Building and Installing OpenCV with Extra Modules on Windows 7 64-bit

Recently, I have been working on a Computer Vision Project which requires me to use OpenCV Track API. However, this Track API is not available on the Pre-Built OpenCV Library. So, I need to build OpenCV from its original source and include the extra modules which contain the Track API.

Pre-requisite

Before I begin, make sure that you have already installed:

1. Microsoft Visual Studio (in this case, I use Visual Studio 2013 in 64-bit PC)
2. CMake for Windows (http://www.cmake.org/download/)

Optional: you might also need to temporarily disable any anti-virus software since it may prevent the process of building OpenCV.

Step-by-step Overview

1. Downloading OpenCV with extra modules
2. Making Visual Studio Project with CMake
3. Building OpenCV project
4. Developing application using OpenCV

Step 1: Downloading OpenCV with extra modules

First, let’s make a directory called “opencv-3.0” and make two other folder “build” and “source” inside it as depicted in figure below:

The build directory will be used for saving the result of building process, and the source folder will contain all of the OpenCV and its extra modules sources.

We can download OpenCV sources together with the extra modules from this following websites:

1. OpenCV: https://github.com/Itseez/opencv
2. Extra Modules: https://github.com/Itseez/opencv_contrib

After finish downloading, unzip each of them to designated folders as depicted in figure below:

As I mentioned before, the opencv folder will store all OpenCV sources, and the opencv_contrib folder will store all extra modules sources.

Step 2: Making Visual Studio Project with CMake

After we completely download and extract the OpenCV and its extra modules sources, we will use CMake application to make an OpenCV Visual Studio Project. The OpenCV Visual Studio Project will later be used to build the OpenCV library together with its extra modules completely.

So, let’s launch CMake application and then specify the source and build directory as shown in figure below. The red box must be filled with the directory path of OpenCV source, and the green box must be filled with the directory path of designated build folder.

After that, click the “Configure” button (yellow box), this is how we configure CMake to build the OpenCV library in our PC. Since I use Visual Studio 2013 in 64-bit machine, I will use the following configuration.

Push the “Next” button, and specify the C and C++ compilers. We can use the compiler in our local machine which is located in {Visual Studio}/VC/bin/cl.exe.

After that, hit “Finish” button and then let’s wait until CMake finishes its initial configuration. Whenever, the process has finished, we can see the configuration log denoting that it’s done as shown in figure below:

Next, we need to specify the extra modules path which is {opencv-3.0}/source/opencv_contrib /modules as depicted in figure below:

Please note that, you actually may uncheck several stuffs that in unimportant to speed up the building process, for example: BUILD_DOCS. However, if you don’t have any idea about what they are, just leave it as it is. After we have finished re-configuring, we need to hit the “Configure” button once again.

Now, let’s double check whether the desired extra modules are on the list. Go to the Configuration Log in the CMake GUI, and find “OpenCV modules” -> “To be built”. If everything goes well, the tracking module should also be there. Here is the example result of re-configuring the CMake with extra modules.

OpenCV modules
To be built	hal core flann imgproc ml photo reg surface_matching video face imgcodecs shape videoio highgui objdetect optflow superres tracking ts ximgproc xobjdetect xphoto adas bgsegm bioinspired features2d latentsvm line_descriptor saliency text calib3d ccalib datasets rgbd videostab xfeatures2d stitching
Disabled	world contrib_world
Disabled by dependency	–
Unavailable	androidcamera cudaarithm cudabgsegm cudacodec cudafeatures2d cudafilters cudaimgproc cudalegacy cudaobjdetect cudaoptflow cudastereo cudawarping cudev java python2 python3 viz cvv matlab

After that, we can start generating the Visual Studio Project by pushing the “Generate” button. And voila, your build directory will be fulfilled with OpenCV Visual Studio Project. Check out this webpage that I use while learning about this step in here [1].

Step 3: Building OpenCV Project

Now, we need to launch the Visual Studio application open the OpenCV.sln file located in the build directory.

