I've built opencv 3.0 from source and can run a few sample apps, build against the headers ok so I presume it's installed successfully.
I'm also using python3 and I now go to install and build caffe. I set a few variables in Makefile.config as I'm using the CPU due to having an AMD GPU and also Anaconda.
When I run make all I get this error:
$ make all
CXX/LD -o .build_release/examples/cpp_classification/classification.bin
/usr/bin/ld: .build_release/examples/cpp_classification/classification.o: undefined reference to symbol '_ZN2cv6imreadERKNS_6StringEi'
//usr/local/lib/libopencv_imgcodecs.so.3.0: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
Makefile:565: recipe for target '.build_release/examples/cpp_classification/classification.bin' failed
make: *** [.build_release/examples/cpp_classification/classification.bin] Error 1
from searching I think this is something to do with using openCV 3 but I'm not sure where to start looking for a solution. Any help?
And yes I'm one of the horde of inexperienced users looking to fiddle with the Google Inception learning technique.
You can also add the opencv_imgcodecs to the MakeFile in line 187, see this pull.
It could be that you are using OpenCV version 3. If yes just uncomment the following line in your Makefile.config:
# OPENCV_VERSION := 3
So it will look like
OPENCV_VERSION := 3
You could verify the version currently in use by doing:
$ python
>>> import cv2
>>> cv2.__version__
'3.1.0-dev'
The problem report is very clear. There is a problem with linking library libraries.The reason may be the difference between 3.0 and 2.x.
You need to add
opencv_core opencv_highgui opencv_imgproc opencv_imgcodecs
into LIBRARIES +=.
I used cmake instead with the -DBUILD_TIFF=ON flag and got a successful build.
You can edit Makefile.config with the following 2 lines like this and it worked for me. Note that your opencv path must be set before default path!
INCLUDE_DIRS := $(PYTHON_INCLUDE) /home/young/Soft/openCV-3.3.1/include \
/usr/local/include /usr/include/hdf5/serial
LIBRARY_DIRS := $(PYTHON_LIB) /home/young/Soft/openCV-3.3.1/lib \
/usr/local/lib /usr/lib /usr/lib/x86_64-linux-gnu /usr/lib/x86_64-linux-gnu/hdf5/serial
A quick workaround is to add -lopencv_imgcodecs flag when you're compiling your code.
This worked for me:
g++ test.cpp -o test <Some flags> -lopencv_imgcodecs
Related
Trying to install FFTW and feel like I have been going in circles. I need it for an R package (poisbinom) and I think I have it installed right (I'm on a cluster and don't have sudo privileges. I followed the instructions here: http://micro.stanford.edu/wiki/Install_FFTW3 )
The problem I'm running into is this:
g++ -std=gnu++14 -shared -L/usr/local/lib64 -o poisbinom.so RcppExports.o init.o poisbinom.o -lfftw3 -lm
/usr/bin/ld: cannot find -lfftw3
collect2: error: ld returned 1 exit status
I have a feeling it is because I did not install FFTW under usr/local/lib64, its under $HOME/usr/. My R install is under $HOME/R/lib64/R. I am not sure what variable to change to make it look in the right place, as I cant change this g++ call as it is part of an R install.packages call. I've been going in circles with this for two days now. Let me know if theres anything else I can provide to help.
I already ran install.packages("fftw") and that was successful and used LFFTW3 without issue because it looked in the right place. I mostly just need to know if I can point this gcc call to the right folder with an environment variable or something.
In case you have pkg-config on your system and your installation produced the required configuration files (e.g. fftw3.pc): Define the environment variable PKG_CONFIG_PATH to (also) include the directory with these configuration files for FFTW3.
Otherwise, you can define FFTW_CFLAGS and FFTW_LIBS:
fftw$ ./configure --help
[...]
FFTW_CFLAGS C compiler flags for FFTW, overriding pkg-config
FFTW_LIBS linker flags for FFTW, overriding pkg-config
[...]
