Ocrd

> * :package: 0.1.20
> * extract_pages: fix interference between plotting and JSON extraction in b024b80aa
> * alpha shape: make even more robust yet
> * alpha shape: make more robust
> * :package: 0.1.19
> * repair (`sanitize`): rewrite…
> * repair (`plausibilize`): use true alpha shape instead of convex hull
> * project: reduce default alpha for alphashape
> * project: add level-of-operation=table
> * repair: add option 'simplify', generalize 'ensure_consistent'
> * Shapely 1.8 compatibility
> * :package: 0.1.18
> * extract-lines: fix regression in 8bfeb9f9bf

[opencv-python](https://github.com/skvark/opencv-python) [0eb6c12](https://github.com/skvark/opencv-python/commits/0eb6c12)..[5de8d66](https://github.com/skvark/opencv-python/commits/5de8d66)

> Release: [64](https://github.com/skvark/opencv-python/releases/64)

> * Merge pull request 644 from nsait-linaro/windows-arm64
> * Merge pull request 652 from asenyaev/asen/pinned_dependencies
> * Merge pull request 637 from orbisvicis/master

[sbb_binarization](https://github.com/qurator-spk/sbb_binarization) [0877199](https://github.com/qurator-spk/sbb_binarization/commits/0877199)..[e4c1eb2](https://github.com/qurator-spk/sbb_binarization/commits/e4c1eb2)

> * :package: 0.1.17
> * project: use alpha-shape instead of convex hull
> * forgot actual source file (86e9474edd)
> * update README
> * :package: v0.1.16
> * project: new processor
> * repair/plausibilize: fix scope of result list (avoids deleting multiple times)
> * :package: v0.1.15
> * Merge branch 'excel_gt_writer'
> * plausibilize: both analyse & apply iff enabled
> * :package: v0.1.14
> * repair: fix line split rule
> * Merge branch 'master' of ssh://github.com/OCR-D/ocrd_segment
> * repair: generalize plausibilization…

[ocrd_tesserocr](https://github.com/OCR-D/ocrd_tesserocr) [ac27465](https://github.com/OCR-D/ocrd_tesserocr/commits/ac27465)..[bfe71b7](https://github.com/OCR-D/ocrd_tesserocr/commits/bfe71b7)

> * readme: add Dockerhub badge
> * add CD to Dockerhub
> * update readme
> * join: add param add-filegrp-index
> * process: adapt to Numpy changes
> * add dockerfile
> * compare/evaluate: reverse order of pair (to match actual order in confusion and history)
> * align: clarify that 1st fileGrp is privileged
> * evaluate: add param match_on, allow comparing across index of same fileGrp, too
> * downgrade plotting to optional dependency
> * adopt word splitting from dinglehopper
> * edits, compare/evaluate: micro- (per char) instead of macro-average (per line), fix num-lines vs num-chars
> * join: add param match-on, offer baseline and coords as alternative to id
> adapt to Numpy deprecations

[ocrd_anybaseocr](https://github.com/OCR-D/ocrd_anybaseocr) [5978a1f](https://github.com/OCR-D/ocrd_anybaseocr/commits/5978a1f)..[eb6ea8e](https://github.com/OCR-D/ocrd_anybaseocr/commits/eb6ea8e)

> * Merge pull request 11 from ASVLeipzig/textline-textequiv-join-processor

[core](https://github.com/OCR-D/core) [0713e0e](https://github.com/OCR-D/core/commits/0713e0e)..[bae7ac2](https://github.com/OCR-D/core/commits/bae7ac2)

> * :package: 0.1.13
> * evaluate: fix multi-matching and over/undersegmentation metrics, add pixel scores
> * evaluate: add pixel-wise measures IoU / IoGT / IoDT…
> * evaluate: also get over/undersegmentation and false DT/GT…
> * evaluate: refactor and expose standalone CLI
> * change pycocotools dependency to fork
> * :package: 0.1.12
> * evaluate: cleanup unrestricted maxDets
> * evaluate: fix pycocotools match pair extraction, work around annotation id > 0 assumption


[v2021-11-30](https://github.com/OCR-D/ocrd_all/releases/v2021-11-30)

Changed:

* Re-enable `TESSDATA_PREFIX`, 261

[core](https://github.com/OCR-D/core) [9069a658](https://github.com/OCR-D/core/commits/9069a658)..[c4ca4fab](https://github.com/OCR-D/core/commits/c4ca4fab)

> * :package: 0.1.12
> * wrapper.evaluate: allow aggregating histograms, add parameter
> * scripts.compare: allow aggregating histograms
> * lib.seq2seq / scripts.eval: allow aggregating histograms
> * lib.alignment: allow aggregating histograms
> * wrapper.evaluate: rename reported metrics and measure WER, too; expose gt_level parameter
> * scripts.compare: rename reported metrics and measure WER, too
> * lib.alignment: apply equivalences when aggregating confusion, too
> * lib.alignment: more level 1/2 equivalences/normalizations
> * lib.alignment: remove commented experimental code for other aligners
> * :package: 0.1.11
> * update README
> * lib.alignment: no gaps for confusion counts

[core](https://github.com/OCR-D/core) [d2c6216f](https://github.com/OCR-D/core/commits/d2c6216f)..[271dcdf6](https://github.com/OCR-D/core/commits/271dcdf6)