Datasets:

sceiba
/

Sceiba-Scientific-Document-Layout-Dataset

text string
19 0.49819300 0.09150327 0.82045367 0.09150327
11 0.17508651 0.16209150 0.18777393 0.02352941
10 0.49819300 0.26143791 0.82383699 0.13856209
10 0.49819300 0.41045752 0.83737024 0.11241830
10 0.49903883 0.53071895 0.83229527 0.08627451
10 0.49819300 0.68562092 0.83398693 0.18169935
10 0.50157632 0.85294118 0.83737024 0.11633987
19 0.49223211 0.08182844 0.81503120 0.06884876
1 0.49953525 0.23589165 0.83255876 0.20541761
18 0.48565927 0.36738149 0.80480680 0.02144470
11 0.13729916 0.43623025 0.10808658 0.02595937
1 0.49880494 0.64954853 0.83401939 0.36907449
18 0.46301952 0.86286682 0.75952729 0.02370203
19 0.49396386 0.08039216 0.83229527 0.07973856
16 0.34340638 0.22026144 0.52779700 0.09019608
17 0.78323722 0.19019608 0.27743176 0.12941176
13 0.48465975 0.31045752 0.81030373 0.04836601
13 0.45336409 0.38366013 0.73756248 0.05098039
8 0.41953095 0.44901961 0.68342945 0.02745098
9 0.37385621 0.48692810 0.58193003 0.02222222
11 0.13617839 0.62026144 0.09980777 0.02483660
1 0.50326797 0.77516340 0.83737024 0.25359477
10 0.15244163 0.14463811 0.13243742 0.01427972
10 0.18556639 0.53995198 0.19711616 0.03748427
19 0.49150179 0.08295711 0.81357057 0.07110609
2 0.49296242 0.50395034 0.82525561 0.74153499
19 0.49311803 0.08300654 0.82045367 0.07450980
10 0.50073049 0.19150327 0.82891196 0.11111111
10 0.50073049 0.34575163 0.82891196 0.16078431
3 0.49734717 0.58954248 0.32479815 0.29019608
4 0.50073049 0.76405229 0.37554787 0.03006536
12 0.12433679 0.80326797 0.08289120 0.02222222
10 0.49988466 0.86928105 0.83060361 0.06797386
19 0.49819300 0.09019608 0.82722030 0.09150327
10 0.49903883 0.17189542 0.83567859 0.04313725
10 0.49903883 0.30522876 0.83567859 0.18169935
10 0.49903883 0.44836601 0.83567859 0.06797386
12 0.13787005 0.51241830 0.11672434 0.02091503
10 0.49988466 0.56470588 0.83737024 0.04705882
10 0.49903883 0.65294118 0.83567859 0.09019608
10 0.12264514 0.72679739 0.08289120 0.02352941
7 0.51257209 0.80326797 0.80522876 0.09542484
19 0.49369274 0.08295711 0.81503120 0.07336343
11 0.12634444 0.14334086 0.08325588 0.02031603
1 0.49953525 0.36907449 0.82671624 0.39051919
18 0.49807462 0.59537246 0.82963750 0.02144470
11 0.18111805 0.66196388 0.19280308 0.02821670
10 0.50172620 0.77539503 0.83401939 0.16027088
19 0.50157632 0.08235294 0.83060361 0.07320261
10 0.49988466 0.16535948 0.84413687 0.06666667
11 0.26728181 0.23202614 0.36539792 0.02745098
2 0.50157632 0.57450980 0.83737024 0.62614379
19 0.49565552 0.08039216 0.82214533 0.07189542
10 0.50242215 0.20392157 0.83229527 0.13594771
10 0.49903883 0.33464052 0.83229527 0.08366013
11 0.26728181 0.41045752 0.36539792 0.02091503
2 0.49903883 0.67254902 0.84244521 0.45882353
19 0.49565552 0.08954248 0.81876201 0.08758170
12 0.18692810 0.20130719 0.20807382 0.02614379
10 0.30957324 0.24575163 0.45674740 0.02091503
12 0.21568627 0.32679739 0.26220684 0.02352941
10 0.49903883 0.37973856 0.83229527 0.04313725
17 0.14175720 0.48016883 0.11194397 0.03295320
0 0.50039777 0.51220667 0.82685592 0.01830734
19 0.49734717 0.08300654 0.82552864 0.07450980
12 0.31972318 0.14379085 0.46689735 0.01830065
10 0.50157632 0.24313725 0.83398693 0.13333333
12 0.32082254 0.33898684 0.46469863 0.01980208
10 0.50099074 0.40402212 0.82761068 0.06737054
12 0.21666223 0.46837607 0.25895366 0.02312720
10 0.50099074 0.55535445 0.83281578 0.10859729
12 0.34093396 0.64132730 0.50749712 0.02312720
10 0.50164138 0.70568125 0.82891196 0.06737054
19 0.49311803 0.08366013 0.82045367 0.07581699
12 0.49988466 0.16143791 0.83398693 0.04313725
10 0.15309496 0.21568627 0.14379085 0.02352941
7 0.51510957 0.27973856 0.80353710 0.06797386
10 0.13617839 0.34183007 0.09980777 0.02222222
7 0.51595540 0.41830065 0.80861207 0.09150327
12 0.47028066 0.49346405 0.77139562 0.02222222
10 0.15394079 0.53398693 0.13871588 0.02222222
7 0.51595540 0.60718954 0.80522876 0.09542484
10 0.13617839 0.68366013 0.10657439 0.02614379
7 0.51680123 0.76013072 0.80692042 0.09281046
19 0.49150179 0.08239278 0.81649183 0.06997743
7 0.51487186 0.15406321 0.79896428 0.04627540
12 0.25341920 0.20823928 0.34032665 0.02144470
10 0.49953525 0.31715576 0.83255876 0.15801354
11 0.16212986 0.42832957 0.15482672 0.02370203
10 0.50172620 0.54232506 0.83109813 0.15688488
10 0.49880494 0.67325056 0.83109813 0.06659142
10 0.49880494 0.74943567 0.83109813 0.04514673
19 0.49650135 0.08431373 0.82722030 0.07712418
10 0.50326797 0.19150327 0.83737024 0.10849673
10 0.49819300 0.34444444 0.83737024 0.15816993
11 0.23175702 0.45816993 0.29434833 0.02483660
10 0.49903883 0.53725490 0.83567859 0.09150327
10 0.50157632 0.65816993 0.83398693 0.11372549
10 0.49903883 0.76732026 0.83567859 0.07581699
10 0.49903883 0.86535948 0.83567859 0.09411765

