takuya-takeuchi / OpenPoseDotNet

OpenPose wrapper written in C++ and C# for Windows
MIT License
66 stars 25 forks source link

Can't Compile OpenPoseDotNet.Native #6

Closed malhosary closed 4 years ago

malhosary commented 4 years ago

Dears, I just got the latest version of the code and couldn't build OpenPoseDotNet.Native. I was able to build OpenPose.dll successfully. OS: Windows 10 64-bit VS: 2017 GPU: nVidea GTX 1050 Ti image

Could anyone support please?

I also tried to use the OpenPoseDotNet.Native dll in a test project using the Nuget package (and i copy all the required dependency dlls in the Bin folder) but i always get following error:

System.DllNotFoundException: 'Unable to load DLL 'OpenPoseDotNetNative': The specified procedure could not be found. (Exception from HRESULT: 0x8007007F)'

takuya-takeuchi commented 4 years ago

I retried the following steps again.

Initialize

git clone https://github.com/takuya-takeuchi/OpenPoseDotNet
cd OpenPoseDotNet
git submodule update --init --recursive

Confirmation

>set CUDA_PATH
CUDA_PATH=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v9.2

Build

cd OpenPoseDotNet
BuildWindowsVS2017.bat
cd OpenPoseDotNet/src/OpenPoseDotNet.Native
BuildWindowsVS2017.bat

It works fine.

Error

System.DllNotFoundException: 'Unable to load DLL 'OpenPoseDotNetNative': The specified

You must deploy the following bionary to same directory of OpenPoseDotNet.Native

boost_chrono-vc140-mt-1_61.dll
boost_filesystem-vc140-mt-1_61.dll
boost_python-vc140-mt-1_61.dll
boost_system-vc140-mt-1_61.dll
boost_system-vc140-mt-gd-1_61.dll
boost_thread-vc140-mt-1_61.dll
caffe.dll
caffehdf5.dll
caffehdf5_hl.dll
caffezlib1.dll
cublas64_XX.dll
cudart64_XX.dll
cudnn64_X.dll
curand64_XX.dll
gflags.dll
gflagsd.dll
glog.dll
glogd.dll
libgcc_s_seh-1.dll
libgfortran-3.dll
libopenblas.dll
libquadmath-0.dll
opencv_core310.dll
opencv_ffmpeg310_64.dll
opencv_imgcodecs310.dll
opencv_imgproc310.dll
opencv_world310.dll
opencv_world310d.dll
openpose.dll
python27.dll
VCRUNTIME140.dll
malhosary commented 4 years ago

@takuya-takeuchi : Thanks a lot for your support, it works now for me. For me, i think the issue was in CUDA 10.0, we should use CUDA 9.2