Installing opencv From Command Line (Windows) - c++

I am trying to use opencv in a project, and am running into problems 'installing' it. I have extracted the opencv files and have created a small test program:
#include "opencv2/highgui/highgui.hpp"
#include <iostream>
int main(int argc, char **argv){
cv::Mat im=cv::imread((argc==2)? argv[1]: "testing.jpg",1);
if (im.empty()){
std::cout << "Cannot open image." << std::endl;
} else {
cv::imshow("image",im);
cv::waitKey(0);
}
return 0;
}
To compile the program I have used the command below:
g++ -I"../../PortableGit/opt/opencv/build/include/" -L"../../PortableGit/opt/opencv/build/x64/vc15/lib" main.cpp -lopencv_core -lopencv_highgui -o main
I get the errors below:
In file included from ../../PortableGit/opt/opencv/build/include/opencv2/core.hpp:3293:0,
from ../../PortableGit/opt/opencv/build/include/opencv2/highgui.hpp:46,
from ../../PortableGit/opt/opencv/build/include/opencv2/highgui/highgui.hpp:48,
from main.cpp:1:
../../PortableGit/opt/opencv/build/include/opencv2/core/utility.hpp:714:14: error: 'recursive_mutex' in namespace 'std' does not name
a type
typedef std::recursive_mutex Mutex;
^~~~~~~~~~~~~~~
../../PortableGit/opt/opencv/build/include/opencv2/core/utility.hpp:715:25: error: 'Mutex' is not a member of 'cv'
typedef std::lock_guard<cv::Mutex> AutoLock;
^~
../../PortableGit/opt/opencv/build/include/opencv2/core/utility.hpp:715:25: error: 'Mutex' is not a member of 'cv'
../../PortableGit/opt/opencv/build/include/opencv2/core/utility.hpp:715:34: error: template argument 1 is invalid
typedef std::lock_guard<cv::Mutex> AutoLock;
I believe that it has something to do with mingw binaries no longer being included with opencv. I am missing the opencv/build/x86/mingw directory.
My questions are:
How do I 'install' opencv and use it without also installing some sort of IDE and/or CMake? (I prefer to use vim and the command line.)
Once installed, what command do I use to compile and link a program with opencv?
Any help is appreciated.
Edit:
This appears to be a problem with GCC's implementation of threads on windows. Using mingw-w64 instead of mingw fixed the std::recursive_mutex issue, but now the linker cannot find the proper files.
/i686-w64-mingw32/bin/ld.exe: cannot find -lopencv_core
/i686-w64-mingw32/bin/ld.exe: cannot find -lopencv_highgui

