ImportError: DLL load failed: The specified module could not be found.

[thwestz]

Expected Behavior

Use dlib with cuda on window 10 64bit

Current Behavior

running install running bdist_egg running egg_info creating dlib.egg-info writing dlib.egg-info\PKG-INFO writing dependency_links to dlib.egg-info\dependency_links.txt writing top-level names to dlib.egg-info\top_level.txt writing manifest file 'dlib.egg-info\SOURCES.txt' package init file 'dlib__init__.py' not found (or not a regular file) reading manifest file 'dlib.egg-info\SOURCES.txt' reading manifest template 'MANIFEST.in' writing manifest file 'dlib.egg-info\SOURCES.txt' installing library code to build\bdist.win-amd64\egg running install_lib running build_py running build_ext Building extension for Python 3.6.6 (v3.6.6:4cf1f54eb7, Jun 27 2018, 03:37:03) [MSC v.1900 64 bit (AMD64)] Invoking CMake setup: 'cmake D:\dlib\tools\python -DCMAKE_LIBRARY_OUTPUT_DIRECTORY=D:\dlib\build\lib.win-amd64-3.6 -DPYTHON_EXECUTABLE=C:\Program Files\Python36\python.exe -G Visual Studio 15 2017 -DCMAKE_LIBRARY_OUTPUT_DIRECTORY_RELEASE=D:\dlib\build\lib.win-amd64-3.6 -A x64' -- Selecting Windows SDK version 10.0.18362.0 to target Windows 10.0.18363. -- The C compiler identification is MSVC 19.16.27034.0 -- The CXX compiler identification is MSVC 19.16.27034.0 -- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio/2017/Community/VC/Tools/MSVC/14.16.27023/bin/Hostx86/x64/cl.exe -- Check for working C compiler: C:/Program Files (x86)/Microsoft Visual Studio/2017/Community/VC/Tools/MSVC/14.16.27023/bin/Hostx86/x64/cl.exe -- works -- Detecting C compiler ABI info -- Detecting C compiler ABI info - done -- Detecting C compile features -- Detecting C compile features - done -- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2017/Community/VC/Tools/MSVC/14.16.27023/bin/Hostx86/x64/cl.exe -- Check for working CXX compiler: C:/Program Files (x86)/Microsoft Visual Studio/2017/Community/VC/Tools/MSVC/14.16.27023/bin/Hostx86/x64/cl.exe -- works -- Detecting CXX compiler ABI info -- Detecting CXX compiler ABI info - done -- Detecting CXX compile features -- Detecting CXX compile features - done -- Found PythonInterp: C:/Program Files/Python36/python.exe (found version "3.6.6") -- Found PythonLibs: C:/Program Files/Python36/libs/Python36.lib -- pybind11 v2.2.2 -- Using CMake version: 3.14.4 -- Compiling dlib version: 19.15.0 -- SSE4 instructions can be executed by the host processor. -- AVX instructions can be executed by the host processor. -- Looking for sys/types.h -- Looking for sys/types.h - found -- Looking for stdint.h -- Looking for stdint.h - found -- Looking for stddef.h -- Looking for stddef.h - found -- Check size of void -- Check size of void - done -- Enabling AVX instructions -- Searching for BLAS and LAPACK -- Searching for BLAS and LAPACK -- Looking for pthread.h -- Looking for pthread.h - not found -- Found Threads: TRUE -- Found CUDA: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0 (found suitable version "10.0", minimum required is "7.5") -- Looking for cuDNN install... -- Found cuDNN: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0/lib/x64/cudnn.lib -- Building a CUDA test project to see if your compiler is compatible with CUDA... -- Checking if you have the right version of cuDNN installed. -- Enabling CUDA support for dlib. DLIB WILL USE CUDA -- C++11 activated. -- Configuring done -- Generating done -- Build files have been written to: D:/dlib/build/temp.win-amd64-3.6/Release Invoking CMake build: 'cmake --build . --config Release -- /m' Microsoft (R) Build Engine version 15.9.21+g9802d43bc3 for .NET Framework Copyright (C) Microsoft Corporation. All rights reserved.

