Problem with interface Python and Cpp with OpenCV and Cuda - python

I'm facing an issue while interfacing python and cpp. I need to compute an optical flow with the GPU. To do so, I'm using the tvl1 algorithm with opencv on with cpp and cuda.
My cpp code is working well with GPU and I now want to call the function in my python code because there isn't a way to compute optical flow on GPU with this algo on Python.
I'm using ctypes to make the link beetwin cpp and python. Here are my codes :
python
from ctypes import cdll
opencvfunc = cdll.LoadLibrary('opt_flow_lib/test.so')
cpp
//test.cpp
#define DLLEXPORT extern "C"
#include "opencv2/xfeatures2d.hpp"
#include "opencv2/cudaarithm.hpp"
#include "opencv2/cudaoptflow.hpp"
#include "opencv2/cudacodec.hpp"
#include "opencv2/video/tracking.hpp"
#include "opencv2/imgproc/imgproc.hpp"
#include "opencv2/highgui/highgui.hpp"
#include "opencv2/core/mat.hpp"
using namespace cv::cuda;
using namespace cv;
DLLEXPORT int tvl(int a, int b) {
int bound = 20;
Size new_size(224, 224);
Mat curr_gray, prev_gray, prev_image, curr_image;
Mat flow_x, flow_y;
GpuMat d_frame_0, d_frame_1;
GpuMat d_flow;
cv::Ptr<cuda::OpticalFlowDual_TVL1> alg_tvl1 =
cuda::OpticalFlowDual_TVL1::create();
return 3;
}
Here is how I build my cpp file :
gcc -o test.so -shared -fPIC test.cpp `pkg-config opencv --cflags --libs`
Here is what I have when I launch my python code :
OSError: opt_flow_lib/test.so: undefined symbol: _ZN2cv4cuda20OpticalFlowDual_TVL16createEdddiididdb
When I remove this line from cpp :
cv::Ptr<cuda::OpticalFlowDual_TVL1> alg_tvl1 =
cuda::OpticalFlowDual_TVL1::create();
Everything is going ok, I have my 3 int returned. But it seems that cuda doesn't want to do its job
Is someone who had ever turned opencv with cuda with ctypes and cpp/python could hepl me ?
Thanks you
EDIT :
Maybe it could be interesting to have my $ pkg-config opencv --cflags --libs :
-I/usr/local/include/opencv4/opencv2 -I/usr/local/include/opencv4 -L/usr/local/lib -lopencv_core -lopencv_imgproc -lopencv_highgui -lopencv_ml -lopencv_video -lopencv_xfeatures2d -lopencv_calib3d -lopencv_objdetect

Related

PyBind11: simple invoke task, standard C++ library not found

i am trying to wrap the C++ library PyrochloreAFM.hpp which itself uses the lib boost/randoom.hpp so that i can import it as a python module. The C++ part itself works fine and i can succesfully import and use all of this from my main.
#include "PyrochloreAFM.hpp"
int main (int argc, const char *argv[]) {
PyrochloreAFM pyrochloreAFM (&parameters, &statistics, &rng);
}
Now following a tutorial i set up my c++ wrapper:
// pybind11_wrapper.cpp
#include <pybind11/pybind11.h>
#include <PyrochloreAFM.hpp>
PYBIND11_MODULE(pybind11_example, m) {
m.doc() = "pybind11 example plugin"; // Optional module docstring
m.def("cpp_function", &PyrochloreAFM, "A function that multiplies two numbers");
}
and my tasks.py file
# tasks.py
import invoke
invoke.run(
"g++ -O3 -Wall -Werror -shared -std=c++11 -fPIC PyrochloreAFM.cpp "
"-o libpyro.so "
)
However now $invoke build-PyrochloreAFM or even $invoke --list seem to have lost the track of the standard C++ library.
In file included from PyrochloreAFM.cpp:1:
./Parameters.hpp:16:10: fatal error: 'boost/random.hpp' file not found
#include "boost/random.hpp" // tested with boost 1.53
^~~~~~~~~~~~~~~~~~
1 error generated.
This might be a simple PATH problem so i would be very glad for any tips!
Thank you, Andres!
"Of course" the tasks.py file which compiles the .cpp file through the python module invoke has to instruct the compiler g++ to use the libraries we desire. In my case it is the flag -I /opt/homebrew/Cellar/boost/1.80.0/include i also use in my makefile or directly in terminal.