After quite a bit of trying things out, this is what I got to work. Oddly, following the LINUX guide to install opencv worked better than the WINDOWS guide, even though I have a windows computer.
Guide to Installing OpenCV on Windows Without VS
Heads-up: This is a multi-step process, 3 separate tools are required. Be prepared for this to take a while.
Part 1: Get everything ready
Download MinGW-w64.
On the downloads page, click on the "MinGW-w64-builds" option. Do not click on the "win-builds" option.
The reason MinGW-w64 has to be used is because it is a newer version of the MinGW compiler suit that has been improved for windows. This means that it supports the posix thread system, where as the standard MinGW compiler only supports the win32 thread system. OpenCV relies on the posix thread system, necessitating the MinGW-w64 compiler.
Extract the MinGW-w64 zip folder to a directory. In my case its PortableGit/opt/MinGW-w64
At this point, you can add the MingGW-w64/mingw32/bin folder to your path. (Assuming that this won't cause any conflicts.) If you do so, you will not have to constantly specify the g++ executable directory to run it. This is up to your discretion.
Download an opencv release.
Do not download the package for windows, click the button that says "sources"
Extract the opencv sources zip folder to a directory. In my case its PortableGit/opt/opencv-4.3.0
Also download the opencv_contrib source files directly from the repository.
Extract that folder and place it inside the top level opencv folder: PortableGit/opt/opencv-4.3.0/opencv_contrib in my case.
Download CMake.
I downloaded the zip folder, but you can download the installer if you wish.
Extract the CMake zip folder if you downloaded that, or run the installer. I put my CMake folder here: PortableGit/opt/cmake-3.17.1-win32-x86
At this point, you can add the cmake-3.17.1-win32-x86/bin folder to your path. (Assuming that this won't cause any conflicts.) If you do so, you will not have to constantly specify the cmake executable directory to run it. This is up to your discretion.
Part 2: Build OpenCV
Navigate to the opencv directory and create a build folder and cd into it.
mkdir build && cd build
Run the following export commands.
export CC=/PortableGit/MinGW-w64/mingw32/bin/gcc.exe
export CXX=/PortableGit/MinGW-w64/mingw32/bin/g++.exe
This is to make sure the next cmake command uses the proper compilers.
Run the following cmake command from within that folder:
PortableGit/opt/cmake-3.17.1-win32-x86/cmake.exe -G "MinGW Makefiles" -DCMAKE_BUILD_TYPE=Release -DOPENCV_VS_VERSIONINFO_SKIP=1 -DOPENCV_EXTRA_MODULES_PATH="/PortableGit/opt/opencv-4.3.0/opencv_contrib/modules/" ..
The -G flag specifies that we are creating build files for the MinGW compiler
The -DCMAKE_BUILD_TYPE=Release specifies that we are building the release version of opencv and not the debug version.
The DOPENCV_EXTRA_MODULES_PATH needs to be set to the modules folder inside the opencv_contrib folder. For me it was PortableGit/opt/opencv-4.3.0/opencv_contrib/modules
The DOPENCV_VS_VERSIONINFO_SKIP specifies to not include version info. If not set, the compiler will throw an error complaining about not having version files. (Shown below for reference.)
gcc: error: long: No such file or directory
mingw32-make[2]: *** [modules\core\CMakeFiles\opencv_core.dir\build.make:1341:
modules/core/CMakeFiles/opencv_core.dir/vs_version.rc.obj] Error 1
If successful, the cmake command will finish like this:
Now run this command, again from the build folder: /PortableGit/opt/MinGW-w64/mingw32/bin/mingw32-make.exe -j7
mingw32-make.exe is the windows equivalent of the Linux make command.
The -j7 option run the process with a maximum of 7 threads.
This will take a while! It took my laptop ~20 minutes to complete
If the make command ends in an error, make sure to reset your build directory before continuing any troubleshooting. This is done through this series of commands
rm -rf build
mkdir build
cd build
Part 3: Using OpenCV
To use the opencv library that you just compiled in a project of your own, compile the project with these flags from your projects main directory.
Remember that your compiler now has to be set to the mingw-w64 compiler for opencv support.
I added indentation and newlines for readability, but when entering this in the terminal do not include the newlines or indents.
The number at the end of the linker options may change depending on the version of opencv you downloaded. I downloaded opencv-4.3.0, making my number 430, but yours may be different.
PortableGit/opt/MinGW-w64/bin/g++.exe
-I../../PortableGit/opt/opencv-4.3.0/include/
-I../../PortableGit/opt/opencv-4.3.0/build/
-I../../PortableGit/opt/opencv-4.3.0/modules/core/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/calib3d/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/dnn/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/features2d/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/flann/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/gapi/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/highgui/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/imgcodecs/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/imgproc/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/ml/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/objdetect/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/photo/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/stitching/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/ts/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/video/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/videoio/include/
-I../../PortableGit/opt/opencv-4.3.0/modules/world/include/
-L../../PortableGit/opt/opencv-4.3.0/build/lib/
*.hpp
*.cpp
-lopencv_calib3d430
-lopencv_core430
-lopencv_dnn430
-lopencv_features2d430
-lopencv_flann430
-lopencv_highgui430
-lopencv_imgcodecs430
-lopencv_imgproc430
-lopencv_ml430
-lopencv_objdetect430
-lopencv_photo430
-lopencv_stitching430
-lopencv_video430
-lopencv_videoio430
-o
main
Or you could download VS. Up to you. Hope this helps.

