mirrors/docling
1
0
Fork 0
mirror of https://github.com/DS4SD/docling.git synced 2025-07-25 19:44:34 +00:00
Code Issues Packages Projects Releases Wiki Activity

- Fixes for scaling transformation for table cell bounding boxes when using do_cell_matching = False

Browse Source 
- Corrected examples/convert.py with appropriate parameter, for good quality example conversion

Signed-off-by: Maxim Lysak <mly@zurich.ibm.com>
This commit is contained in:
Maxim Lysak 2024-07-30 13:43:50 +02:00
parent b07c4a7a4a
commit cfdfda3629
2 changed files with 11 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

5
docling/models/table_structure_model.py
View File

@ -114,12 +114,13 @@ class TableStructureModel:
for element in table_out["tf_responses"]:
if not self.do_cell_matching:
the_bbox = BoundingBox.model_validate(element["bbox"])
the_bbox = BoundingBox.model_validate(element["bbox"]).scaled(1 / self.scale)
text_piece = page._backend.get_text_in_rect(the_bbox)
element["bbox"]["token"] = text_piece
tc = TableCell.model_validate(element)
tc.bbox = tc.bbox.scaled(1 / self.scale)
if self.do_cell_matching:
tc.bbox = tc.bbox.scaled(1 / self.scale)
table_cells.append(tc)
# Retrieving cols/rows, after post processing:

9
examples/convert.py
View File

@ -53,7 +53,14 @@ def main():
artifacts_path = DocumentConverter.download_models_hf()
doc_converter = DocumentConverter(artifacts_path=artifacts_path)
pipeline_options = PipelineOptions(do_table_structure=True)
# use text cells predicted from table structure model, instead of matching with pdf cells
pipeline_options.table_structure_options.do_cell_matching = False
doc_converter = DocumentConverter(
artifacts_path=artifacts_path,
pipeline_options=pipeline_options
)
input = DocumentConversionInput.from_paths(input_doc_paths)