Skip to content

Word alignment in PAGE XML #75

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
vlachvojta wants to merge 7 commits into
base: develop
Choose a base branch
from
Open

Word alignment in PAGE XML #75

vlachvojta wants to merge 7 commits into from

Conversation

@vlachvojta
Copy link
Collaborator

@vlachvojta vlachvojta commented Mar 22, 2025 •
edited
Loading

Old behavior

  • words get aligned using CTC logits into word bounding boxes only if trying to export to ALTO XML
  • the alignment is not stored elsewhere

New behavior

  • word alignment = storing words with bounding boxes, transcription (, ...) in core/layout/word elements (list in textline elements)
  • user can enable the word alignment feature with OCR.ALIGN_WORDS = yes in .ini config
    • (it is disabled by default, as word alignment takes some time)
  • aligned words are stored in PAGE XML Word elements with id, polygon, transcription, and transcription confidence
  • TextLine in core/layout stores aligned words in the new class Word + information on how the words got aligned (from_logits or mean_width for error fallback)
  • word polygons are rendered to the image using a thin black line

@vlachvojta
add optinal baseline rendering
8127482
@vlachvojta
Add ALIGN_WORDS OCR config argument, which creates words using logi…
4521369 
…ts alignment.

New PAGE XML has both textline transcription and individual words with bounding boxes.

- fallback for impossible alignment is still average width for every word
@vlachvojta
Make option to force new word alignment from scratch
334ef0c
@vlachvojta
update docs for render_to_image
5cafdcb
@vlachvojta
Add type of word alignment to every line (save+load to TextLine custo…
0d68142 
…m tag)
@vlachvojta vlachvojta self-assigned this Mar 22, 2025
@vlachvojta vlachvojta added the enhancement New feature or request label Mar 22, 2025
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

@vlachvojta vlachvojta

Labels

enhancement New feature or request

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@vlachvojta