Stefan Weil
40c1cf671f
unittest: Fix and enable pango_font_info_test
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-28 08:09:28 +02:00
Stefan Weil
04d85b4c0f
Add more test code for normstrngs_test
...
unilib.h is now available, so more code can be enabled.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-28 07:35:27 +02:00
Stefan Weil
aa54bf0f8b
Fix code from tensorflow/models/research/syntaxnet/util/utf8
...
See https://github.com/tensorflow/models/issues/7090 .
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-28 07:30:40 +02:00
Stefan Weil
0702194246
Add code from tensorflow/models
...
The new code was copied from the latest code on GitHub
(https://github.com/tensorflow/models/tree/master/research/syntaxnet ).
It is required for pango_font_info_test and other unit tests.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-28 07:30:40 +02:00
Stefan Weil
7093727da1
cmake: Fix build with OpenMP (issue #2462 )
...
Builds with gcc wrongly used /openmp instead of -fopenmp.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-25 13:20:52 +02:00
Stefan Weil
676b18834c
Fix check for icu 52.1 or newer
...
It detected old versions but did not disable the training build.
This completes commit 66da4df11d
.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-25 10:55:33 +02:00
Stefan Weil
5895534b5e
Update enum from unicode/uchar.h
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-25 10:55:33 +02:00
Stefan Weil
252d80cb6d
unittest: Fix function QCHECK (issue #2517 )
...
The function must print an error message if the condition fails.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-23 19:05:42 +02:00
Egor Pugin
111ef842ae
Add VS2019
2019-06-23 11:34:58 +03:00
Stefan Weil
c1264c189e
Fix format string for 64 bit integer
...
This fixes also a warning from gcc.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-23 09:31:09 +02:00
Stefan Weil
dfd35d3e27
baseapi: Remove old code
...
The workaround is no longer needed because _splitpath and _MAX_FNAME
were removed in commit cc0d87c5b8
.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-23 09:15:32 +02:00
Stefan Weil
dd261e8d42
Replace code using _splitpath_s (win32)
...
That simplifies the code and removes a dependency on "newer"
versions of Windows.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-23 09:15:15 +02:00
Stefan Weil
f522b039a5
Remove outdated comment
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 21:03:19 +02:00
Stefan Weil
ea20bf0373
Remove dummy code from LSTMTrainer::InitTensorFlowNetwork
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 21:01:40 +02:00
Stefan Weil
12f63c6097
Add package libpango1.0-dev to Travis configuration
...
It is needed to build the training tools.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 20:38:10 +02:00
Stefan Weil
efa3cae06d
Simplify unittest/Makefile.am
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 20:29:39 +02:00
Stefan Weil
41f91c96c8
cmake: Build training tools also on Linux and macOS
...
This enables CI tests for the code in src/training on Linux and macOS.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 20:27:56 +02:00
Egor Pugin
ab28a03e93
Merge pull request #2514 from stweil/tessresultcallback
...
Move LSTMTrainer from libtesseract to libtesseract_training
2019-06-22 18:34:49 +03:00
Stefan Weil
df98bb7368
Move LSTMTrainer from libtesseract to libtesseract_training
...
LSTMTrainer is only used for training, so the shared library for
Tesseract can be made smaller.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 16:23:51 +02:00
Egor Pugin
ee37abb7b3
Merge pull request #2515 from stweil/baseapi
...
Replace callback by direct function calls in TessBaseAPI::GetComponen…
2019-06-22 17:04:48 +03:00
Stefan Weil
cb2957b3d2
Replace callback by direct function calls in TessBaseAPI::GetComponentImages
...
The new code avoids dynamic memory allocation, uses faster function calls
and allows removing more code from tesscallback.h.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 14:54:31 +02:00
Stefan Weil
3159f42257
Remove unused GenericVector::dot_product
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 12:59:21 +02:00
Stefan Weil
bef73d9956
Remove unused GenericVector::compact
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 12:59:08 +02:00
Egor Pugin
3c6a04ea1a
Merge pull request #2512 from stweil/tessresultcallback
...
Simplify class LSTMTrainer
2019-06-22 13:41:21 +03:00
Stefan Weil
2a9b2fb32a
Remove wrong description for GenericVector::set_compare_callback and simplify code
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 11:22:07 +02:00
Stefan Weil
bd13069fe8
Simplify class LSTMTrainer
...