I ended up having to add the variables to my bashrc file:
export PATH=$HOME/fftw_folder/bin:$PATH
export LD_LIBRARY_PATH=$HOME/fftw_folder/lib:$LD_LIBRARY_PATH
export LIBRARY_PATH=$HOME/fftw_folder/lib:$LIBRARY_PATH
export PKG_CONFIG_PATH=$HOME/_folder/lib/pkgconfig:$PKG_CONFIG_PATH
As laid out here: http://hpc.loni.org/docs/faq/installation-details.php
And now things seem to be moving forward smoothly. Thanks for the help!
I'd like to learn how to write PNG images pixel-by-pixel using both RGB and HSV color models with C++. I read that this should be fairly easy using PNGwriter (https://github.com/pngwriter/pngwriter), but I've spent many hours struggling with installing it (on Ubuntu) and compiling my code with it. Any help would be much appreciated.
Disclaimer: I have a weird background in the sense that I have many years of experience in using Unix-like operating systems, doing stuff in the terminal, and writing code, but I know little/nothing about installing software from the source code or compiling programs manually or with makefiles from multiple source code files.
The installation instructions on GitHub advise to do one of the following:
Spack:
spack install pngwriter
spack load pngwriter
From Source:
First install the dependencies zlib, libpng, and (optional for text support) freetype. PNGwriter >can then be installed using CMake:
git clone https://github.com/pngwriter/pngwriter.git
mkdir -p pngwriter-build
cd pngwriter-build
# for own install prefix append: -DCMAKE_INSTALL_PREFIX=$HOME/somepath
cmake ../pngwriter
make -j
# optional
make test
# sudo is only required for system paths
sudo make install
I managed to install Spack and then PNGwriter, but couldn't compile the simplest program with it and wasn't able to figure out why. I then installed PNGwriter manually, but still couldn't compile anything with it. This was many hours of struggling ago so I, unfortunately, don't remember what kind of errors or problems I was encountering at this point.
The instructions on GitHub say the following about linking:
First set the following environment hint if PNGwriter was not installed in a system path:
# optional: only needed if installed outside of system paths
export CMAKE_PREFIX_PATH=$HOME/somepath:$CMAKE_PREFIX_PATH
Use the following lines in your projects CMakeLists.txt:
find_package(PNGwriter 0.7.0)
if(PNGwriter_FOUND)
target_link_libraries(YourTarget PRIVATE PNGwriter::PNGwriter)
endif(PNGwriter_FOUND)
Questions: How do I know if PNGwriter was or wasn't installed in a system path? I have libPNGwriter.a in /usr/local/lib and pngwriter.h in /usr/local/include --- does this mean that it was installed in a system path? When installing I simply tried to follow the instructions above. Do I just type the environment hint to the terminal or add it to some file? If the former then does it need to be given every time I open a new terminal session? Is "somepath" /usr/local/lib, /usr/local/include or something else?
Questions: Does the second part regarding "CMakeLists.txt" depend on whether PNGwriter was installed in a system path? What is "CMakeLists.txt"? I assume it's some file one's IDE creates, but my NetBeans projects don't seem to contain such files. What if I have a single source file and compile it manually in the terminal?
Now, let's say I'd like to compile the PNGwriter quickstart example:
#include <pngwriter.h>
int main()
{
int i;
int y;
pngwriter png(300, 300, 0, "test.png");
for(i = 1; i ≤ 300; i++)
{
y = 150 + 100*sin((double)i*9/300.0);
png.plot(i, y, 0.0, 0.0, 1.0);
}
png.close();
return 0;
}
The PNGwriter manual instructs to compile as
g++ myprogram.cc -o my_program `freetype-config --cflags` -I/usr/local/include -L/usr/local/lib -lpng -lpngwriter -lz -lfreetype
but I get errors
$ g++ example.cpp -o example `freetype-config --cflags` -I/usr/local/include -L/usr/local/lib -lpng -lpngwriter -lz -lfreetype
Package freetype2 was not found in the pkg-config search path.