End of preview. Expand in Data Studio

Sceiba Scientific Document Layout Dataset

Dataset Description

This dataset contains annotated scientific-document pages prepared for Document Layout Analysis (DLA) within the Sceiba project.

The dataset provides page images and region-level annotations for 20 semantic document classes. The classes are aligned with the Document Components Ontology (DoCO) and complementary csd:* classes used by Sceiba.

The original scientific PDF files belong to the journals Cooperativismo y Desarrollo (COODES), with ISSN 2310-340X; Mendive. Revista de Educación, with ISSN 1815-7696; and The Cuban Journal of Forest Sciences (CFORES), with ISSN 2310-346. They remain available through the respective journal publishers. This repository does not redistribute the original documents.

The current dataset release contains:

90 source documents
1,518 document pages
11,812 annotated regions
20 semantic classes, the 20 classes are represented in the three partitions
fixed document-level train, eval, and test partitions

Dataset Splits

The partitions are defined at document level.

Split	Documents	Pages	Annotated regions
train	63	1,062	8,244
eval	14	228	1,757
test	13	228	1,811
Total	90	1,518	11,812

Classes

class_id	class_name	global	train	eval	test	represented_all_splits	documents_global
0	`csd:License`	90	63	14	13	True	90
1	`doco:Abstract`	398	286	55	57	True	88
2	`doco:BibliographicReferenceList`	333	239	48	46	True	90
3	`doco:Figure`	195	146	24	25	True	63
4	`doco:FigureLabel`	195	146	24	25	True	63
5	`csd:PageFooter`	579	402	85	92	True	32
6	`doco:Formula`	72	61	6	5	True	23
7	`doco:List`	407	242	74	91	True	63
8	`doco:ListOfAuthors`	90	63	14	13	True	90
9	`doco:ListOfOrganizations`	90	63	14	13	True	90
10	`doco:Paragraph`	3,876	2,726	576	574	True	90
11	`csd:MainSectionTitle`	643	447	96	100	True	90
12	`csd:SubsectionTitle`	755	506	102	147	True	88
13	`doco:Subtitle`	180	126	28	26	True	90
14	`doco:Table`	258	173	49	36	True	60
15	`doco:TableLabel`	193	129	37	27	True	58
16	`doco:Title`	90	63	14	13	True	90
17	`csd:image`	1,586	1,112	233	241	True	90
18	`csd:keywords`	264	189	36	39	True	88
19	`csd:PageHeader`	1,518	1,062	228	228	True	90

global, train, eval, and test report the number of annotated regions. documents_global reports the number of source documents in which the class is present.

Semantic Taxonomy

The dataset uses semantic document-region labels based primarily on DoCO (Document Components Ontology).

Additional csd:* classes are used for document components required by the Sceiba processing workflow that are not represented directly by the selected DoCO classes.

The semantic identifiers are preserved in the annotations to facilitate their use in downstream structured-document and knowledge-representation workflows.

License

This dataset is distributed under Creative Commons Attribution-NonCommercial (CC BY-NC).

Users must preserve the required attribution and respect the non-commercial use restriction.

Downloads last month: 16