Correction for JackCamichael's answer
those 2 commands won't work in Windows
export CC=/PortableGit/MinGW-w64/mingw32/bin/gcc.exe
export CXX=/PortableGit/MinGW-w64/mingw32/bin/g++.exe
This should be
setx -m CC C:\msys64\mingw64\bin\gcc.exe
setx -m CXX C:\msys64\mingw64\bin\g++.exe
C:\msys64\mingw64\bin is mingw64 path on my machine

Related

Failing to install gtkmm / gtk+

So i am new on the platform, i use c/c++ and i have interests in graph interfaces so i decided to use gtk+and gtkmm with visual studio.
First, i downloaded gtk for the gnome project. I followed all the steps and i got this after a manual run :
\\\ test23.cpp
#include <gtkmm.h>
int main(int argc, char* argv[])
{
Gtk::Main app(argc, argv);
Gtk::Window fenetre;
Gtk::Main::run(fenetre);
return 0;
}
$ g++ -std=c++ test23.cpp $(pkg-config gtkmm-3.0 --cflags --libs | sed 's/ -I/ -isystem /g')
Package gtkmm-3.0 was not found in the pkg-config search path.
Perhaps you should add the directory containing `gtkmm-3.0.pc'
to the PKG_CONFIG_PATH environment variable
No package 'gtkmm-3.0' found
bash: g++: command not found
so i decided to install using vcpkg.
But I failed to install gtk+ and gtkmm after multiple tries.
Here what i've got :
C:\WINDOWS\system32>vcpkg install gtk
Computing installation plan...
The following packages will be built and installed:
gtk[core]:x86-windows
* harfbuzz[core,glib]:x86-windows
* libepoxy[core]:x86-windows
* pango[core]:x86-windows
Additional packages (*) will be modified to complete this operation.
Detecting compiler hash for triplet x86-windows...
Starting package 1/4: libepoxy:x86-windows
Building package libepoxy[core]:x86-windows...
Could not locate cached archive: C:\Users\Manolo97233\AppData\Local\vcpkg\archives\f7\f743ec00b235ca7fd37812284b7d2e09d89b368a.zip
-- Using cached C:/Windows/SysWOW64/vcpkg/downloads/anholt-libepoxy-1.5.4.tar.gz
-- Cleaning sources at C:/Windows/SysWOW64/vcpkg/buildtrees/libepoxy/src/1.5.4-337c486045.clean. Use --editable to skip cleaning for the packages you specify.
-- Extracting source C:/Windows/SysWOW64/vcpkg/downloads/anholt-libepoxy-1.5.4.tar.gz
-- Applying patch libepoxy-1.5.4_Add_call_convention_to_mock_function.patch
-- Using source at C:/Windows/SysWOW64/vcpkg/buildtrees/libepoxy/src/1.5.4-337c486045.clean
-- Acquiring MSYS Packages...
CMake Error at scripts/cmake/vcpkg_execute_required_process.cmake:72 (message):
Command failed: C:/Windows/SysWOW64/vcpkg/downloads/tools/msys2/msys64/usr/bin/bash.exe --noprofile --norc -c "pacman -S --noconfirm --needed pkg-config"
Working Directory: C:/Windows/SysWOW64/vcpkg/downloads/tools/msys2
Error code: 1
See logs for more information:
C:\Windows\SysWOW64\vcpkg\buildtrees\libepoxy\msys-pacman-x86-windows-err.log
Call Stack (most recent call first):
scripts/cmake/vcpkg_acquire_msys.cmake:127 (vcpkg_execute_required_process)
scripts/cmake/vcpkg_configure_meson.cmake:106 (vcpkg_acquire_msys)
ports/libepoxy/portfile.cmake:16 (vcpkg_configure_meson)
scripts/ports.cmake:79 (include)
I believed i had a problem with msys2 so i tried to install it separately following an install kit unsuccessfully.
then i tried to run separately other packages involved with gtkmm like pango and i got this :
//vcpkg install pango
Computing installation plan...
error writing file: C:\Windows\SysWOW64\vcpkg\buildtrees\0.vcpkg_dep_info.cmake: The data is invalid.
Someone explained on a topic it could be an vcpkg accessibility problem for others packages. i modified the accessibility of vcpkg for other packages in my systWOW64'file parameters but it went unsuccessful.
I don't know if vcpkg is problem because i did not remove it and re install it.
I am looking for solutions/options, I am short on ideas, i literally need some help.
Thanks
Use JHbuild, its your friend when building GNOME applications from source !
Run jhbuild build gtkmm --nodeps, the built files will be installed in $HOME/jhbuild/build. To compile your application with it, you must export the environment variable PKG_CONFIG_PATH as follows:
export PKG_CONFIG_PATH="$HOME/jhbuild/build/lib/pkgconfig"
(in the terminal you are running the g++ command from)
And you have to install g++.
I do not know how you are using bash together with visual studio, but
if you have apt, you can install g++ and gtkmm with the following command :
apt install libgtkmm-3.0-dev libgstreamermm-1.0-dev g++
If you have already downloaded gtk, then the downloaded package must contain a .pc file. The directory this file is in when you installed the downloaded package you must add to the PKG_CONFIG_PATH.
Thanks Frederic for your answer it really help me.
I have got an other question. In my quest to add external packages to my C/C++ project, i read differents topics about how to convert static library (.a) into static library (.lib).
I downloaded packages and i wanted to add them manually using their pathway.
I added my packages as followed:
1 / Add the path of the headers in the compiler directories for the project:
-> Project / Properties menu
-> On the left tree, choose Configuration properties - C / C ++ - General
-> On the right table, the first line "Other Include directories": add the directory (s) of your library containing the headers
2 / Add the path of the .lib in the compiler directories for the project:
-> Project / Properties menu
-> On the left tree, choose Configuration Properties - Link Editor - General
-> On the right table, the line "Directory of additional libraries": add the directory (s) of your library containing the .libs
3 / Specify the libraries with which your project is linked:
-> Project / Properties menu
-> On the left tree, choose Configuration Properties - Link Editor - Enter
-> On the right table, the first line "Additional dependencies": add the .lib library (s) with which your project must be linked
But I looked for .lib files to add to my linker input instead, I found .a files and .dll files and I didn't know what to do. . a files and .lib files are almost identical. It seems like .a files are used under linux while .lib are used under windows.
I tried to add the .a files unsuccesfilly. I wondered if I could convert an .a file into a .lib file.
Thanks