Perhaps you should add the directory containing `freetype2.pc'
to the PKG_CONFIG_PATH environment variable
Package 'freetype2', required by 'virtual:world', not found
Package freetype2 was not found in the pkg-config search path.
Perhaps you should add the directory containing `freetype2.pc'
to the PKG_CONFIG_PATH environment variable
Package 'freetype2', required by 'virtual:world', not found
Package freetype2 was not found in the pkg-config search path.
Perhaps you should add the directory containing `freetype2.pc'
to the PKG_CONFIG_PATH environment variable
Package 'freetype2', required by 'virtual:world', not found
Package freetype2 was not found in the pkg-config search path.
Perhaps you should add the directory containing `freetype2.pc'
to the PKG_CONFIG_PATH environment variable
Package 'freetype2', required by 'virtual:world', not found
sed: -e expression #1, char 0: no previous regular expression
sed: -e expression #1, char 0: no previous regular expression
In file included from example.cpp:1:0:
/usr/local/include/pngwriter.h:66:22: fatal error: ft2build.h: No such file or directory
compilation terminated.
The answer to this Stack Overflow question (Trying to install pygame on ubuntu which gives error) suggests to install libfreetype6-dev, but I apparently have the latest version already whereby the errors remain unchanged. If I instead actually add the directory containing freetype2.pc (I found it by going to / and using find -name "freetype2.pc") to the environment variable (added export PKG_CONFIG_PATH="/usr/lib/x86_64-linux-gnu/pkgconfig/:$PKG_CONFIG_PATH to ~/.bashrc and did source ~/.bashrc) then I get new errors
$ g++ example.cpp -o example `freetype-config --cflags` -I/usr/local/include -L/usr/local/lib -lpng -lpngwriter -lz -lfreetype
/usr/bin/ld: cannot find -lpngwriter
collect2: error: ld returned 1 exit status
Here I figured out that I needed to replace -lpngwriter with -lPNGwriter (i.e., the manual is erroneous). Then I get:
$ g++ example.cpp -o example `freetype-config --cflags` -I/usr/local/include -L/usr/local/lib -lpng -lPNGwriter -lz -lfreetype
/usr/bin/ld: /usr/local/lib/libPNGwriter.a(pngwriter.cc.o): undefined reference to symbol 'png_set_sig_bytes##PNG12_0'
/usr/lib/gcc/x86_64-linux-gnu/5/../../../x86_64-linux-gnu/libpng.so: error adding symbols: DSO missing from command line
collect2: error: ld returned 1 exit status
This is where I'm currently stuck. I can't seem to find a solution by googling (at least one that I'd understand).
Question: How do I get this working? How do I get it working in NetBeans? Do I get these problems because I effed up the linking step above?
Edit1: As per john's comment, I tried swapping -lpng and -lPNGwriter and I again get new errors:
$ g++ example.cpp -o example `freetype-config --cflags` -I/usr/local/include -L/usr/local/lib -lPNGwriter -lpng -lz -lfreetype
/usr/local/lib/libPNGwriter.a(pngwriter.cc.o): In function `pngwriter::close()':
pngwriter.cc:(.text+0x41c2): undefined reference to `png_convert_to_rfc1123_buffer'
/usr/local/lib/libPNGwriter.a(pngwriter.cc.o): In function `pngwriter::read_png_info(_IO_FILE*, png_struct_def**, png_info_def**)':
pngwriter.cc:(.text+0x4fdf): undefined reference to `png_set_longjmp_fn'
collect2: error: ld returned 1 exit status
I'm still left clueless.
Thanks to john, I think I got it figured out. My guess is that the installation from the source (after the Spack installation) messed things up somehow. I reinstalled PNGwriter using Spack and, now apparently having all the pieces for the compilation command, was finally able to compile the example code.
