ImageEn, unit ievision

TIEVisionLanguages


Declaration

type
  TIEVisionLanguages = record
    Code        : PAnsiChar;  // Generally ISO 639-2 or ISO 639-3 codes
    DisplayName : string;
  end;

const
  IEOCRLanguageList: array[TIEOCRLanguages] of TIEVisionLanguages = (
    (Code: 'afr';            DisplayName: 'Afrikaans'),
    (Code: 'sqi';            DisplayName: 'Albanian'),
    (Code: 'amh';            DisplayName: 'Amharic'),
    (Code: 'grc';            DisplayName: 'Ancient Greek'),
    (Code: 'ara';            DisplayName: 'Arabic'),
    (Code: 'hye';            DisplayName: 'Armenian'),
    (Code: 'asm';            DisplayName: 'Assamese'),
    (Code: 'aze';            DisplayName: 'Azerbaijani'),
    (Code: 'aze_cyrl';       DisplayName: 'Azerbaijani (Cyrillic)'),
    (Code: 'eus';            DisplayName: 'Basque'),
    (Code: 'bel';            DisplayName: 'Belarusian'),
    (Code: 'ben';            DisplayName: 'Bengali'),
    (Code: 'bos';            DisplayName: 'Bosnian'),
    (Code: 'bre';            DisplayName: 'Breton'),
    (Code: 'bul';            DisplayName: 'Bulgarian'),
    (Code: 'mya';            DisplayName: 'Burmese'),
    (Code: 'cat';            DisplayName: 'Catalan'),
    (Code: 'ceb';            DisplayName: 'Cebuano'),
    (Code: 'chr';            DisplayName: 'Cherokee'),
    (Code: 'chi_sim';        DisplayName: 'Chinese (Simplified)'),
    (Code: 'chi_sim_vert';   DisplayName: 'Chinese (Simplified/Vertical)'),
    (Code: 'chi_tra';        DisplayName: 'Chinese (Traditional)'),
    (Code: 'chi_tra_vert';   DisplayName: 'Chinese (Traditional/Vertical)'),
    (Code: 'cos';            DisplayName: 'Corsican'),
    (Code: 'hrv';            DisplayName: 'Croatian'),
    (Code: 'ces';            DisplayName: 'Czech'),
    (Code: 'dan';            DisplayName: 'Danish'),
    (Code: 'dan_frak';       DisplayName: 'Danish (Fraktur)'),
    (Code: 'div';            DisplayName: 'Dhivehi'),
    (Code: 'nld';            DisplayName: 'Dutch'),
    (Code: 'dzo';            DisplayName: 'Dzongkha'),
    (Code: 'eng';            DisplayName: 'English'),
    (Code: 'equ';            DisplayName: 'Equations'),
    (Code: 'epo';            DisplayName: 'Esperanto'),
    (Code: 'est';            DisplayName: 'Estonian'),
    (Code: 'fao';            DisplayName: 'Faroese'),
    (Code: 'fil';            DisplayName: 'Filipino'),
    (Code: 'fin';            DisplayName: 'Finnish'),
    (Code: 'frk';            DisplayName: 'Frankish'),
    (Code: 'fra';            DisplayName: 'French'),
    (Code: 'gla';            DisplayName: 'Gaelic'),
    (Code: 'glg';            DisplayName: 'Galician'),
    (Code: 'kat';            DisplayName: 'Georgian'),
    (Code: 'kat_old';        DisplayName: 'Georgian (Old)'),
    (Code: 'deu';            DisplayName: 'German'),
    (Code: 'deu_frak';       DisplayName: 'German (Fraktur)'),
    (Code: 'ell';            DisplayName: 'Greek'),
    (Code: 'guj';            DisplayName: 'Gujarati'),
    (Code: 'hat';            DisplayName: 'Haitian'),
    (Code: 'heb';            DisplayName: 'Hebrew'),
    (Code: 'hin';            DisplayName: 'Hindi'),
    (Code: 'hun';            DisplayName: 'Hungarian'),
    (Code: 'isl';            DisplayName: 'Icelandic'),
    (Code: 'ind';            DisplayName: 'Indonesian'),
    (Code: 'iku';            DisplayName: 'Inuktitut'),
    (Code: 'gle';            DisplayName: 'Irish'),
    (Code: 'ita';            DisplayName: 'Italian'),
    (Code: 'ita_old';        DisplayName: 'Italian (Old)'),
    (Code: 'jpn';            DisplayName: 'Japanese'),
    (Code: 'jpn_vert';       DisplayName: 'Japanese (Vertical)'),
    (Code: 'jav';            DisplayName: 'Javanese'),
    (Code: 'kan';            DisplayName: 'Kannada'),
    (Code: 'kaz';            DisplayName: 'Kazakh'),
    (Code: 'khm';            DisplayName: 'Khmer'),
    (Code: 'kor';            DisplayName: 'Korean'),
