore dataset, initially built to test and validate a new Optical Music Recognition (OMR) system. This dataset includes 200 pages of scores from an homogeneous corpus which consists in various works from a single composer, Camille Saint-Saëns (1835-1921). In addition to the essential data, namely images and reference scores encoded in MEI, we describe an original approach to content referencing based on the IIIF standard, and we discuss some general principles to conduct OMR evaluation in the proposed framework, including the tuning of comparison metrics.
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|---|
| Licence |