This is a repo for "Recognition systems development tools" course. Students should have access to this repo. Based on https://github.com/ansgri/cpp-boilerplate-minimal (rsdt branch at the time of writing).

Assuming Visual Studio 9 2008 as your sole installed compiler. Then from the root of the copied & customized cpp-boilerplate, run this:

• mkdir build.vc9 && cd build.vc9
• cmake ..
• start .sln
• Build Solution (F7)

To run the sample, OpenCV dlls must be in the system path.

• mkdir build.cmake.debug && cd build.cmake.debug
• cmake .. -DCMAKE_BUILD_TYPE=Debug
• make -j5

If you already have Boost installed or built, just set the BOOST_ROOT environment variable, like this:

set BOOST_ROOT=c:\Users\ansgri\thirdparty-src\boost_1_55_0

Otherwise, build it first following this approximate procedure:

• you must have Visual Studio (or other compiler of choice) installed (obviously)
• download and unpack the source archive
• from the unpacked source dir execute bootstrap.bat
• review the build options: b2 --help
• build like this: b2 link=shared variant=release threading=multi runtime-link=shared

On Linux this is similar and generally more streamlined. Just RTFM.

TODO: instructions for the case when OpenCV is installed via binary installer

Building from source on Windows:

• git clone git://github.com/Itseez/opencv.git
• git checkout 2.4.6 (or other stable tag; to see all, execute git tag)
• mkdir build.user && cd build.user
• cmake .. && start OpenCV.sln
• build solution (F7) in both Release and Debug configurations
• build 'INSTALL' target in both Release and Debug configurations (in Solution Explorer, right-click and Build, it's in CMake Targets folder)
• you'll have install folder under build.user. Its full path must be set to OpenCV_DIR environment variable.