mirrors/docling
1
0
Fork 0
mirror of https://github.com/DS4SD/docling.git synced 2025-12-08 20:58:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: enable configurable runtime for rapidocr and handle new result better;

Browse Source 
Signed-off-by: Georg Heiler <georg.kf.heiler@gmail.com>
This commit is contained in:
Georg Heiler
2025-08-20 08:36:43 +02:00
parent 63fb8ff599
commit 2892cb5abe
2 changed files with 17 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docling/datamodel/pipeline_options.py
View File

@@ -99,6 +99,8 @@ class RapidOcrOptions(OcrOptions):
# For more details on the following options visit # For more details on the following options visit
# https://rapidai.github.io/RapidOCRDocs/install_usage/api/RapidOCR/ # https://rapidai.github.io/RapidOCRDocs/install_usage/api/RapidOCR/
# https://rapidai.github.io/RapidOCRDocs/main/install_usage/rapidocr/usage/#__tabbed_3_4
backend: Literal["onnxruntime", "openvino", "paddle", "torch"] = "onnxruntime"
text_score: float = 0.5 # same default as rapidocr text_score: float = 0.5 # same default as rapidocr
use_det: Optional[bool] = None # same default as rapidocr use_det: Optional[bool] = None # same default as rapidocr

17
docling/models/rapid_ocr_model.py
View File

@@ -42,7 +42,7 @@ class RapidOcrModel(BaseOcrModel):
if self.enabled: if self.enabled:
try: try:
from rapidocr import RapidOCR # type: ignore from rapidocr import RapidOCR, EngineType # type: ignore
except ImportError: except ImportError:
raise ImportError( raise ImportError(
"RapidOCR is not installed. Please install it via `pip install rapidocr onnxruntime` to use this OCR engine. " "RapidOCR is not installed. Please install it via `pip install rapidocr onnxruntime` to use this OCR engine. "
@@ -54,6 +54,13 @@ class RapidOcrModel(BaseOcrModel):
use_cuda = str(AcceleratorDevice.CUDA.value).lower() in device use_cuda = str(AcceleratorDevice.CUDA.value).lower() in device
use_dml = accelerator_options.device == AcceleratorDevice.AUTO use_dml = accelerator_options.device == AcceleratorDevice.AUTO
intra_op_num_threads = accelerator_options.num_threads intra_op_num_threads = accelerator_options.num_threads
_ALIASES = {
"onnxruntime": EngineType.ONNXRUNTIME,
"openvino": EngineType.OPENVINO,
"paddle": EngineType.PADDLE,
"torch": EngineType.TORCH,
}
backend_enum = _ALIASES.get(self.options.backend, EngineType.TORCH)
self.reader = RapidOCR( self.reader = RapidOCR(
params={ params={
@@ -76,6 +83,9 @@ class RapidOcrModel(BaseOcrModel):
"Rec.use_cuda": use_cuda, "Rec.use_cuda": use_cuda,
"Rec.use_dml": use_dml, "Rec.use_dml": use_dml,
"Rec.intra_op_num_threads": intra_op_num_threads, "Rec.intra_op_num_threads": intra_op_num_threads,
"Det.engine_type": backend_enum,
"Cls.engine_type": backend_enum,
"Rec.engine_type": backend_enum,
} }
) )
@@ -103,12 +113,15 @@ class RapidOcrModel(BaseOcrModel):
scale=self.scale, cropbox=ocr_rect scale=self.scale, cropbox=ocr_rect
) )
im = numpy.array(high_res_image) im = numpy.array(high_res_image)
result, _ = self.reader( result = self.reader(
im, im,
use_det=self.options.use_det, use_det=self.options.use_det,
use_cls=self.options.use_cls, use_cls=self.options.use_cls,
use_rec=self.options.use_rec, use_rec=self.options.use_rec,
) )
result = list(
zip(result.boxes.tolist(), result.txts, result.scores)
)
del high_res_image del high_res_image
del im del im