How do I build OpenCV with TesseractOCR? - c++

Question

I'm trying to build OpenCV with the Tesseract OCR module to use on a raspberry pi.
There is very little information online on how to build this, and what is available is very incomplete.
I'm building using Visual Studio 2017 on windows 10 using the VisualGDB plugin for cross compilation.
I can load the OpenCV CMake project and build it without issues.
I downloaded the OpenCV-contrib modules, and pointed the OPENCV_EXTRA_MODULES_PATH to C:/opencv/opencv_contrib-master/modules
I can see there is a text module that contains some sources for tesseract-ocr.
There are also entries for Tesseract_INCLUDE_DIR and Tesseract_LIBRARY
Does this mean I need to download Tesseract separately? Or can I just use the files that came with OpenCV contrib?
I cloned https://github.com/tesseract-ocr/tesseract
But there is no include folder anywhere. There is an src folder that contains headers and source files, but no include folder anywhere.
Do I need to build tesseract first? And when its built, it creates lib and include files?
I tried to build tesseract, but it required leptonica.
I was able to build leptonica. but I cannot manage to include it in Tesseract cmake.
I added the lines
add_subdirectory(${CMAKE_SOURCE_DIR}/../leptonica)
add_subdirectory(${CMAKE_SOURCE_DIR}/../leptonica/build)
find_package(Leptonica ${MINIMUM_LEPTONICA_VERSION} REQUIRED CONFIG)
But I'm getting errors:
add_subdirectory not given a binary directory but the given source C:\opencv\tesseract\CMakeLists.txt
Could not find a package configuration file provided by "Leptonica" C:\opencv\tesseract\CMakeLists.txt 124
Thanks,

Answer 1

I got tesseract to build, here is what I added:
To tesseract/CMakeLists.txt on line 105 comment out and add
#if(NOT EXISTS ${PROJECT_SOURCE_DIR}/.cppan)
# if (NOT Leptonica_DIR AND NOT MSVC)
# find_package(PkgConfig REQUIRED)
# pkg_check_modules(Leptonica REQUIRED lept>=${MINIMUM_LEPTONICA_VERSION})
# link_directories(${Leptonica_LIBRARY_DIRS})
# else()
# find_package(Leptonica ${MINIMUM_LEPTONICA_VERSION} REQUIRED CONFIG)
# endif()
#else()
# if (STATIC)
# set(CPPAN_BUILD_SHARED_LIBS 0)
# else()
# set(CPPAN_BUILD_SHARED_LIBS 1)
# endif()
# add_subdirectory(.cppan)
#endif()
set(Leptonica_DIR "C:/opencv/leptonica/build")
find_package(Leptonica ${MINIMUM_LEPTONICA_VERSION} REQUIRED CONFIG)
find_package(OpenCL QUIET)
Then some build options need to be changed. Same file, line 260.
Set all the PROPERTIES COMPILE_FLAGS "-marm" otherwise it wont compile
set_source_files_properties(
${CMAKE_CURRENT_SOURCE_DIR}/src/arch/dotproductsse.cpp
PROPERTIES COMPILE_FLAGS "-msse4.1")
set_source_files_properties(
${CMAKE_CURRENT_SOURCE_DIR}/src/arch/intsimdmatrixsse.cpp
PROPERTIES COMPILE_FLAGS "-msse4.1")
set_source_files_properties(
${CMAKE_CURRENT_SOURCE_DIR}/src/arch/dotproductavx.cpp
PROPERTIES COMPILE_FLAGS "-mavx")
set_source_files_properties(
${CMAKE_CURRENT_SOURCE_DIR}/src/arch/intsimdmatrixavx2.cpp
PROPERTIES COMPILE_FLAGS "-mavx2")
Then, copy the endianness.h and config_auto.h from C:\opencv\leptonica\build\src to C:\opencv\leptonica\src , again otherwise it wont build
Then it will complain about undefined PACKAGE_VERSION in tessdatamanager.cpp, so define it in the CMakeLists or just directly in the source file.
After that it should build, if your directories match.
My directory structure is as follows:
Tesseract git repository located in C:\opencv\tesseract
Tesseract build directory C:\opencv\tesseract\build
Leptonica repository located in C:\opencv\leptonica
Leptonica build directory in C:\opencv\leptonica\build (contains LeptonicaConfig.cmake,
Leptonica library files, (libleptonica.so, libleptonica.so.1.77.0, libleptonica.so.5.3.0) are in C:\opencv\leptonica\build\src