Cython Help: Skipping Incompatible Library

I have a small cython module called deLorean.pyx
cdef public struct Vehicle:
int speed
float power
cdef public api void activate(int v):
print "Time travel achieved at " + str(v) + " mph."
I also have a setup.py file that looks like this:
from distutils.core import setup
from Cython.Build import cythonize
setup(name = 'First try', ext_modules = cythonize(["deLorean.pyx"]),)
When I go to compile the cython code using this: cython deLorean.pyx
This will then generate *.h, *.c, and *_api.h files.
I also have a c program called marty.c that looks like this:
#include "Python.h"
#include "deLorean_api.h"
#include <stdlib.h>
struct Vehicle car;
int main(int argc, char** agrv){
printf("HELLO");
Py_Initialize();
import_deLorean();
car.speed = 33;
car.power = 12.3;
printf("speed: %d, power: %f", car.speed, car.power);
activate(12);
Py_Finalize();
return 0;
}
I then compile the entire module using this:
gcc -fPIC -L/usr/lib -I/usr/local/include/python2.7 -lpython2.7 deLorean.c marty.c -o deLorean -g
This compiles with these notes:
/usr/bin/ld: skipping incompatible /usr/lib/libc.so when searching for -lc
/usr/bin/ld: skipping incompatible /usr/lib/libc.a when searching for -lc
This creates an a.out file; however, it seg faults when run.
When I run it in gdb, this is the output:
(gdb) r
Starting program: /root/Asta/Cython-0.22.1/deLorean
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/lib64/libthread_db.so.1".
Program received signal SIGSEGV, Segmentation fault.
0x0000000000000000 in ?? ()
(gdb)
I have played around with marty.c and I have narrowed down the culprit to when the activate() function is called. Is there something I have overlooked? What could possible be causing this behavior?

Compile c++ Programm with CMake that uses Boost/python.hpp

I'm currently trying to compile a c++ file using CMake.
But since I'm using Boost::python it won't compile.
I set up a little test file to figure out what I need to do but I just can't get it to work.
Any Help would be greatly appreciated!!
The test file:
#include <Python.h>
#include <boost/python.hpp>
#include <iostream>
using std::cout;
using std::endl;
int main()
{
namespace py = boost::python;
Py_Initialize();
// Retrieve the main module's namespace
py::object global(py::import("__main__").attr("__dict__"));
py::exec("print 'Hello from Python!' \n", global, global);
return 0;
}
It will compile if I just use,
clang++ -I/usr/include/python2.7 -lpython2.7 -lboost_python -std=c++11 boosttest.cpp -o boosttest
I tried this CMakeLists.txt to get it to work.
cmake_minimum_required(VERSION 3.2)
FIND_PACKAGE(PythonLibs)
FIND_PACKAGE(Boost)
INCLUDE_DIRECTORIES(${Boost_INCLUDE_DIRS} ${PYTHON_INCLUDE_DIRS})
LINK_LIBRARIES(${Boost_LIBRARIES} ${PYTHON_LIBRARIES})
add_executable(Test1 boosttest.cpp)
target_link_libraries(Test1 ${Boost_LIBRARIES} ${PYTHON_LIBRARIES})
and what I get is
undefined reference to `boost::python::import(boost::python::str)'
and a couple more of the same category.
Thanks for your help Mark, thanks to the new errors after I included
find_package(Boost REQUIRED python)
I was able to figure out that the problem was that CMake selected the libs for python 3.4 but Boost was build against 2.7.
So the Solution was to include the version as so:
FIND_PACKAGE(PythonLibs 2.7 REQUIRED)
Did you try
find_package(Boost REQUIRED python)
also run with verbosity to see what is going on
cmake . --debug-output
make VERBOSE=1

