Two medieval manuscripts containing music and text, prepared for Historical Document Understanding tasks. The dataset is an outcome of the Laudare ERC AdG project and includes diplomatic and editorial annotations. It was last updated on December 19, 2025.
Use Cases
- Train Optical Character Recognition models based on diplomatic and editorial text annotations.
- Benchmark Optical Music Recognition systems using the provided music annotations.
- Evaluate Layout Recognition algorithms on the structured manuscript pages.
- Simulate sequential learning scenarios using the provided incremental annotation splits.
- Perform 5-fold cross-validation for model comparison using the pre-defined experimental protocol.
Strengths
- Includes two distinct annotation versions: diplomatic (graphical signs) and editorial (scholarly interpretations).
- Provides pre-defined splits for 5-fold cross-validation and sequential learning tasks.
- Comes with a standardized experimental protocol for proper evaluation measures.
Limitations
- Description metadata is limited; actual data quality requires manual inspection after download.
- Row count is unknown, which may limit suitability assessment.
- Column-level documentation is absent; field semantics must be inferred after download.
Provenance
- Source
- CERN - European Organization for Nuclear Research and European Commission
- Collection Method
- Created as an outcome of the Laudare ERC AdG project.
- Freshness
- Last updated 2025-12-19 00:00:00; freshness should be verified.