Summary:
Source Spack
# For bash/zsh/sh
$ . spack/share/spack/setup-env.sh
# For tcsh/csh
$ source spack/share/spack/setup-env.csh
# For fish
$ . spack/share/spack/setup-env.fish
Install using Spack (skip if already installed)
spack install pngwriter
Load PNGwriter (I guess one needs to do this in every new terminal session)
spack load pngwriter
Compile (note that this isn't quite what the PNGwriter manual suggests)
g++ example.cpp -o example `freetype-config --cflags` -I/usr/local/include -L/usr/local/lib -lPNGwriter -lpng -lz -lfreetype
I'm getting an ld error when attempting to compile an sfml program on ubuntu 16.04. This is apparently a known issue, and there is supposed to be a workaround, but I don't understand what is it...
http://web.archive.org/web/20160509014317/https://gitlab.peach-bun.com/pinion/SFML/commit/3383b4a472f0bd16a8161fb8760cd3e6333f1782.patch
The error spat out by ld is
hidden symbol `__cpu_model' in /usr/lib/gcc/x86_64-linux-gnu/4.9/libgcc.a(cpuinfo.o) is referenced by DSO
There is no relevant code to this - as I understand it this error is produced on all ubuntu 16.04 systems with g++ 5, if the program to be linked contains objects such as sf::Texture and sf::Sprite. (I don't know any more detail than this.)
I have tried also compiling with g++ 4.9, but the same error occurs.
My compile line is g++-4.9 --std=c++11 -Wall main.cpp -lsfml-graphics -lsfml-window -lsfml-system -o a.out
Has anyone else experienced this error and resolved it successfully?
I've had to fix this issue several times. Instead of applying the patch, you can manually fix it by editing the file SFML/src/SFML/Graphics/CMakeLists.txt. At line 149, you will find the following:
if(SFML_COMPILER_GCC)
set_source_files_properties(${SRCROOT}/ImageLoader.cpp PROPERTIES COMPILE_FLAGS -fno-strict-aliasing)
endif()
After the endif(), insert the following:
if(SFML_COMPILER_GCC AND BUILD_SHARED_LIBS)
list(APPEND GRAPHICS_EXT_LIBS "-lgcc_s -lgcc")
endif()
Then, in the top-level SFML folder, run the following:
mkdir build && cd build
cmake .. -DSFML_BUILD_EXAMPLES=ON -DSFML_BUILD_DOCS=ON
make
sudo make install
sudo ldconfig
This will get it built and installed without the compiler error. (Note: Remove the -D flags from cmake if you don't want docs or examples)
I ran this in the SFML source directory before running the standard cmake...make:
curl https://gitlab.peach-bun.com/pinion/SFML/commit/3383b4a472f0bd16a8161fb8760cd3e6333f1782.patch \
| patch -p1
and that solved the problem
I've got the same linker error when trying to build SFML 2.4.2 with examples, specifically with opengl and shader ones.
Inspired by the #Joshua solution, I tried to change the compiler from GCC to Clang. It worked.
I am sharing here because it seems to be a simpler solution if you have no restrictions to use Clang.
Just download SFML and change to its directory. And...
mkdir build && cd build
cmake .. -DSFML_BUILD_EXAMPLES=ON -DCMAKE_CXX_COMPILER=clang++
make
sudo make install
sudo ldconfig
Removing -fvisibility=hidden from compiler options worked for me.
I'm used to R but I'm a complete beginner in C++. I'm having a hard time installing GSL to use it with RcppGSL in R in order to refine a package comprising C++ code. I downloaded the latest GSL version, unpacked it to C:/GSL, set the environmental variable LIB_GSL to C:/GSL and the PATHtoC:/GSL/bin.
Like I said, I want to refine an existing package. I've forked and pulled the latest version of the package in question from github. My primary goal is simply to be able to use and refine the package. However, when I'm trying to make a source package out of the source files I get the following Error:
Installing to library 'C:/Users/Simon/Documents/R/win-library/3.1'
* installing *source* package 'Rflim' ...
