Commits · 168ede5d41352ae9b6eec6dfb9422555ad4758ea · nlp / ahisto-modules / Optical Character Recognition Experiments · GitLab

Nov 13, 2021
- Evaluate language identification with Google Vision AI without limiting languages · 168ede5d
  Vít Starý Novotný authored Nov 13, 2021
  
  168ede5d
Nov 12, 2021
- Add language detection results with both old and new algorithms · 005c77b2
  Vít Starý Novotný authored Nov 12, 2021
  
  005c77b2
Nov 11, 2021
- Evaluate language detection on human annotations · 739afa39
  Vít Starý Novotný authored Nov 11, 2021
  
  739afa39
- Add 122 annotated pages for language detection · c051b076
  Vít Starý Novotný authored Nov 11, 2021
```
Cf https://gitlab.fi.muni.cz/nlp/ahisto-language-detection/-/jobs/134219
```
  c051b076
- Add support for annotated HOCR files · 3df50968
  Vít Starý Novotný authored Nov 11, 2021
  
  3df50968
Sep 04, 2021
- Fix layout detection algorithm · a1513a8e
  Vít Starý Novotný authored Sep 03, 2021
  
  a1513a8e
Aug 26, 2021
- Add... · 02d9ced7
  Vít Starý Novotný authored Aug 26, 2021
```
Add /var/tmp/tesseract/output-downscaled-waifu2x-high-noise-ocr4-twopass-google-lowres.00.accuracy-results-human
```
  02d9ced7
- Assume common.is_multicolumn() may get an empty file · 7cba118d
  Vít Starý Novotný authored Aug 26, 2021
  
  7cba118d
Aug 25, 2021
- Use both old and new language detection algorithms for various tasks · bc4f0d66
  Vít Starý Novotný authored Aug 24, 2021
  
  bc4f0d66
- Recompute results using old language detection · c91ba8c8
  Vít Starý Novotný authored Aug 25, 2021
  
  c91ba8c8
Aug 18, 2021
- Add computational geometry algorithm that detects the number of columns (cont.) · c7a67412
  Vít Starý Novotný authored Aug 18, 2021
  
  c7a67412
Aug 17, 2021
- Make pytype happy · 0006a3d0
  Vít Starý Novotný authored Aug 17, 2021
  
  0006a3d0
- Add computational geometry algorithm that detects the number of columns · d2c9a68e
  Vít Starý Novotný authored Aug 17, 2021
  
  d2c9a68e
Jul 27, 2021
- Evaluate the current best technique · c490abf5
  Vít Starý Novotný authored Jul 27, 2021
  
  c490abf5
- Add and evaluate layout detection (cont.) · 4bfa3667
  Vít Starý Novotný authored Jul 27, 2021
  
  4bfa3667
- Add and evaluate layout detection · e6630c66
  Vít Starý Novotný authored Jul 27, 2021
  
  e6630c66
Jul 14, 2021
- Evaluate the current best technique · e8692aaa
  Vít Starý Novotný authored Jul 14, 2021
  
  e8692aaa
Jun 26, 2021
- Merge remote-tracking branch 'origin/master' into manylang · 5efa5491
  Vít Starý Novotný authored Jun 26, 2021
  
  5efa5491
- Fix an encoding error · 77a950cc
  Vít Starý Novotný authored Jun 26, 2021
  
  77a950cc
- Merge remote-tracking branch 'origin/master' into manylang · 03cf20fb
  Vít Starý Novotný authored Jun 26, 2021
  
  03cf20fb
- Better guard against empty XML files in read_page_languages · bac869ec
  Vít Starý Novotný authored Jun 26, 2021
  
  bac869ec
- Fix recipe for OUTPUT_DIFFICULT_PAGES_DIRNAME · ba83d939
  Vít Starý Novotný authored Jun 26, 2021
  
  ba83d939
- Download tessdata with one line of code · ee67a982
  Vít Starý Novotný authored Jun 26, 2021
  
  ee67a982
- Evaluate nine languages · ef2ff455
  Vít Starý Novotný authored Jun 26, 2021
  
  ef2ff455
Jun 21, 2021
- Evaluate more thresholds for two-pass processing · 4cdd29eb
  Vít Starý Novotný authored Jun 21, 2021 and Vít Starý Novotný committed Jun 21, 2021
  
  4cdd29eb
- Evaluate more thresholds for two-pass processing · d270cb2e
  Vít Starý Novotný authored Jun 21, 2021
  
  d270cb2e
Jun 16, 2021
- Only apply OCR to the linked images, not all · 7dd7905f
  Vít Starý Novotný authored Jun 16, 2021
  
  7dd7905f
- Extend the number of detected languages from 3 to 9 · 8f8f9621
  Vít Starý Novotný authored Jun 16, 2021
  
  8f8f9621
- Guard against empty XML files in read_page_languages · 790419f6
  Vít Starý Novotný authored Jun 16, 2021
  
  790419f6
Jun 11, 2021
- Guard against zero probabilities of languages · d721be26
  Vít Starý Novotný authored Jun 11, 2021
  
  d721be26
- Take word-level language annotations in HOCR output into account (cont.) · c9796cbb
  Vít Starý Novotný authored Jun 11, 2021
  
  c9796cbb
- Take word-level language annotations in HOCR output into account · 63fdca1a
  Vít Starý Novotný authored Jun 11, 2021
  
  63fdca1a
May 27, 2021
- Make scripts.untypeset_difficult_pages preserve line breaks · 40472280
  Vít Starý Novotný authored May 27, 2021
  
  40472280
May 20, 2021
- Support extraction of languages into JSON files · 3545b3a9
  Vít Starý Novotný authored May 20, 2021
  
  3545b3a9
Apr 28, 2021
- Evaluate best system · 5770c594
  Vít Starý Novotný authored Apr 28, 2021
  
  5770c594
Apr 11, 2021
- Make scripts.untypeset_difficult_pages provide richer output · 1f9cd3ff
  Vít Starý Novotný authored Apr 11, 2021
  
  1f9cd3ff
- Evaluate superscaling with SRGAN · 8a400a16
  Vít Starý Novotný authored Apr 11, 2021
  
  8a400a16
Apr 08, 2021
- Reevaluate on final human judgements · 77381d3d
  Vít Starý Novotný authored Apr 08, 2021
  
  77381d3d
Apr 05, 2021
- Use user-defined ordering of languages in two-pass Tesseract 4 · 4516b3d3
  Vít Starý Novotný authored Apr 05, 2021
  
  4516b3d3
Mar 31, 2021
- Evaluate two-pass Tesseract 4 with thresholding · 9cf0a480
  Vít Starý Novotný authored Mar 31, 2021
  
  9cf0a480