tboenig / gt_corpus_benchmark

This repo provides a collection of ground truth data. The collection was compiled under different aspects (complexity of the layouts and use of the fonts). The individual data are also characterized by metadata. The metadata is based on the labeling scheme of OCR-D/PrimaLab.

📚 Corpus

This corpus includes Ground Truth (GT) data compiled considering the following feature:

1. Classification into font groups: Gothic/Blackletter, Antiqua and FontMix (Antiqua and Blackletter)
   distinction of the selected print type or combinations
2. Classification into simple and complex
   compelexity of the layout (columns, footnotes,...)

The data are also divided according to the time of creation or production.

🖉 Creation

The data were created according to the OCR-D Ground Truth Guideline (https://ocr-d.de/en/gt-guidelines/trans/).

💻 Repositories

Gothic/Blackletter

simple

• https://github.com/tboenig/16_frak_simple
• https://github.com/tboenig/17_frak_simple
• https://github.com/tboenig/18_frak_simple
• https://github.com/tboenig/19_frak_simple

complex

• https://github.com/tboenig/16_frak_complex
• https://github.com/tboenig/17_frak_complex
• https://github.com/tboenig/18_frak_complex

Antiqua

simple

• https://github.com/tboenig/16_ant_simple
• https://github.com/tboenig/18_ant_simple

complex

• https://github.com/tboenig/16_ant_complex
• https://github.com/tboenig/19_ant_simple

FontMix (Antiqua and Blackletter)

fontmix

• https://github.com/tboenig/17_fontmix_simple
• https://github.com/tboenig/18_fontmix_complex

Analyzed collection

The GT data has been labeled. The labeling is based on an ontology defined by the Pattern Recognition and Image Analysis Research Lab (PRImA-Research-Lab) at the University of Salford. The labeling metadata is created for each available page. The following labeling metadata is available for the different collections.

see: gt-labelling : semantic-labelling OCR ground truth data (https://github.com/OCR-D/gt-labelling)

FontMix (Antiqua and Blackletter)

simple

• activityDomain/computing/visual/analysisRecognition/layoutAnalysis

  In computer vision, document layout analysis is the process of identifying and categorizing the regions of interest in the scanned image of a text document. A reading system requires the segmentation of text zones from non-textual ones and the arrangement in their correct reading order.

  Examples: Page layout analysis (segmentation into regions, classification into text, graphic, table etc.)

  Related: "OCR": Often used as a synonym for layout analysis and text recognition, but strictly only the text recognition component.

• activityDomain/computing/visual/analysisRecognition/ocr

• activityDomain/computing/visual/analysisRecognition/text

  Translation of any kind of depicted symbols to machine readable format

  Examples: OCR Mathematical equation recognition

  Related: Text processing (separate category) Table recognition Map reading

• condition/acquisition/method-flaws/imaging/uneven-illumination

  Uneven illumination leading to brightness or contrast variations

• condition/production-related/document-characteristics/low-contrast

  The contrast bwtween the paper and the page content is very low

• condition/production-related/document-faults/ink-from-facing

  Ink from facing page was transferred to this page

• condition/wear/additions/informative/annotations

  Annotations regarding the content

• content-encoding/structured

  E.g. XML

• content-type/corpus

  Corpus: a collection of written texts, especially the entire works of a particular author or a body of writing on a particular subject.

  Examples: A text corpus, An image database

• contentOfInterest/visual/graphical

  Description coming soon.

• contentOfInterest/visual/graphical/separator

  Description coming soon.

• contentOfInterest/visual/text

  Description coming soon.

• data-attributes/document-related/structural/running-titles

  Titles repeated each page

• data-attributes/document-related/visual/text/drop-caps

  Drap capitals (large capitals at beginning of paragraph)

• data-attributes/document-related/visual/text/font/multi-font/font-sizes

  More than one font size used

• data-attributes/document-related/visual/text/font/multi-font/typefaces

  More than one typeface used

• data-attributes/document-related/visual/text/font/typeface/antiqua

  Antiqua font (more modern)

• data-attributes/document-related/visual/text/font/typeface/blackletter

  Blackletter, gothic, Fraktur

• data-attributes/language/mixed

  More than one language used

• granularity/logical/document-related/paragraph

  Description coming soon.

• granularity/physical/document-related/page

  Description coming soon.

• granularity/physical/document-related/region

  Region, zone, block

• granularity/physical/document-related/text-line

  Description coming soon.

