mirrors/docling
1
0
Fork 0
mirror of https://github.com/DS4SD/docling.git synced 2025-08-02 07:22:14 +00:00
Code Issues Packages Projects Releases Wiki Activity

test: image input as stream

Browse Source 
Signed-off-by: Michele Dolfi <dol@zurich.ibm.com>
This commit is contained in:
Michele Dolfi 2025-01-20 11:16:35 +01:00
parent 768608351d
commit 12e3419149
5 changed files with 93 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
tests/data/groundtruth/docling_v2/2305.03393v1-pg9-img.doctags.txt Normal file
View File

@ -0,0 +1,19 @@
<document>
<text><location><page_1><loc_22><loc_81><loc_79><loc_85></location>order to compute the TED score. Inference timing results for all experiments were obtained from the same machine On single core with AMD EPYC 7763 CPU @2.45 GHz.</text>
<section_header_level_1><location><page_1><loc_22><loc_77><loc_52><loc_79></location>5.1 Hyper Parameter Optimization</section_header_level_1>
<text><location><page_1><loc_22><loc_68><loc_79><loc_77></location>We have chosen the PubIabNet data set to perform HPO, since it includes à highly diverse set of tables. Also we report TED scores separately for simple and complex tables (tables with cell spans) Results are presented in Table. It is evident that with OTSL, our model achieves the same TED score and slightly 2r speed up in the inference runtime over HIML.</text>
<table>
<location><page_1><loc_23><loc_41><loc_78><loc_57></location>
<caption>Table 1 HPO performed in OTSL HTML representation OIl the samle transformer-based TableFormer 19 architecture; trained only on PubTabNet [22] Effects of reducing the # of layers in encoder and decoder stages of the model show that smaller models trained OIl OTSL perform better, especially in recognizing complex table structures , and maintain a much higher mnAP score than the HTML counterpart_ and</caption>
<row_0><col_0><body></col_0><col_1><body></col_1><col_2><col_header>Language</col_2><col_3><col_header>TEDs</col_3><col_4><col_header>TEDs</col_4><col_5><col_header>TEDs</col_5><col_6><col_header>mAP</col_6><col_7><col_header>Inference</col_7></row_0>
<row_1><col_0><col_header>enc-layers</col_0><col_1><col_header>dec-layers</col_1><col_2><body></col_2><col_3><col_header>simple</col_3><col_4><col_header>complex</col_4><col_5><col_header>all</col_5><col_6><col_header>(0.75)</col_6><col_7><col_header>time (secs)</col_7></row_1>
<row_2><col_0><body>6</col_0><col_1><body>6</col_1><col_2><body>OTSL HTML</col_2><col_3><body>0.965 0.969</col_3><col_4><body>0.934 0.927</col_4><col_5><body>0.955 0.955</col_5><col_6><body>0.88 0.857</col_6><col_7><body>2.73 5.39</col_7></row_2>
<row_3><col_0><body></col_0><col_1><body></col_1><col_2><body>OTSL HTML</col_2><col_3><body>0.938 0.952</col_3><col_4><body>0.904 0.909</col_4><col_5><body>0.927</col_5><col_6><body>0.853</col_6><col_7><body>1.97</col_7></row_3>
<row_4><col_0><body>2</col_0><col_1><body></col_1><col_2><body>OTSL</col_2><col_3><body>0.923</col_3><col_4><body>0.897 0.901</col_4><col_5><body>0.938 0.915</col_5><col_6><body>0.843</col_6><col_7><body>3.77</col_7></row_4>
<row_5><col_0><body></col_0><col_1><body></col_1><col_2><body>HTML</col_2><col_3><body>0.945</col_3><col_4><body></col_4><col_5><body>0.931</col_5><col_6><body>0.859 0.834</col_6><col_7><body>1.91 3.81</col_7></row_5>
<row_6><col_0><body></col_0><col_1><body>2</col_1><col_2><body>OTSL HTML</col_2><col_3><body>0.952 0.944</col_3><col_4><body>0.92 0.903</col_4><col_5><body>0.942 0.931</col_5><col_6><body>0.857 0.824</col_6><col_7><body>1.22 2</col_7></row_6>
</table>
<section_header_level_1><location><page_1><loc_22><loc_35><loc_43><loc_36></location>5.2 Quantitative Results</section_header_level_1>
<text><location><page_1><loc_22><loc_22><loc_79><loc_34></location>We picked the model parameter configuration that produced the best prediction trained and evaluated it on three publicly available data sets: Pub TabNet (395k samples) , FinlabNet (1l3k samples) and PubTables-IM (about IM samples) Performance results are presented in Table.@] It is clearly evident that the model trained on OTSL outperforms HTML across the board; keeping high TEDs and mAP scores evell OIl difficult financial tables (FinTabNet) that contain sparse and tables. large</text>
<text><location><page_1><loc_22><loc_16><loc_79><loc_22></location>advantage OvCI HTML when applied on & bigger data set like PubTables-IM and achieves significantly improved scores. Finally; OTSL achieves faster inference due to fewer decoding steps which is a result of the reduced sequence representation:</text>
</document>