MinGW g++/gcc not compiling with proper path and version command working

i changed the path to the correct directory and when I typed "g++ --version" and "gcc --version" I get the version info. I saved simple.cpp into a folder I made in the directory from path and typed "g++ simple.cpp" into the console and it returned with
C:\Users\guede>g++ simple.cpp
g++: error: simple.cpp: No such file or directory
g++: fatal error: no input files
compilation terminated.
This is the code i am trying to compile as a g++ test
#include <iostream>
int main()
{
std::cout << "Hello World!\n";
return 0;
}
g++ simple.cpp
will look for simple.cpp in the current directory. Since You are executing g++ from C:\MinGW\bin, the compiler will only look in C:\MinGW\bin. It will not look in subdirectories. Since simple.cpp is in C:\MinGW\bin\Cpp_prog, the compiler is not looking for the file in the right place. You could
g++ Cpp_prog\simple.cpp
to specify which directory the file is in, but you do not want the compiler output cluttering up GCC's bin directory. Instead, make yourself a "Workspace" directory somewhere else on the computer, could be in your Documents folder or anywhere else convenient. Inside this workspace make a separate directory for each project so you can keep things organized more easily. Execute the compiler from within the appropriate project directory.
Then either invoke the compiler with
C:\MinGW\bin\g++ simple.cpp
or add C:\MinGW\bin to the user path or system path. See Adding directory to PATH Environment Variable in Windows for help on that.
Later you may find yourself with many different compilers and sometimes several versions of GCC. You don't want to have many locations for g++ in the path because it's easy to call the wrong one. When you get to this point you will likely have had to learn about using build automation tools to assist you with complicated projects. Typically you'll specify the compiler location to the automation tool.
Side note:
dir is the DOS /Windows directory listing command. In the command prompt, type dir it will list all of the files in the current directory. If you're going to take up programming, you'll find it very helpful to learn about the various commandline tools provided by your operating system and development environment.

