Invoice images have been processed through optical character recognition (OCR) to generate multiple output formats. The dataset includes the original images alongside extracted markdown and structured JSON representations. The specific source, volume, and creation date of the invoices are not provided in the available metadata.
Use Cases
- Train or evaluate OCR models based on the provided invoice images.
- Develop parsers for converting unstructured invoice text into structured data based on the JSON outputs.
- Benchmark multi-format document processing systems using the image, markdown, and JSON triad.
- Create synthetic training data for invoice analysis by modifying the structured JSON and regenerating layouts.
Strengths
- Provides multiple synchronized data representations (image, markdown, JSON) for the same source document.
- Focuses on a specific, commercially relevant document type (invoices).
Limitations
- Row count and dataset scale are unknown, which may limit suitability assessment.
- Column-level documentation is absent; field semantics must be inferred after download.
- Description metadata is limited; actual data quality requires manual inspection after download.
Provenance
- Collection Method
- Invoice images processed through optical character recognition (OCR).