    (Code: 'kor_vert';       DisplayName: 'Korean (Vertical)'),
    (Code: 'kur';            DisplayName: 'Kurdish'),
    (Code: 'kur_ara';        DisplayName: 'Kurdish (Arabic Script)'),
    (Code: 'kir';            DisplayName: 'Kyrgyz'),
    (Code: 'lao';            DisplayName: 'Lao'),
    (Code: 'lat';            DisplayName: 'Latin'),
    (Code: 'lav';            DisplayName: 'Latvian'),
    (Code: 'lit';            DisplayName: 'Lithuanian'),
    (Code: 'ltz';            DisplayName: 'Luxembourgish'),
    (Code: 'mkd';            DisplayName: 'Macedonian'),
    (Code: 'msa';            DisplayName: 'Malay'),
    (Code: 'mal';            DisplayName: 'Malayalam'),
    (Code: 'mlt';            DisplayName: 'Maltese'),
    (Code: 'mri';            DisplayName: 'Maori'),
    (Code: 'mar';            DisplayName: 'Marathi'),
    (Code: 'mcr';            DisplayName: 'MICR'),
    (Code: 'enm';            DisplayName: 'Middle English'),
    (Code: 'frm';            DisplayName: 'Middle French'),
    (Code: 'mon';            DisplayName: 'Mongolian'),
    (Code: 'nep';            DisplayName: 'Nepali'),
    (Code: 'nor';            DisplayName: 'Norwegian'),
    (Code: 'oci';            DisplayName: 'Occitan'),
    (Code: 'ori';            DisplayName: 'Oriya'),
    (Code: 'pus';            DisplayName: 'Pashto'),
    (Code: 'fas';            DisplayName: 'Persian'),
    (Code: 'pol';            DisplayName: 'Polish'),
    (Code: 'por';            DisplayName: 'Portuguese'),
    (Code: 'pan';            DisplayName: 'Punjabi'),
    (Code: 'que';            DisplayName: 'Quechua'),
    (Code: 'ron';            DisplayName: 'Romanian'),
    (Code: 'rus';            DisplayName: 'Russian'),
    (Code: 'san';            DisplayName: 'Sanskrit'),
    (Code: 'srp';            DisplayName: 'Serbian'),
    (Code: 'srp_latn';       DisplayName: 'Serbian (Latin)'),
    (Code: 'snd';            DisplayName: 'Sindhi'),
    (Code: 'sin';            DisplayName: 'Sinhala'),
    (Code: 'slk';            DisplayName: 'Slovakian'),
    (Code: 'slk_frak';       DisplayName: 'Slovakian (Fraktur)'),
    (Code: 'slv';            DisplayName: 'Slovenian'),
    (Code: 'spa';            DisplayName: 'Spanish'),
    (Code: 'spa_old';        DisplayName: 'Spanish (Old)'),
    (Code: 'sun';            DisplayName: 'Sundanese'),
    (Code: 'swa';            DisplayName: 'Swahili'),
    (Code: 'swe';            DisplayName: 'Swedish'),
    (Code: 'syr';            DisplayName: 'Syriac'),
    (Code: 'tgl';            DisplayName: 'Tagalog'),
    (Code: 'tgk';            DisplayName: 'Tajik'),
    (Code: 'tam';            DisplayName: 'Tamil'),
    (Code: 'tat';            DisplayName: 'Tatar'),
    (Code: 'tel';            DisplayName: 'Telugu'),
    (Code: 'tha';            DisplayName: 'Thai'),
    (Code: 'bod';            DisplayName: 'Tibetan'),
    (Code: 'tir';            DisplayName: 'Tigrinya'),
    (Code: 'ton';            DisplayName: 'Tongan'),
    (Code: 'tur';            DisplayName: 'Turkish'),
    (Code: 'ukr';            DisplayName: 'Ukrainian'),
    (Code: 'urd';            DisplayName: 'Urdu'),
    (Code: 'uig';            DisplayName: 'Uyghur'),
    (Code: 'uzb';            DisplayName: 'Uzbek'),
    (Code: 'uzb_cyrl';       DisplayName: 'Uzbek (Cyrillic)'),
    (Code: 'vie';            DisplayName: 'Vietnamese'),
    (Code: 'cym';            DisplayName: 'Welsh'),
    (Code: 'fry';            DisplayName: 'Western Frisian'),
    (Code: 'yid';            DisplayName: 'Yiddish'),
    (Code: 'yor';            DisplayName: 'Yoruba ')
    );

// Note:
// - Language files can be downloaded from: www.imageen.com/download/
// - 'osd.TrainedData' is not a language file, it is used for page orientation detection. If it is not included IEVision will be unable to automatically process pages that are incorrectly oriented (e.g. upside-down pages)