• granularity/physical/document-related/word

  Word or partial word, if separated by line break, for example

• platform/platform-independent

  Description coming soon.

complex

• activityDomain/computing/visual/analysisRecognition/layoutAnalysis

  In computer vision, document layout analysis is the process of identifying and categorizing the regions of interest in the scanned image of a text document. A reading system requires the segmentation of text zones from non-textual ones and the arrangement in their correct reading order.

  Examples: Page layout analysis (segmentation into regions, classification into text, graphic, table etc.)

  Related: "OCR": Often used as a synonym for layout analysis and text recognition, but strictly only the text recognition component.

• activityDomain/computing/visual/analysisRecognition/ocr

• activityDomain/computing/visual/analysisRecognition/text

  Translation of any kind of depicted symbols to machine readable format

  Examples: OCR Mathematical equation recognition

  Related: Text processing (separate category) Table recognition Map reading

• condition/acquisition/content-or-background/included-objects/preceeding-or-proceeding

  Part of preceeding or succeeding object included (e.g. other page)

• condition/acquisition/geometric/page-curl

  Visible page curl (e.g. book scanning)

• condition/acquisition/geometric/perspective-distortions

  Perspective distortions (e.g. due to camera-based acquisition)

• condition/acquisition/method-flaws/imaging/uneven-illumination

  Uneven illumination leading to brightness or contrast variations

• condition/production-related/document-characteristics/low-contrast

  The contrast bwtween the paper and the page content is very low

• condition/production-related/document-faults/ink-from-facing

  Ink from facing page was transferred to this page

• content-encoding/structured

  E.g. XML

• content-type/corpus

  Corpus: a collection of written texts, especially the entire works of a particular author or a body of writing on a particular subject.

  Examples: A text corpus, An image database

• contentOfInterest/visual/graphical/separator

  Description coming soon.

• contentOfInterest/visual/text

  Description coming soon.

• data-attributes/document-related/structural/footnote-continued

• data-attributes/document-related/structural/footnotes

  Footnotes at bottom of page

• data-attributes/document-related/structural/running-titles

  Titles repeated each page

• data-attributes/document-related/visual/text/drop-caps

  Drap capitals (large capitals at beginning of paragraph)

• data-attributes/document-related/visual/text/font/multi-font/font-sizes

  More than one font size used

• data-attributes/document-related/visual/text/font/multi-font/typefaces

  More than one typeface used

• data-attributes/document-related/visual/text/font/typeface/antiqua

  Antiqua font (more modern)

• data-attributes/document-related/visual/text/font/typeface/blackletter

  Blackletter, gothic, Fraktur

• data-attributes/language/mixed

  More than one language used

• granularity/logical/document-related/paragraph

  Description coming soon.

• granularity/physical/document-related/page

  Description coming soon.

• granularity/physical/document-related/region

  Region, zone, block

• granularity/physical/document-related/text-line

  Description coming soon.

• granularity/physical/document-related/word

  Word or partial word, if separated by line break, for example

• platform/platform-independent

  Description coming soon.

Gothic/Blackletter

simple

• activityDomain/computing/visual/analysisRecognition/layoutAnalysis

  In computer vision, document layout analysis is the process of identifying and categorizing the regions of interest in the scanned image of a text document. A reading system requires the segmentation of text zones from non-textual ones and the arrangement in their correct reading order.

  Examples: Page layout analysis (segmentation into regions, classification into text, graphic, table etc.)

  Related: "OCR": Often used as a synonym for layout analysis and text recognition, but strictly only the text recognition component.

• activityDomain/computing/visual/analysisRecognition/ocr

• activityDomain/computing/visual/analysisRecognition/text

  Translation of any kind of depicted symbols to machine readable format

  Examples: OCR Mathematical equation recognition

  Related: Text processing (separate category) Table recognition Map reading

• condition/acquisition/geometric/page-curl

  Visible page curl (e.g. book scanning)

• condition/acquisition/geometric/perspective-distortions

  Perspective distortions (e.g. due to camera-based acquisition)

• condition/ageing/warping

  Arbitrary warping (e.g. due to moisture)

• condition/production-related/document-faults/ink-from-facing

  Ink from facing page was transferred to this page

• condition/wear/additions/informative/annotations

  Annotations regarding the content

• condition/wear/medium-damage/stains

  Noticeable stains on medium

• content-encoding/structured

  E.g. XML

• content-type/corpus

  Corpus: a collection of written texts, especially the entire works of a particular author or a body of writing on a particular subject.

  Examples: A text corpus, An image database

• contentOfInterest/visual/graphical

  Description coming soon.

