How to compile

Compiling C++ Example Programs on Any Operating System Using CMake

The best way to compile a program that uses dlib is to use CMake. For example, the following commands will compile the example programs on any operating system:
cd examples
mkdir build
cd build
cmake ..
cmake --build . --config Release
Note that you also need to have a C++11 compiler installed on your system. There are free C++11 compilers for most operating systems. For example, Visual Studio is free on Windows and GCC is free and works well on Mac OS X and Linux systems. If you have multiple compilers/IDEs installed then you can tell CMake which one you want it to use via the -G option.

The examples/CMakeLists.txt file tells CMake how to build the examples. You can create your own projects by starting with this file and editing it however you like. You can also perform additional configuration of a cmake project using the cmake-gui or ccmake tool. For example, if you are using dlib's face detector then you should turn on either SSE4 or AVX instructions since this makes it run much faster (also see this FAQ).

Finally, note that when using Visual Studio, CMake will by default generate a 32bit executable. This means the programs you compile will only be able to use 2GB of RAM. To avoid this, you need to tell CMake to generate a 64bit executable. You do this by using a command like

cmake -G "Visual Studio 14 2015 Win64" ..
instead of
cmake ..
You can see the list of valid arguments to -G by running cmake with no options.

Compiling Dlib's Python Interface

Go to the base folder of the dlib repository and run python install. That should compile and install the dlib python API on your system.

Alternatively, if you want to add more python bindings to dlib's python interface then you probably want to avoid the file and work directly using CMake. In particular, dlib's python API is built by the CMake project in the tools/python folder. You build this project using the usual CMake commands and when compiled it outputs the dlib shared library that defines the python API for dlib.

Compiling C++ Examples Without CMake

In most cases, to use this library all you have to do is extract it somewhere, make sure the folder containing the dlib folder is in your include path, and finally add dlib/all/source.cpp to your project. It is worth noting that most of dlib is "header-only" which means that, in many cases, you don't actually have to build dlib/all/source.cpp into your application. So if you don't get linker errors when you exclude dlib/all/source.cpp from your project then you don't need it.

An example makefile that uses this library can be found here: dlib/test/makefile. It is the makefile used to build the regression test suite for this library.

Again, note that you should not add the dlib folder itself to your compiler's include path. Doing so will cause the build to fail because of name collisions (e.g. dlib/string.h with string.h from the standard library). Instead you should add the folder that contains the dlib folder to your include search path and then use include statements of the form #include <dlib/queue.h>. This will ensure that everything builds correctly.

Note also that if you want to work with jpeg/png/gif files using dlib then you will need to link your program with libjpeg, libpng, and/or libgif. You also need to tell dlib about this by defining the DLIB_JPEG_SUPPORT, DLIB_PNG_SUPPORT, and DLIB_GIF_SUPPORT preprocessor directives. How you "link to libjpeg/libpng/libgif" varies from platform to platform. On UNIX machines you usually just add a -ljpeg, -lpng, or -lgif switch to your compiler (after installing the libraries). On windows it's less well defined. So dlib comes with a copy of libjpeg and libpng in the dlib/external folder so you can statically compile them into your application if no system wide version is available on your machine. If all this talk about linking is confusing to you then just use CMake. It will set this all up for you.

Dlib is also capable of using any optimized BLAS or LAPACK libraries that are installed on your system. Linking to these libraries will make many things run faster. To do this you define the DLIB_USE_BLAS and/or DLIB_USE_LAPACK preprocessor directives and then link your program with whatever BLAS or LAPACK libraries you have. If you use CMake it will set this up automatically.

Compiling on Linux From Command Line

From within the examples folder, you can compile nearly all of the examples with a single command like so:
g++ -std=c++11 -O3 -I.. ../dlib/all/source.cpp -lpthread -lX11 example_program_name.cpp
Note that not all examples require this much work. For example, the svm_ex.cpp example can be compiled with just:
g++ -std=c++11 -O3 -I.. svm_ex.cpp
On non-Linux systems like Solaris, you might have to link to other libraries. For example, I have seen systems where it was also necessary to supply -lnsl or -lsocket options to g++. Additionally, the X11 development library isn't installed on Ubuntu by default. So if you require it and are using Ubuntu you can install it by typing:
sudo apt-get install libx11-dev

Compiling on Windows Using GCC

The commands for gcc on windows are the same as above but you may also have to link (via the -l option) to the following libraries: gdi32, comctl32, user32, winmm, ws2_32, or imm32.

Compiling on Windows Using Visual Studio 2015 or Newer

All you need to do is create an empty console project. Then add dlib/all/source.cpp to it and add the folder containing the dlib folder to the #include search path. Then you can compile any example program by adding it to your project.

Again, note that dlib will only be able to work with jpeg and png files if you link in libjpeg and libpng. In Visual Studio, the easiest way to do this is to add all the libjpeg, libpng, and zlib source files in the dlib/external folder into your project and also define the DLIB_PNG_SUPPORT and DLIB_JPEG_SUPPORT preprocessor directives. If you don't know how to configure Visual Studio then you should use CMake as shown above since it will take care of everything automatically.

Miscellaneous Preprocessor Directives

In addition to the preprocessor directives mentioned above, there are a few more you can supply during the build process to cause the library to build in various optional ways. By default, the library will always do something reasonable, but they are listed here in the event that you need to use them.


Defining this directive causes all the DLIB_ASSERT macros to be active. If you are using Visual Studio or CMake then ENABLE_ASSERTS will be automatically enabled for you when you compile in debug mode. However, if you are using a different build system then you might have to manually enable it if you want to turn the asserts on.


This is a #define directive that you can set to cause the library to exclude all non ISO C++ code (The things in the API wrappers section and any objects that depend on those wrappers). This is useful if you are trying to build on a system that isn't fully supported by the library or if you just decide you don't want any of that stuff compiled into your program for your own reasons.


This is just like the DLIB_ISO_CPP_ONLY option except that it excludes only the GUI part of the library. An example of when you might want to use this would be if you don't need GUI support and you are building on a UNIX platform that doesn't have the X11 headers installed.


This preprocessor directive causes the dlib headers to pull in all the code that would normally be built in dlib/all/source.cpp. Thus if you #define NO_MAKEFILE you won't have to add dlib/all/source.cpp to your project. The only time this is useful is when your project consists of a single translation unit (i.e. a single cpp file). In this instance NO_MAKEFILE allows you to easily build your project on the command line by saying something like g++ -DNO_MAKEFILE project.cpp. But again, this is only for single cpp file projects. If you use NO_MAKEFILE with projects that contain more than one cpp file you will get linker errors about multiply defined symbols.

Also note that if you use this macro then the stack trace functionality in the library will be disabled.

#define DLIB_THREAD_POOL_TIMEOUT <time-in-milliseconds>

If you use dlib to create your threads then you receive the benefit of the dlib dynamic thread pool (Note that the dlib::thread_pool object is something else unrelated to this so don't confuse the two). This pool enables dlib to spawn new threads very rapidly since it draws threads back out of its thread pool when the pool isn't empty.

Thus, when a thread that was created by dlib ends it actually goes back into the dlib thread pool and waits DLIB_THREAD_POOL_TIMEOUT milliseconds before totally terminating and releasing its resources back to the operating system. The default timeout used by this library is 30,000 milliseconds (30 seconds). You may however change this to whatever you like by defining DLIB_THREAD_POOL_TIMEOUT to some new value.