NeoAnd/Docling
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat: add coverage_threshold to skip OCR for small images (#161)

Browse Source 
* feat: add coverage_threshold to skip OCR for small images

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* filter individual boxes

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

* rename option

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>

---------

Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>
This commit is contained in:
Michele Dolfi 2024-10-18 13:58:23 +02:00 committed by GitHub
parent f799e777c1
commit b346faf622
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 32 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
docling/datamodel/pipeline_options.py
View File

@ -22,6 +22,9 @@ class TableStructureOptions(BaseModel):
class OcrOptions(BaseModel):
kind: str
bitmap_area_threshold: float = (
0.05 # percentage of the area for a bitmap to processed with OCR
)
class EasyOcrOptions(OcrOptions):

10
docling/models/base_ocr_model.py
View File

@ -69,7 +69,7 @@ class BaseOcrModel:
coverage, ocr_rects = find_ocr_rects(page.size, bitmap_rects)
# return full-page rectangle if sufficiently covered with bitmaps
if coverage > BITMAP_COVERAGE_TRESHOLD:
if coverage > max(BITMAP_COVERAGE_TRESHOLD, self.options.bitmap_area_threshold):
return [
BoundingBox(
l=0,
@ -81,6 +81,14 @@ class BaseOcrModel:
]
# return individual rectangles if the bitmap coverage is smaller
else: # coverage <= BITMAP_COVERAGE_TRESHOLD:
# skip OCR if the bitmap area on the page is smaller than the options threshold
ocr_rects = [
rect
for rect in ocr_rects
if rect.area() / (page.size.width * page.size.height)
> self.options.bitmap_area_threshold
]
return ocr_rects
# Filters OCR cells by dropping any OCR cell that intersects with an existing programmatic cell.

20
tests/test_options.py
View File

@ -42,3 +42,23 @@ def test_e2e_conversions(test_doc_path):
doc_result: ConversionResult = converter.convert(test_doc_path)
assert doc_result.status == ConversionStatus.SUCCESS
def test_ocr_coverage_threshold(test_doc_path):
pipeline_options = PdfPipelineOptions()
pipeline_options.do_ocr = True
pipeline_options.ocr_options.bitmap_area_threshold = 1.1
converter = DocumentConverter(
format_options={
InputFormat.PDF: PdfFormatOption(
pipeline_options=pipeline_options,
)
}
)
test_doc_path = Path("./tests/data_scanned/ocr_test.pdf")
doc_result: ConversionResult = converter.convert(test_doc_path)
# this should have generated no results, since we set a very high threshold
assert len(doc_result.document.texts) == 0