SWIG Python undefined symbol error

I'm trying to create a *.so file for further use in Python using SWIG, but something isn't working.
I have two files:
DataGatherer.h
#include <iostream>
#include <stdlib.h>
#include <time.h>
#include "gnublin.h"
#include <pthread.h>
class dataGatherer
{
private:
int threshold;
int timeThreshold;
int data[4096];
bool running;
gnublin_spi* spiDevice;
pthread_t spiThread;
void *params;
public:
dataGatherer(void);
dataGatherer(int, int);
void initData();
int getThreshold(void);
int* getData(void);
int getPeak(void);
void initSPI(void);
void gatherData();
void * run(void * arg);
void stop(void);
// for use of thread we have to implement some methods from C
static void * start_static(void * params)
{
dataGatherer * thread_this = static_cast<dataGatherer*>(params);
return thread_this->run(thread_this->params);
}
void start(void * params)
{
this->params = params;
pthread_create(&spiThread, 0, &dataGatherer::start_static, this);
}
};
and spiController.h
#include "dataGatherer.h"
class spiController
{
private:
bool runGather;
dataGatherer* gatherer;
int data[4096];
public:
spiController(void);
spiController(int, int);
void initData();
bool checkStop();
void stop();
void start();
};
My spiController.i interface file looks like this:
/* spiController.i */
%module spiController
%{
#include "dataGatherer.h"
#include "spiController.h"
#include "gnublin.h"
#include <stdlib.h>
#include <time.h>
#include <pthread.h>
extern void initData();
extern bool checkStop();
extern void stop();
extern void start();
%}
extern void initData();
extern bool checkStop();
extern void stop();
extern void start();
At the end I try to create the *.so file using commands in the terminal like in the example on the SWIG page with:
swig -python -c++ spiController.i
c++ -c spiController_wrap.c -I/usr/include/python2.7
c++ -shared spiController_wrap.o -o _spiController.so
*.cxx, *.o and *.so file are created with no error, but when I import the spiController into the python code I get:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "spiController.py", line 26, in <module>
_spiController = swig_import_helper()
File "spiController.py", line 22, in swig_import_helper
_mod = imp.load_module('_spiController', fp, pathname, description)
ImportError: ./_spiController.so: undefined symbol: _Z9checkStopv
It's my first try using SWIG and I'm already stuck at this point. How can I resolve this?
I just got the same error and finally figured out why. As above people said, when it says unfound symbol like yours and gives the undefined function name '_Z9checkStopv', always check for the implementation of this function in .cpp file as well as any function declaration of the same name!!
For my case, my cpp does define my 'unfound symbol' constructor function, but in my .h file, i have an overloaded operator= (for the constructor) which is undefined in .cpp file. So swig wraps both default constructor(implemented in .cpp) and operator= (not implemented). Therefore when import, this unimplemented operator= produces the error. Hope this helps!
You must link the library that defines your C++ functions that you have declared like checkStop etc. You would add -L<path to your C++ DLL> -l<name of your C++ DLL> on 3rd line of your example's compile steps.
Like this:
c++ -L<path to DLL> -l<name of your dll> -shared spiController_wrap.o -o _spiController.so
Just as Adam's comment and my experience, you should first compile your XXX.cpp file into XXX.o, the whole command lines maybe like following:
swig -python -c++ XXX.i
g++ -c -fpic XXX.cpp* (this command will generate XXX.o file)
g++ -c -fpic XXX_wrap.cxx -I/usr/include/python2.7* (this command will generate XXX_wrap.o file)
g++ -shared XXX.o XXX_wrap.o -o XXX.so
Although there may be many causes for this problem, I got the exact same error when I compiled the shared library with the python v3.5 headers, e.g.
swig -python example.i
gcc -fPIC -c example.c example_wrap.c -I/usr/include/python3.5 # <-- ISSUE HERE
gcc -shared example.o example_wrap.o -o _example.so
But then later tried to use the example library using python test.py, which ran python v2.7 on system (so it was a python version mismatch issue).
In my case I was also getting that error, and spent some time trying out stuff without success. My problem was that although my source file was plain C I had it named with the .cpp extension, assuming it wouldn't matter. Changing the extension to .c solved automatically the issue.
Another way of solving it was to add the line #include "example.cpp" to the header section of SWIG's .i file.
So, summarizing:
example.c
int fact(int n) {
if (n <= 1) return 1;
else return n*fact(n-1);
}
example.i
%module example
%{
extern int fact(int n);
%}
extern int fact(int n);
Then the following worked for me (in Ubuntu 17.10):
swig -c++ -python example.i
gcc -fPIC -c example.c example_wrap.c -I /usr/include/python2.7
gcc -shared example.o example_wrap.o -o _example.so
python -c "import example; print example.fact(5)"
Hope this helps someone!
Cheers
I know the solution
At the end of make up the share,you should usr g++ -shared -fpic *.o *.o -o _***.so