1
tests/data/groundtruth/docling_v2/2305.03393v1-pg9-img.json Normal file
View File

File diff suppressed because one or more lines are too long

22
tests/data/groundtruth/docling_v2/2305.03393v1-pg9-img.md Normal file
View File

@ -0,0 +1,22 @@
order to compute the TED score. Inference timing results for all experiments were obtained from the same machine On single core with AMD EPYC 7763 CPU @2.45 GHz.
## 5.1 Hyper Parameter Optimization
We have chosen the PubIabNet data set to perform HPO, since it includes à highly diverse set of tables. Also we report TED scores separately for simple and complex tables (tables with cell spans) Results are presented in Table. It is evident that with OTSL, our model achieves the same TED score and slightly 2r speed up in the inference runtime over HIML.
Table 1 HPO performed in OTSL HTML representation OIl the samle transformer-based TableFormer 19 architecture; trained only on PubTabNet [22] Effects of reducing the # of layers in encoder and decoder stages of the model show that smaller models trained OIl OTSL perform better, especially in recognizing complex table structures , and maintain a much higher mnAP score than the HTML counterpart\_ and
| | | Language | TEDs | TEDs | TEDs | mAP | Inference |
|------------|------------|------------|-------------|-------------|-------------|-------------|-------------|
| enc-layers | dec-layers | | simple | complex | all | (0.75) | time (secs) |
| 6 | 6 | OTSL HTML | 0.965 0.969 | 0.934 0.927 | 0.955 0.955 | 0.88 0.857 | 2.73 5.39 |
| | | OTSL HTML | 0.938 0.952 | 0.904 0.909 | 0.927 | 0.853 | 1.97 |
| 2 | | OTSL | 0.923 | 0.897 0.901 | 0.938 0.915 | 0.843 | 3.77 |
| | | HTML | 0.945 | | 0.931 | 0.859 0.834 | 1.91 3.81 |
| | 2 | OTSL HTML | 0.952 0.944 | 0.92 0.903 | 0.942 0.931 | 0.857 0.824 | 1.22 2 |
## 5.2 Quantitative Results
We picked the model parameter configuration that produced the best prediction trained and evaluated it on three publicly available data sets: Pub TabNet (395k samples) , FinlabNet (1l3k samples) and PubTables-IM (about IM samples) Performance results are presented in Table.@] It is clearly evident that the model trained on OTSL outperforms HTML across the board; keeping high TEDs and mAP scores evell OIl difficult financial tables (FinTabNet) that contain sparse and tables. large
advantage OvCI HTML when applied on & bigger data set like PubTables-IM and achieves significantly improved scores. Finally; OTSL achieves faster inference due to fewer decoding steps which is a result of the reduced sequence representation:

1
tests/data/groundtruth/docling_v2/2305.03393v1-pg9-img.pages.json Normal file
View File

File diff suppressed because one or more lines are too long

50
tests/test_image_input.py Normal file
View File

@ -0,0 +1,50 @@
from io import BytesIO
from pathlib import Path
import pytest
from docling.datamodel.base_models import DocumentStream
from docling.document_converter import DocumentConverter
from .verify_utils import verify_conversion_result_v2
GENERATE = False
def get_doc_path():
pdf_path = Path("./tests/data/2305.03393v1-pg9-img.png")
return pdf_path
@pytest.fixture
def converter():
converter = DocumentConverter()
return converter
def test_convert_path(converter: DocumentConverter):
doc_path = get_doc_path()
print(f"converting {doc_path}")
doc_result = converter.convert(doc_path)
verify_conversion_result_v2(
input_path=doc_path, doc_result=doc_result, generate=GENERATE
)
def test_convert_stream(converter: DocumentConverter):
doc_path = get_doc_path()
print(f"converting {doc_path}")
buf = BytesIO(doc_path.open("rb").read())
stream = DocumentStream(name=doc_path.name, stream=buf)
doc_result = converter.convert(stream)
verify_conversion_result_v2(
input_path=doc_path, doc_result=doc_result, generate=GENERATE
)