Cannot build Soundpipe DSP library on Windows

I've found the lightweight DSP c library - Soundpipe.
I want to use some filters from it. I do not really need binary files, but the problem is that the repository doesn't have its main header file - soundpipe.h. As I understand, this is because the library uses specific modules ported from Csound and FAUST languages. Also the repository's readme doesn't have installation guide for Windows. It says:
By default, Soundpipe needs libsndfile, and a standard build
environment. Other modules that use other external libraries will need
to be explicitly compiled by modifying config.mk (note: but the Makefile's folder doesn't contain config.mk, there is only config.def.mk)
To compile:
make
sudo make install
Ok, I've downloaded and installed libsndfile.
Then I have tried to use MSVC's nmake - it doesn't work:
makefile(7) : fatal error U1036: syntax error : too many names to left of '='
Stop.
Here is the beginning of the makefile:
> .PHONY: all clean install docs bootstrap
>
> default: all
>
> VERSION = 1.5.0
>
> INTERMEDIATES_PREFIX ?= . PREFIX ?= /usr/local
> ...
After that I've downloaded MinGW, mingw32-make result:
config.mk: No such file or directory
mingw32-make: *** No rule to make target 'config.def.mk', needed by 'config.mk'. Stop.
Ok, when I try to run the configure command, it doesn't work on my Win7 x 64 (sh: ./configure: No such file or directory, etc...)
I have the paths in mt system path variable:
C:\MinGW\bin;C:\MinGW\msys\1.0\bin
Links that I've read:
How Do I Run ./configure with MinGW?
Getting mingw-get to install correctly - mingw/msys path missing plus more!
MinGW's configure doesn't work after reinstalling Git for Windows and GCC_TDM too (No idea how to run configure).
Any ideas how to build the library or at least find missing source files (or make through CSound, Faust, ...) ?
The Soundpipe build system is designed to be used with POSIX environments, of which there are a few to choose from on Windows.
I've been able to build Soundpipe using both MSYS2 and Windows Bash on Windows 10.

Compiling Tensorflow with a custom Clang + Libc++ (instead of stdlibc++)

