jspast/paddlepaddle-table-models-onnx

ONNX weights for PaddlePaddle table models

Models

• table_cls.onnx: Classifies table images into wired and wireless tables (original model)
• wired_table_cell_det.onnx: Detects cells from wired table images (original model)
• wireless_table_cell_det.onnx: Detects cells from wireless table images (original model)

Usage

With cells2table, a docling plugin is provided to allow integrating these models in a complete pipeline.

Install with uv:

uv pip install cells2table

Example:

from cells2table.docling import CustomDoclingTableStructureOptions

pipeline_options = PdfPipelineOptions(
    allow_external_plugins=True,
    table_structure_options=CustomDoclingTableStructureOptions(),
)

converter = DocumentConverter(
    format_options={
        InputFormat.PDF: PdfFormatOption(pipeline_options=pipeline_options),
        InputFormat.IMAGE: PdfFormatOption(pipeline_options=pipeline_options),
    }
)

result = converter.convert("path/to/document.pdf")
print(result.document.export_to_markdown())