The function pointers and callbacks file_reader_, file_writer_,
checkpointer_reader_ and checkpoint_writer_ are always set to
the same values. Replacing them by direct function calls
simplifies the code and allows removing more code from tesscallback.h.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-22 09:18:13 +02:00
zdenop
dff33d6d93
Merge pull request #2510 from stweil/tessresultcallback
...
Simplify indirect call of LMPainPoints::GeneratePainPoint
2019-06-22 06:55:27 +02:00
Stefan Weil
3871caae86
Simplify indirect call of LMPainPoints::GeneratePainPoint
...
It does neither need a temporary TessResultCallback2 nor the function
LMPainPoints::GenerateForBlamer.
This also allows removing more code from tesscallback.h.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-21 17:09:33 +02:00
zdenop
60b4c68d31
tesstrain_utils.sh: remove redundant code
2019-06-20 18:42:29 +02:00
Stefan Weil
5f23290655
tesscallback: Remove more unused code
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-20 08:38:00 +02:00
Egor Pugin
21514457d4
Remove old and misguiding build steps on windows.
2019-06-17 18:53:36 +03:00
Stefan Weil
b967c62880
unittest: Add missing Leptonica library for textlineprojection_test
...
It is needed for builds with --enable-shared.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-17 15:10:16 +02:00
Stefan Weil
2c78735d97
ocrfeatures: Remove locally used functions from global interface
...
ReadFeature, WriteFeature are only used locally.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-17 15:09:39 +02:00
zdenop
a3593d994b
Merge pull request #2499 from stweil/embedded
...
Remove code for embedded build
2019-06-17 10:24:45 +02:00
Stefan Weil
674d6a90d8
Remove code for embedded build
...
That code is unrelated to Tesseract and can be easily implemented
by external projects which require it.
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-17 09:55:33 +02:00
zdenop
60aee9f821
create OUTPUT_DIR did not exist; fixes #2497
2019-06-16 15:07:16 +02:00
Stefan Weil
ceabab8373
unittest: Catch missing eng.traineddata in baseapi_test
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-16 08:11:16 +02:00
Stefan Weil
bbd3626d77
unittest: Fix and enable normstrngs_test
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-16 08:01:11 +02:00
Stefan Weil
73e5241004
unittest: Fix and enable textlineprojection_test
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-15 10:22:44 +02:00
zdenop
982bdf3845
Merge pull request #2495 from stweil/scanutils_test
...
unittest: Fix and enable scanutils_test
2019-06-14 20:45:28 +02:00
zdenop
fad96db497
Merge pull request #2494 from Shreeshrii/master
...
Allow saving of box/tiff pairs during legacy tesseract training
2019-06-14 20:44:49 +02:00
Stefan Weil
e0e29126ac
unittest: Fix and enable scanutils_test
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-14 16:51:39 +02:00
Shree
6fa4587949
Allow saving of box/tiff pairs during base tesseract training
2019-06-14 09:35:39 +00:00
Shree
45cdf741ae
Allow saving of box/tiff pairs during base tesseract training
2019-06-14 09:32:41 +00:00
Shree
832c6edb97
Allow saving of box/tiff pairs during base tesseract training
2019-06-14 09:25:54 +00:00
zdenop
3afc946d5e
Merge pull request #2492 from jbarlow83/fix-text2image
...
Fix text2image compilation on C++17 compilers
2019-06-14 07:45:14 +02:00
James R. Barlow
a9890afd12
Fix text2image compilation on C++17 compilers
...
C++17 drops support for `std::random_shuffle`, breaking C++17 compilers
that run to compile text2image.cpp. std::shuffle is valid on C++11
through C++17, so use std::shuffle instead.
Due to the use `std::random_shuffle`, `text2image --render_ngrams`
would not give consistent results for different compilers or platforms.
With the current change, the same random number generator is used for
all platforms and initialized to the same seed, so training output
should be consistent.
2019-06-13 16:07:20 -07:00
Stefan Weil
3c507100c6
unittest: Fix and enable ligature_table_test
...
Signed-off-by: Stefan Weil <sw@weilnetz.de>
2019-06-11 16:40:23 +02:00
Egor Pugin
5717b098aa
[sw] Remove inexistent include dirs.
2019-06-08 17:45:37 +03:00
Egor Pugin
c570772bce
[sw] Exclude missing include dir.
2019-06-08 17:41:02 +03:00