Fast integer versions of trained LSTM models
Find a file
Stefan Weil 8741641865 Mention Tesseract 5 too
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2024-08-01 21:41:24 +02:00
script Move trained data for scripts to new subdirectory 2018-03-10 10:18:55 +01:00
tessconfigs@3decf1c825 Update tessconfigs 2019-10-23 13:31:56 +02:00
.gitmodules Update URL for tessconfigs submodule (use HTTPS) 2019-10-11 13:08:13 +02:00
afr.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
amh.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ara.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
asm.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
aze.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
aze_cyrl.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
bel.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ben.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
bod.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
bos.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
bre.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
bul.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
cat.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ceb.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ces.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
chi_sim.traineddata Fix extra intra-word spaces by adding config file 2018-02-20 20:18:27 +05:30
chi_sim_vert.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
chi_tra.traineddata Fix extra spaces in words for chi_tra 2018-02-20 22:50:06 +05:30
chi_tra_vert.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
chr.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
configs Add tessconfigs submodule and links for required tessdata files 2019-05-22 22:47:58 +02:00
cos.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
cym.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
dan.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
deu.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
deu_latf.traineddata Rename frk -> deu_latf (ISO 639-3, ISO 15924) 2024-03-09 11:23:20 +01:00
div.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
dzo.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ell.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
eng.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
enm.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
epo.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
equ.traineddata Add equ.traineddata (copy from tessdata) 2019-10-30 08:44:48 +01:00
est.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
eus.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
fao.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
fas.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
fil.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
fin.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
fra.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
frk.traineddata Create symbolic link for frk.traineddata 2024-03-09 11:23:20 +01:00
frm.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
fry.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
gla.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
gle.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
glg.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
grc.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
guj.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
hat.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
heb.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
hin.traineddata Add config files to fix auto PSM issue 1273 2018-02-26 20:21:11 +05:30
hrv.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
hun.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
hye.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
iku.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ind.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
isl.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ita.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ita_old.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
jav.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
jpn.traineddata Fix extra intra-word spaces by adding config file 2018-02-20 20:18:27 +05:30
jpn_vert.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
kan.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
kat.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
kat_old.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
kaz.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
khm.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
kir.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
kmr.traineddata correct name kur_ara to kmr - Kurmanji (Latin script) 2018-04-25 22:52:09 +05:30
kor.traineddata Fix extra intra-word spaces by adding config file 2018-02-20 20:18:27 +05:30
kor_vert.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
lao.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
lat.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
lav.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
LICENSE Add Apache license file 2019-06-13 20:30:10 +02:00
lit.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ltz.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
mal.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
mar.traineddata Add config files to fix auto PSM issue 1273 2018-02-26 20:21:11 +05:30
mkd.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
mlt.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
mon.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
mri.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
msa.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
mya.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
nep.traineddata Add config files to fix auto PSM issue 1273 2018-02-26 20:21:11 +05:30
nld.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
nor.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
oci.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ori.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
osd.traineddata Use legacy Orientation Script Detector (OSD) because that is the only thing that currently works. 2017-09-15 11:49:11 -07:00
pan.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
pdf.ttf Add tessconfigs submodule and links for required tessdata files 2019-05-22 22:47:58 +02:00
pol.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
por.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
pus.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
que.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
README.md Mention Tesseract 5 too 2024-08-01 21:41:24 +02:00
ron.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
rus.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
san.traineddata Fix config file for default oem mode, change to --oem 1 2017-09-15 18:44:40 +05:30
sin.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
slk.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
slv.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
snd.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
spa.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
spa_old.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
sqi.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
srp.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
srp_latn.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
sun.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
swa.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
swe.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
syr.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
tam.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
tat.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
tel.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
tgk.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
tha.traineddata Fix extra intra-word spaces by adding config file 2018-02-20 20:18:27 +05:30
tir.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ton.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
tur.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
uig.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
ukr.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
urd.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
uzb.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
uzb_cyrl.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
vie.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
yid.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00
yor.traineddata Initial import to github (on behalf of Ray) 2017-09-14 14:35:44 -07:00

tessdata_fast Fast integer versions of trained models

This repository contains fast integer versions of trained models for the Tesseract Open Source OCR Engine.

These models only work with the LSTM OCR engine of Tesseract 4 and 5.

  • These are a speed/accuracy compromise as to what offered the best "value for money" in speed vs accuracy.
  • For some languages, this is still best, but for most not.
  • The "best value for money" network configuration was then integerized for further speed.
  • Most users will want to use these traineddata files to do OCR and these will be shipped as part of Linux distributions eg. Ubuntu 18.04.
  • Fine tuning/incremental training will NOT be possible from these fast models, as they are 8-bit integer.
  • When using the models in this repository, only the new LSTM-based OCR engine is supported. The legacy tesseract engine is not supported with these files, so Tesseract's oem modes '0' and '2' won't work with them.

Deprecated models

The former model frk has been renamed to deu_latf because the old name was never ISO compliant. A symbolic link from deu_latf.traineddata to frk.traineddata may help to migrate from the old name to the new one. However, all projects and distributions are encouraged to use only the new name. The use of frk is deprecated and will not be supported in the future.

Two types of models

The repository contains two types of models,

  • those for a single language and
  • those for a single script supporting one or more languages.

Most of the script models include English training data as well as the script, but not Cyrillic, as that would have a major ambiguity problem.

On Debian and Ubuntu, the language based traineddata packages are named tesseract-ocr-LANG where LANG is the three letter language code eg. tesseract-ocr-eng (English language), tesseract-ocr-hin (Hindi language), etc.

On Debian and Ubuntu, the script based traineddata packages are named tesseract-ocr-script-SCRIPT where SCRIPT is the four letter script code eg. tesseract-ocr-script-latn (Latin Script), tesseract-ocr-script-deva (Devanagari Script), etc.

Data files for a particular script

Initial capitals in the filename indicate the one model for all languages in that script. These are now available under script subdirectory.

  • Latin is all latin-based languages, except vie.
  • Vietnamese is for latin-based Vietnamese language.
  • Fraktur is basically a combination of all the latin-based languages that have an 'old' variant.
  • Devanagari is for hin+san+mar+nep+eng.

LSTM training details for different languages and scripts

For Latin-based languages, the existing model data provided has been trained on about 400000 textlines spanning about 4500 fonts. For other scripts, not so many fonts are available, but they have still been trained on a similar number of textlines. eg. Latin ~4500 fonts, Devanagari ~50 fonts, Kannada 15.

With a theory that poor accuracy on test data and over-fitting on training data was caused by the lack of fonts, the training data has been mixed with English, so that some of the font diversity might generalize to the other script. The overall effect was slightly positive, hence the script models include English language also.

Example - jpn and Japanese

'jpn' contains whatever appears on the www that is labelled as the language, trained only with fonts that can render Japanese.

Japanese contains all the languages that use that script (in this case just the one) PLUS English.The resulting model is trained with a mix of both training sets, with the expectation that some of the generalization to 4500 English training fonts will also apply to the other script that has a lot less.

'jpn_vert' is trained on text rendered vertically (but the image is rotated so the long edge is still horizontal).

'jpn' loads 'jpn_vert' as a secondary language so it can try it in case the text is rendered vertically. This seems to work most of the time as a reasonable solution.


See the Tesseract documentation for additional information.

All data in the repository are licensed under the Apache-2.0 License, see file LICENSE.