Building a DLL with MinGW and loading it with Python's ctypes module

I am trying to build a C DLL which can be loaded within python using ctypes.windll.loadlibrary(...)
I can create a DLL and a client program all in C which work following the MinGW tutorial at http://www.mingw.org/wiki/MSVC_and_MinGW_DLLs.
When I try to load the same DLL within python I get an error:
OSError: [WinErrror 193] %1 is not a valid Win32 application
Can someone give me some idea as to what I am doing incorrectly?
Here are the files:
noise_dll.h
#ifndef NOISE_DLL_H
#define NOISE_DLL_H
// declspec will identify which functions are to be exported when
// building the dll and imported when 'including' this header for a client
#ifdef BUILDING_NOISE_DLL
#define NOISE_DLL __declspec(dllexport)
#else
#define NOISE_DLL __declspec(dllimport)
#endif
//this is a test function to see if the dll is working
// __stdcall => use ctypes.windll ...
int __stdcall NOISE_DLL hello(const char *s);
#endif // NOISE_DLL_H
noise_dll.c
#include <stdio.h>
#include "noise_dll.h"
__stdcall int hello(const char *s)
{
printf("Hello %s\n", s);
return 0;
}
I build the DLL with:
gcc -c -D BUILDING_NOISE_DLL noise_dll.c
gcc -shared -o noise_dll.dll noise_dll.o -Wl,--out-implib,libnoise_dll.a
The python code is simply:
import ctypes
my_dll = ctypes.windll.LoadLibrary("noise_dll")
and I get the error above: '%1 is not not a valid Win32 application'
I know the DLL is not completely wrong, because if i create a client file:
noise_client.c
#include <stdio.h>
#include "noise_dll.h"
int main(void)
{
hello("DLL");
return 0;
}
and build with:
gcc -c noise_client.c
gcc -o noise_client.exe noise_client.o -L. -lnoise_dll
I get a working executable. I have some understanding of all of what goes on in the code, options and preprocessor directives above, but am still a little fuzzy on how the .dll file and the .a file are used. I know if I remove the .a file I can still build the client, so I am not even sure what its purpose is. All I know is that it is some kind of archive format for multiple object files
I can ctypes.windll.loadlibrary(...) an ordinary windows DLL found in the windows/system32 without an issue.
One final point:
I am using 64 bit python 3.3. I am using the version of minGW tat comes with the recommended installer (mingw-get-inst-20120426.exe). I am not sure if it is 32 bit, or if that matters.
Thanks!
Use 32-bit python - your DLL probably wasn't compiled as 64 bit code.

Categories