** libs
g++ -m64 -I"C:/PROGRA~1/R/R-31~1.2/include" -DNDEBUG -I"C:/Users/Simon/Documents/R/win-library/3.1/Rcpp/include" -I"d:/RCompile/CRANpkg/extralibs64/local/include" -IC:/Users/Simon/Documents/R/win-library/3.1/Rcpp/include -I/usr/local/include -I C:/Users/Simon/Documents/R/win-library/3.1/RcppGSL/include -O2 -Wall -mtune=core2 -c flim.cpp -o flim.o
In file included from C:/Users/Simon/Documents/R/win-library/3.1/RcppGSL/include/RcppGSL.h:23:0,
from flim.cpp:1:
C:/Users/Simon/Documents/R/win-library/3.1/RcppGSL/include/RcppGSLForward.h:26:29: fatal error: gsl/gsl_vector.h: No such file or directory
compilation terminated.
make: *** [flim.o] Error 1
Warnung: Ausf�hrung von Kommando 'make -f "Makevars" -f "C:/PROGRA~1/R/R-31~1.2/etc/x64/Makeconf" -f "C:/PROGRA~1/R/R-31~1.2/share/make/winshlib.mk" SHLIB_LDFLAGS='$(SHLIB_CXXLDFLAGS)' SHLIB_LD='$(SHLIB_CXXLD)' SHLIB="Rflim.dll" WIN=64 TCLBIN=64 OBJECTS="flim.o"' ergab Status 2
ERROR: compilation failed for package 'Rflim'
* removing 'C:/Users/Simon/Documents/R/win-library/3.1/Rflim'
Exited with status 1.
Sorry, there are a couple of german words in there:
Warnung: Ausf�hrung von Kommando==Warning:executing command
What I realize is that R is trying to access the GSL library via the RcppGSL path. Is that correct? Shouldn't it go directly to C:/GSL instead? I also realized that R is trying to go to paths from both Rcpp as well as RcppGSL and I don't see why...
I would appreciate any clarification or suggestions on how to solve the problem.
Thank you!!!
Please install the prebuild GSL version for R on Windows from the support site provided by Prof Ripley and then set these values accordingly in src/Makevars.win
## This assumes that the LIB_GSL variable points to working GSL libraries
PKG_CPPFLAGS=-I$(LIB_GSL)/include -I../inst/include
PKG_LIBS=-L$(LIB_GSL)/lib -lgsl -lgslcblas
either directly, or set LIB_GSL accordingly.
This is how CRAN builds the package, and it evidently works.
The packages I'm toying with here are rather unknown, but nevertheless the problem is rather generic. Basically, I'm trying to compile Python module (called rql) with C++ extension. The extension uses external framework called gecode, which contains several libraries. I compiled gecode and installed locally. Now, let the output speak for itself:
red#devel:~/build/rql-0.23.3$ echo $LD_LIBRARY_PATH
/home/red/usr/lib
red#devel:~/build/rql-0.23.3$ ls $LD_LIBRARY_PATH | grep libgecodeint
libgecodeint.so
libgecodeint.so.22
libgecodeint.so.22.0
red#devel:~/build/rql-0.23.3$ python setup.py build
running build
running build.py
package init file './test/__init__.py' not found (or not a regular file)
running build_ext
building 'rql_solve' extension
g++ -pthread -shared build/temp.linux-i686-2.5/gecode-solver.o -lgecodeint -lgecodekernel -lgecodesearch -o build/lib.linux-i686-2.5/rql_solve.so
/usr/bin/ld: cannot find -lgecodeint
collect2: ld returned 1 exit status
error: command 'g++' failed with exit status 1
LD_LIBRARY_PATH is for runtime linker/loader (same effect could be achieved with ldconfig ). What you need is the -L flag:
-L/home/red/usr/lib
on the compiler command line.
Edit:
And - thanks to #bjg for reminding me - you can use LIBRARY_PATH if you don't want to mess with compiler options.
You've apparently modified LD_LIBRARY_PATH to point to a non-standard location in your home directory. Do you know if LD_LIBRARY_PATH in the environment used to call g++ in setup.py matches your shell's environment?
See if you can pass arguments to setup.py to modify the library path or simply pass -L/home/red/usr/lib to g++.