• contentOfInterest/visual/graphical/separator

  Description coming soon.

• contentOfInterest/visual/text

  Description coming soon.

• data-attributes/document-related/structural/running-titles

  Titles repeated each page

• data-attributes/document-related/visual/text/drop-caps

  Drap capitals (large capitals at beginning of paragraph)

• data-attributes/document-related/visual/text/font/multi-font/font-sizes

  More than one font size used

• data-attributes/document-related/visual/text/font/multi-font/typefaces

  More than one typeface used

• data-attributes/document-related/visual/text/font/typeface/antiqua

  Antiqua font (more modern)

• data-attributes/document-related/visual/text/font/typeface/blackletter

  Blackletter, gothic, Fraktur

• granularity/logical/document-related/paragraph

  Description coming soon.

• granularity/physical/document-related/page

  Description coming soon.

• granularity/physical/document-related/region

  Region, zone, block

• granularity/physical/document-related/text-line

  Description coming soon.

• granularity/physical/document-related/word

  Word or partial word, if separated by line break, for example

• platform/platform-independent

  Description coming soon.

complex

• activityDomain/computing/visual/analysisRecognition/layoutAnalysis

  In computer vision, document layout analysis is the process of identifying and categorizing the regions of interest in the scanned image of a text document. A reading system requires the segmentation of text zones from non-textual ones and the arrangement in their correct reading order.

  Examples: Page layout analysis (segmentation into regions, classification into text, graphic, table etc.)

  Related: "OCR": Often used as a synonym for layout analysis and text recognition, but strictly only the text recognition component.

• activityDomain/computing/visual/analysisRecognition/ocr

• activityDomain/computing/visual/analysisRecognition/text

  Translation of any kind of depicted symbols to machine readable format

  Examples: OCR Mathematical equation recognition

  Related: Text processing (separate category) Table recognition Map reading

• condition/acquisition/content-or-background/included-objects/preceeding-or-proceeding

  Part of preceeding or succeeding object included (e.g. other page)

• condition/acquisition/geometric/page-curl

  Visible page curl (e.g. book scanning)

• condition/acquisition/geometric/perspective-distortions

  Perspective distortions (e.g. due to camera-based acquisition)

• condition/acquisition/method-flaws/imaging/uneven-illumination

  Uneven illumination leading to brightness or contrast variations

• condition/ageing/warping

  Arbitrary warping (e.g. due to moisture)

• condition/production-related/document-characteristics/low-contrast

  The contrast bwtween the paper and the page content is very low

• condition/production-related/document-faults/ink-from-facing

  Ink from facing page was transferred to this page

• condition/wear/additions/informative/annotations

  Annotations regarding the content

• condition/wear/additions/informative/stamps

  The medium was stamped

• condition/wear/medium-damage/stains

  Noticeable stains on medium

• content-encoding/structured

  E.g. XML

• content-type/corpus

  Corpus: a collection of written texts, especially the entire works of a particular author or a body of writing on a particular subject.

  Examples: A text corpus, An image database

• contentOfInterest/visual/composite/music

  Description coming soon.

• contentOfInterest/visual/graphical

  Description coming soon.

• contentOfInterest/visual/graphical/separator

  Description coming soon.

• contentOfInterest/visual/text

  Description coming soon.

• data-attributes/document-related/structural/footnotes

  Footnotes at bottom of page

• data-attributes/document-related/structural/running-titles

  Titles repeated each page

• data-attributes/document-related/visual/decorations

  Decorations of some kind

• data-attributes/document-related/visual/illustrations

  Illustrations in content

• data-attributes/document-related/visual/illustrations/multi-colour

  Multi-colour illustrations in content

• data-attributes/document-related/visual/text/drop-caps

  Drap capitals (large capitals at beginning of paragraph)

• data-attributes/document-related/visual/text/font/multi-font/font-sizes

  More than one font size used

• data-attributes/document-related/visual/text/font/multi-font/typefaces

  More than one typeface used

• data-attributes/document-related/visual/text/font/typeface/antiqua

  Antiqua font (more modern)

• data-attributes/document-related/visual/text/font/typeface/blackletter

  Blackletter, gothic, Fraktur

• data-attributes/language/mixed

  More than one language used

• granularity/logical/document-related/paragraph

  Description coming soon.

• granularity/physical/document-related/page

  Description coming soon.

• granularity/physical/document-related/region

  Region, zone, block

• granularity/physical/document-related/text-line

  Description coming soon.

• granularity/physical/document-related/word

  Word or partial word, if separated by line break, for example

• platform/platform-independent

  Description coming soon.