Note: if you use the C: directory probably you will need an Administrator privilege to build the OpenCV project, so you must also run the Visual Studio under the Administrator privilege.

After loading all of the project files, go to Solution Explorer, find the INSTALL project inside the CMakeTargets directory, and build that project. It may take several minutes for the whole building process to finish, just be patient.

Whenever the whole building process has been done, there will be several resulting new directories, such as: install, bin, and lib directory inside the build directory. We will use the install directory for building our own computer vision application. Here is how it looks like inside the install directory whenever the whole building process has been finished.

We can make Debug and Release version just by changing the build configuration of OpenCV project.

Now, we have already finished building our own OpenCV library as well as its extra modules. If you need another reference for installing OpenCV in Windows provided by the official OpenCV website, please refer to the following webpage [2].

Step 4: Developing Application using OpenCV

Now, that we have already built the OpenCV library, we can use it to build our own application which uses OpenCV great features. There are two main steps that are required:

1. Registering OpenCV Environment Variables
2. Creating OpenCV Project Property Sheet

Step 4.1: Registering OpenCV Environment Variables

Before we are able to use OpenCV in our application, we need to register OpenCV path to our system environment variables. First, we need to open the Environment Variables window, click “New”, and specify the new variable name and value. The new variable name should be “OPENCV_DIR” and the value should be the vc12 directory inside install\x64 directory. Here is the example:

By doing this, it will be easy for you to maintain your Open CV library. Just update this environment variable once you have the latest Open CV version. After that, we also need to modify the “Path” variable to register our OpenCV .dll directory. This can be done by adding %OPENCV_DIR%\bin entry at the last part of Path variable value.

Please note that I use %OPENCV_DIR% instead of static folder path. This will help when updating our Open CV version in the future. For more detailed step on how to set the environment variables refer to this webpage [3].

Step 4.2: Creating OpenCV Project Property Sheet

Now, we can start building our own application which uses OpenCV feature in Visual Studio. Here are the steps:

1. Make an empty C++ project in Visual Studio
2. Open the Property Manager View, right click on “Debug | 64” folder and “Add New Project Property Sheet” as shown in figure below:
3. Specify the Property Sheet name, for example: OPENCV3_DEBUG and click “Add” to proceed to next step.
4. Right click on the newly created Property Sheet and select Properties.
5. Configure the Additional Include Libraries (Common Properties->C/C++/General) by adding the following entry: $(OPENCV_DIR)\..\..\include .This will help the C\C++ compiler to look for additional libraries while building our application.
6. Configure the Additional Library Directories (Common Properties->Linker->General) by adding the following entry: $(OPENCV_DIR)\lib
7. Specify all Additional Dependencies (Common Properties->Linker->Input) by adding all of the .lib files on the %(OPENCV_DIR)\lib directory. Please note, if we are creating Debug Mode Property Sheet, we must only specify .lib files which name are ended with “d” letter. For example: opencv_bgsegm300d.lib, opencv_bioinspired300d.lib. However, if we are creating Release Mode Property Sheet, we must only specify .lib files without “d”. For example: opencv_bgsegm300.lib, opencv_bioinspired300.lib. Here is the example:
   

   Debug Mode Property Sheet	Release Mode Property Sheet

8. Click OK and start making your OpenCV application.

The advantage of using Property Sheet is that we don’t have to repeat this process over again when we want to start another OpenCV project. We can just add the existing OpenCV Project Property Sheet. For more info regarding this step, please refer to this webpage [4].

References

1. eavise, “Building OpenCV yourself on Windows 7 x64 with OpenCV 2.4.5 and CUDA 5.0,” 2015. [Online].
2. opencv, “Installation in Windows,” 2015. [Online].
3. opencv, “Set the OpenCV enviroment variable and add it to the systems path,” 2015. [Online].
4. opencv, “How to build applications with OpenCV inside the Microsoft Visual Studio,” 2015. [Online].