Skip to content
Snippets Groups Projects
TesseractOCRPipelineModel.defaults.yml 41.4 KiB
Newer Older
  • Learn to ignore specific revisions
  • # - title: 'Afrikaans'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/afr.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Amharic'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/amh.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Arabic'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ara.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Assamese'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/asm.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Azerbaijani'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/aze.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Azerbaijani - Cyrillic'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/aze_cyrl.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Belarusian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/bel.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Bengali'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ben.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Tibetan'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/bod.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Bosnian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/bos.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Bulgarian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/bul.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Catalan; Valencian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/cat.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Cebuano'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ceb.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Czech'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ces.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Chinese - Simplified'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/chi_sim.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Chinese - Traditional'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/chi_tra.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Cherokee'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/chr.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Welsh'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/cym.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Danish'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/dan.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    - title: 'German'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/deu.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Dzongkha'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/dzo.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Greek, Modern (1453-)'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ell.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    - title: 'English'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/eng.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    - title: 'English, Middle (1100-1500)'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/enm.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Esperanto'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/epo.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Estonian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/est.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Basque'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/eus.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Persian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/fas.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Finnish'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/fin.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'French'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/fra.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    - title: 'German Fraktur'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/frk.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    - title: 'French, Middle (ca. 1400-1600)'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/frm.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Irish'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/gle.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Galician'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/glg.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Greek, Ancient (-1453)'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/grc.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Gujarati'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/guj.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Haitian; Haitian Creole'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/hat.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Hebrew'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/heb.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Hindi'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/hin.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Croatian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/hrv.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Hungarian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/hun.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Inuktitut'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/iku.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Indonesian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ind.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Icelandic'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/isl.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Italian'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ita.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    - title: 'Italian - Old'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ita_old.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Javanese'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/jav.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Japanese'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/jpn.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Kannada'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/kan.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Georgian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/kat.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Georgian - Old'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/kat_old.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Kazakh'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/kaz.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Central Khmer'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/khm.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Kirghiz; Kyrgyz'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/kir.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Korean'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/kor.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Kurdish'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/kur.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Lao'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/lao.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Latin'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/lat.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Latvian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/lav.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Lithuanian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/lit.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Malayalam'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/mal.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Marathi'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/mar.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Macedonian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/mkd.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Maltese'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/mlt.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Malay'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/msa.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Burmese'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/mya.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Nepali'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/nep.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Dutch; Flemish'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/nld.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Norwegian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/nor.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Oriya'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ori.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Panjabi; Punjabi'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/pan.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Polish'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/pol.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Portuguese'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/por.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Pushto; Pashto'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/pus.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Romanian; Moldavian; Moldovan'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/ron.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Russian'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/rus.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Sanskrit'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/san.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Sinhala; Sinhalese'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/sin.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Slovak'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/slk.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Slovenian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/slv.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    - title: 'Spanish; Castilian'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/spa.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    - title: 'Spanish; Castilian - Old'
      description: ''
      url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/spa_old.traineddata'
      publisher: 'tesseract-ocr'
    
      publisher_url: 'https://github.com/tesseract-ocr'
      publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
      publishing_year: 2021
      version: '4.1.0'
      compatible_service_versions:
        - '0.1.0'
    
    # - title: 'Albanian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/sqi.traineddata'
    #   publisher: 'tesseract-ocr'
    
    #   publisher_url: 'https://github.com/tesseract-ocr'
    #   publishing_url: 'https://github.com/tesseract-ocr/tessdata/releases/tag/4.1.0'
    
    #   publishing_year: 2021
    #   version: '4.1.0'
    #   compatible_service_versions:
    #     - '0.1.0'
    
    # - title: 'Serbian'
    #   description: ''
    #   url: 'https://github.com/tesseract-ocr/tessdata/raw/4.1.0/srp.traineddata'
    #   publisher: 'tesseract-ocr'