Antiqua

simple

• activityDomain/computing/visual/analysisRecognition/layoutAnalysis

  In computer vision, document layout analysis is the process of identifying and categorizing the regions of interest in the scanned image of a text document. A reading system requires the segmentation of text zones from non-textual ones and the arrangement in their correct reading order.

  Examples: Page layout analysis (segmentation into regions, classification into text, graphic, table etc.)

  Related: "OCR": Often used as a synonym for layout analysis and text recognition, but strictly only the text recognition component.

• activityDomain/computing/visual/analysisRecognition/ocr

• activityDomain/computing/visual/analysisRecognition/text

  Translation of any kind of depicted symbols to machine readable format

  Examples: OCR Mathematical equation recognition

  Related: Text processing (separate category) Table recognition Map reading

• condition/production-related/document-faults/ink-from-facing

  Ink from facing page was transferred to this page

• condition/wear/medium-damage/stains

  Noticeable stains on medium

• content-encoding/structured

  E.g. XML

• content-type/corpus

  Corpus: a collection of written texts, especially the entire works of a particular author or a body of writing on a particular subject.

  Examples: A text corpus, An image database

• contentOfInterest/visual/graphical/separator

  Description coming soon.

• contentOfInterest/visual/text

  Description coming soon.

• data-attributes/document-related/visual/text/drop-caps

  Drap capitals (large capitals at beginning of paragraph)

• data-attributes/document-related/visual/text/font/multi-font/font-sizes

  More than one font size used

• data-attributes/document-related/visual/text/font/typeface/antiqua

  Antiqua font (more modern)

• data-attributes/document-related/visual/text/font/typeface/blackletter

  Blackletter, gothic, Fraktur

• granularity/logical/document-related/paragraph

  Description coming soon.

• granularity/physical/document-related/page

  Description coming soon.

• granularity/physical/document-related/region

  Region, zone, block

• granularity/physical/document-related/text-line

  Description coming soon.

• granularity/physical/document-related/word

  Word or partial word, if separated by line break, for example

• platform/platform-independent

  Description coming soon.

complex

• activityDomain/computing/visual/analysisRecognition/layoutAnalysis

  In computer vision, document layout analysis is the process of identifying and categorizing the regions of interest in the scanned image of a text document. A reading system requires the segmentation of text zones from non-textual ones and the arrangement in their correct reading order.

  Examples: Page layout analysis (segmentation into regions, classification into text, graphic, table etc.)

  Related: "OCR": Often used as a synonym for layout analysis and text recognition, but strictly only the text recognition component.

• activityDomain/computing/visual/analysisRecognition/ocr

• activityDomain/computing/visual/analysisRecognition/text

  Translation of any kind of depicted symbols to machine readable format

  Examples: OCR Mathematical equation recognition

  Related: Text processing (separate category) Table recognition Map reading

• condition/production-related/document-faults/ink-from-facing

  Ink from facing page was transferred to this page

• condition/wear/additions/informative/annotations

  Annotations regarding the content

• condition/wear/medium-damage/stains

  Noticeable stains on medium

• content-encoding/structured

  E.g. XML

• content-type/corpus

  Corpus: a collection of written texts, especially the entire works of a particular author or a body of writing on a particular subject.

  Examples: A text corpus, An image database

• contentOfInterest/visual/text

  Description coming soon.

• data-attributes/document-related/structural/footnote-continued

• data-attributes/document-related/structural/footnotes

  Footnotes at bottom of page

• data-attributes/document-related/structural/running-titles

  Titles repeated each page

• data-attributes/document-related/visual/text/drop-caps

  Drap capitals (large capitals at beginning of paragraph)

• data-attributes/document-related/visual/text/font/multi-font/font-sizes

  More than one font size used

• data-attributes/document-related/visual/text/font/multi-font/typefaces

  More than one typeface used

• data-attributes/document-related/visual/text/font/typeface/antiqua

  Antiqua font (more modern)

• data-attributes/document-related/visual/text/font/typeface/blackletter

  Blackletter, gothic, Fraktur

• data-attributes/language/mixed

  More than one language used

• granularity/logical/document-related/paragraph

  Description coming soon.

• granularity/physical/document-related/page

  Description coming soon.

• granularity/physical/document-related/region

  Region, zone, block

• granularity/physical/document-related/text-line

  Description coming soon.

• granularity/physical/document-related/word

  Word or partial word, if separated by line break, for example

• platform/platform-independent

  Description coming soon.