I am trying to compile tensorflow with a custom clang/llvm toolchain and using clang's native libc++ (instead of borrowing Gcc's stdlibc++).
It looks like bazel plain assumes that every clang will use Gcc's libraries because I get these errors:
$ bazel build --cxxopt=-std=c++11 --cxxopt=-stdlib=libc++ tensorflow:libtensorflow.so
INFO: Found 1 target...
INFO: From Compiling
external/protobuf/src/google/protobuf/compiler/js/embed.cc [for host]:
external/protobuf/src/google/protobuf/compiler/js/embed.cc:37:12:
warning: unused variable 'output_file' [-Wunused-const-variable]
const char output_file[] = "well_known_types_embed.cc";
^
1 warning generated.
ERROR: /home/hbucher/.cache/bazel/_bazel_hbucher/ad427c7fddd5b68de5e1cfaa7cd8c8cc/external/com_googlesource_code_re2/BUILD:11:1: undeclared inclusion(s) in rule '#com_googlesource_code_re2//:re2':
this rule is missing dependency declarations for the following files included by 'external/com_googlesource_code_re2/re2/bitstate.cc':
'/home/hbucher/install/include/c++/v1/stddef.h'
'/home/hbucher/install/include/c++/v1/__config'
I tried to hack into tools/cpp/CROSSTOOL inside bazel as some posts suggested to add the line
cxx_builtin_include_directory: "/home/hbucher/install/include/c++/v1"
but to no avail, it does not seem to make any difference.
Then I tried to follow a bazel tutorial to create a custom toolchain. The text does not help much because they are actually writing a cross tool while what I am trying to do is to tweak the existing host rules and somehow bazel seems to undo every attempt I try to tweak its parameters.
I have got to the point that is currently in my github repository https://github.com/HFTrader/BazelCustomToolchain
However it does not compile and I cannot even figure out how to start debugging this message.
$ bazel build --crosstool_top=#hbclang//:toolchain tensorflow:libtensorflow.so
.....................
ERROR: The crosstool_top you specified was resolved to
'#hbclang//:toolchain', which does not contain a CROSSTOOL file. You can
use a crosstool from the depot by specifying its label.
INFO: Elapsed time: 2.216s
I have appended these lines to my tensorflow/WORKSPACE
new_local_repository(
name="hbclang",
path="/home/hbucher/BazelCustomToolchain",
build_file = "/home/hbucher/BazelCustomToolchain/BUILD",
)
I have asked this question on bazel's google groups but they redirected me to stackoverflow. At this point I am about to give up.
Have someone attempted to do this or I'm breaking ground here?
Thank you.
Solved. Not in the intended way but it works for me.
export INSTALL_DIR="$HOME/install"
export CC=$INSTALL_DIR/bin/clang
export CXX=$INSTALL_DIR/bin/clang++
export CXXFLAGS="-stdlib=libc++ -L$INSTALL_DIR/lib"
export LDFLAGS="-L$INSTALL_DIR/lib -lm -lrt"
export LD_LIBRARY_PATH="/usr/lib:/lib/x86_64-linux-gnu/:$INSTALL_DIR/lib"
git clone https://github.com/tensorflow/tensorflow.git tensorflow-github
cd tensorflow-github
mkdir build-tmp && cd build-tmp
cmake ../tensorflow/contrib/cmake/
make -j4
Easy as 1-2-3 with cmake
[2020-05-24: Edit to make the answer up to date.]
TLDR: To build a project with Bazel with a specific Clang binary, and with libc++, this works for me (where INSTALL_DIR is where I've installed llvm):
CC="$INSTALL_DIR/bin/clang" \
BAZEL_CXXOPTS="-stdlib=libc++:-isystem$INSTALL_DIR/include" \
BAZEL_LINKOPTS="-stdlib=libc++" \
BAZEL_LINKLIBS="-L$INSTALL_DIR/lib:-Wl,-rpath,$INSTALL_DIR/lib:-lc++:-lm" \
bazel test //...
Background:
You can use --repo_env option, e.g. --repo_env=CC=clang, to put these defaults into your project- or system-wide .bazelrc.
This approach uses Bazel's C++ toolchain autoconfiguration which doesn't attempt to declare all the toolchain inputs in BUILD files. This is to simplify the configuration for the user. Therefore whenever you modify the C++ toolchain in a way that Bazel cannot know about (rebuild llvm etc.), you have to run bazel clean --expunge to flush the cache and rerun the autoconfiguration the next time.
The robust solution to specifying C++ toolchain in Bazel is to use the CcToolchainConfigInfo. See the documentation at https://docs.bazel.build/versions/master/tutorial/cc-toolchain-config.html and https://docs.bazel.build/versions/master/cc-toolchain-config-reference.html.

Build folder and makefile

The question comes from my puzzlement when compiling a makefile for Deep Learning framework Caffe on Ubuntu, but it relates, I believe, to a more general phenomenon of the nature of compiling a C++ makefile.
After "make all", the resulting files from the compilation were put in a hidden folder: .build_release, not in the respective folders where the cpp files are.
Then when I tried to run the following lines:
./data/mnist/get_mnist.sh
./examples/mnist/create_mnist.sh
I was getting an error that the system does not find the file:
./create_mnist.sh: 16: ./create_mnist.sh: build/examples/mnist/convert_mnist_data.bin: not found
But the file actually existed in the .build_release folder.
What happened and how to fix this problem?
The issue is not with make, you simply need to follow the instructions carefully. The BUILD_DIR is specified by Makefile.config. By default this folder is named build. Once you followed the compilation instructions:
cp Makefile.config.example Makefile.config
# Adjust Makefile.config (for example, if using Anaconda Python)
make all
make test
make runtest
Navigate to build:
cd build
./data/mnist/get_mnist.sh
./examples/mnist/create_mnist.sh