Checking Build System CMake does not need to re-run because D:/dlib/build/temp.win-amd64-3.6/Release/CMakeFiles/generate.stamp is up-to-date. CMake does not need to re-run because D:/dlib/build/temp.win-amd64-3.6/Release/pybind11_build/CMakeFiles/generate.stamp is up-to-date. CMake does not need to re-run because D:/dlib/build/temp.win-amd64-3.6/Release/dlib_build/CMakeFiles/generate.stamp is up-to-date. Building NVCC (Device) object dlib_build/CMakeFiles/dlib.dir/cuda/Release/dlib_generated_cuda_dlib.cu.obj cuda_dlib.cu cuda_dlib.cu Building NVCC (Device) object dlib_build/CMakeFiles/dlib.dir/cuda/Release/dlib_generated_cusolver_dlibapi.cu.obj cusolver_dlibapi.cu cusolver_dlibapi.cu Building Custom Rule D:/dlib/dlib/CMakeLists.txt CMake is re-running because D:/dlib/build/temp.win-amd64-3.6/Release/dlib_build/CMakeFiles/generate.stamp is out-of-date. the file 'D:/dlib/build/temp.win-amd64-3.6/Release/dlib_build/CMakeFiles/dlib.dir/cuda/dlib_generated_cuda_dlib.cu.obj.depend' is newer than 'D:/dlib/build/temp.win-amd64-3.6/Release/dlib_build/CMakeFiles/generate.stamp.depend' result='-1' -- Selecting Windows SDK version 10.0.18362.0 to target Windows 10.0.18363. -- pybind11 v2.2.2 -- Using CMake version: 3.14.4 -- Compiling dlib version: 19.15.0 -- SSE4 instructions can be executed by the host processor. -- AVX instructions can be executed by the host processor. -- Enabling AVX instructions -- Looking for cuDNN install... -- Found cuDNN: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v10.0/lib/x64/cudnn.lib -- Enabling CUDA support for dlib. DLIB WILL USE CUDA -- C++11 activated. -- Configuring done -- Generating done -- Build files have been written to: D:/dlib/build/temp.win-amd64-3.6/Release base64_kernel_1.cpp bigint_kernel_1.cpp bigint_kernel_2.cpp bit_stream_kernel_1.cpp entropy_decoder_kernel_1.cpp entropy_decoder_kernel_2.cpp entropy_encoder_kernel_1.cpp entropy_encoder_kernel_2.cpp md5_kernel_1.cpp tokenizer_kernel_1.cpp unicode.cpp image_dataset_metadata.cpp mnist.cpp global_function_search.cpp kalman_filter.cpp test_for_odr_violations.cpp auto.cpp sockets_kernel_1.cpp bsp.cpp dir_nav_kernel_1.cpp dir_nav_kernel_2.cpp dir_nav_extensions.cpp linker_kernel_1.cpp extra_logger_headers.cpp logger_kernel_1.cpp logger_config_file.cpp misc_api_kernel_1.cpp misc_api_kernel_2.cpp sockets_extensions.cpp sockets_kernel_2.cpp sockstreambuf.cpp sockstreambuf_unbuffered.cpp server_kernel.cpp server_iostream.cpp server_http.cpp multithreaded_object_extension.cpp threaded_object_extension.cpp threads_kernel_1.cpp threads_kernel_2.cpp threads_kernel_shared.cpp thread_pool_extension.cpp async.cpp timer.cpp stack_trace.cpp cpu_dlib.cpp tensor_tools.cpp fonts.cpp widgets.cpp drawable.cpp canvas_drawing.cpp style.cpp base_widgets.cpp gui_core_kernel_1.cpp gui_core_kernel_2.cpp png_loader.cpp save_png.cpp jcomapi.cpp jdapimin.cpp jdapistd.cpp jdatasrc.cpp jdcoefct.cpp jdcolor.cpp jddctmgr.cpp jdhuff.cpp jdinput.cpp jdmainct.cpp jdmarker.cpp jdmaster.cpp jdmerge.cpp jdphuff.cpp jdpostct.cpp jdsample.cpp jerror.cpp jidctflt.cpp jidctfst.cpp jidctint.cpp jidctred.cpp jmemmgr.cpp jmemnobs.cpp jquant1.cpp jquant2.cpp jutils.cpp jcapimin.cpp jdatadst.cpp jcparam.cpp jcapistd.cpp jcmarker.cpp jcinit.cpp jcmaster.cpp jcdctmgr.cpp jccoefct.cpp jccolor.cpp jchuff.cpp jcmainct.cpp jcphuff.cpp jcprepct.cpp jcsample.cpp jfdctint.cpp jfdctflt.cpp jfdctfst.cpp jpeg_loader.cpp save_jpeg.cpp cudnn_dlibapi.cpp cublas_dlibapi.cpp curand_dlibapi.cpp cuda_data_ptr.cpp gpu_data.cpp png.c pngerror.c pngget.c pngmem.c pngpread.c pngread.c pngrio.c pngrtran.c pngrutil.c pngset.c pngtrans.c pngwio.c pngwrite.c pngwtran.c pngwutil.c adler32.c compress.c crc32.c deflate.c gzclose.c gzlib.c gzread.c gzwrite.c infback.c inffast.c inflate.c inftrees.c trees.c uncompr.c zutil.c dlib.vcxproj -> D:\dlib\build\temp.win-amd64-3.6\Release\dlib_build\Release\dlib19.15.0_release_64bit_msvc1916.lib Building Custom Rule D:/dlib/tools/python/CMakeLists.txt CMake does not need to re-run because D:/dlib/build/temp.win-amd64-3.6/Release/CMakeFiles/generate.stamp is up-to-date. dlib.cpp matrix.cpp vector.cpp svm_c_trainer.cpp D:\dlib\tools\python\src\vector.cpp(62): warning C4101: 'e': unreferenced local variable [D:\dlib\build\temp.win-amd64-3.6\Release\dlib_python.vcxproj] svm_rank_trainer.cpp decision_functions.cpp other.cpp basic.cpp D:\dlib\tools\python\src\basic.cpp(22): warning C4101: 'e': unreferenced local variable [D:\dlib\build\temp.win-amd64-3.6\Release\dlib_python.vcxproj] cca.cpp sequence_segmenter.cpp svm_struct.cpp D:\dlib\tools\python\src\svm_struct.cpp(59): warning C4101: 'e': unreferenced local variable [D:\dlib\build\temp.win-amd64-3.6\Release\dlib_python.vcxproj] D:\dlib\tools\python\src\svm_struct.cpp(51): note: while compiling class template member function 'void svm_struct_prob::separation_oracle(const long,const dlib::matrix<double,0,1,dlib::default_memory_manager,dlib::row_major _layout> &,double &,std::vector<ulong_double_pair,std::allocator<_Ty>> &) const' with [ psi_type=std::vector<ulong_double_pair,std::allocator>, _Ty=ulong_double_pair ] D:\dlib\tools\python\src\svm_struct.cpp(110): note: see reference to class template instantiation 'svm_struct_prob' being compiled with [ psi_type=std::vector<ulong_double_pair,std::allocator> ] D:\dlib\tools\python\src\svm_struct.cpp(136): note: see reference to function template instantiation 'dlib::matrix<double,0,1,dlib::default_memory_manager,dlib::row_major_layout> solve_structural_svm_problem_impl<std::vector<ulong_dou ble_pair,std::allocator<_Ty>>>(pybind11::object)' being compiled with [ _Ty=ulong_double_pair ] image.cpp image2.cpp image3.cpp image4.cpp D:\dlib\dlib..\dlib/image_transforms/assign_image.h(86): warning C4018: '>=': signed/unsigned mismatch (compiling source file D:\dlib\tools\python\src\image3.cpp) [D:\dlib\build\temp.win-amd64-3.6\Release\dlib_python.vcxproj] D:\dlib\dlib..\dlib/image_transforms/assign_image.h(160): note: see reference to function template instantiation 'void dlib::impl_assign_image_scaled<out_image_type,src_image_type>(dlib::image_view &,const src_image_t ype &,const double)' being compiled with [ out_image_type=dlib::numpy_image, src_image_type=dlib::matrix_op<dlib::op_image_to_mat<dlib::numpy_image,uint32_t>> ] (compiling source file D:\dlib\tools\python\src\image3.cpp) D:\dlib\dlib..\dlib/image_transforms/assign_image.h(177): note: see reference to function template instantiation 'void dlib::impl_assign_image_scaled<dest_image_type,dlib::matrix_op<dlib::op_image_to_mat<src_image_type,T>>>(dest_imag e_type &,const dlib::matrix_op<dlib::op_image_to_mat<src_image_type,T>> &,const double)' being compiled with [ dest_image_type=dlib::numpy_image, src_image_type=dlib::numpy_image, T=uint32_t ] (compiling source file D:\dlib\tools\python\src\image3.cpp) D:\dlib\tools\python\src\image3.cpp(27): note: see reference to function template instantiation 'void dlib::assign_image_scaled<dlib::numpy_image,dlib::numpy_image>(dest_image_type &,const src_image_type &,const doub le)' being compiled with [ dest_image_type=dlib::numpy_image, src_image_type=dlib::numpy_image ] D:\dlib\tools\python\src\image3.cpp(435): note: see reference to function template instantiation 'pybind11::array convert_image_scaled(const dlib::numpy_image &,const std::string &,const double)' being compiled D:\dlib\dlib..\dlib/image_transforms/assign_image.h(87): warning C4018: '<=': signed/unsigned mismatch (compiling source file D:\dlib\tools\python\src\image3.cpp) [D:\dlib\build\temp.win-amd64-3.6\Release\dlib_python.vcxproj] rectangles.cpp object_detection.cpp shape_predictor.cpp correlation_tracker.cpp face_recognition.cpp cnn_face_detector.cpp global_optimization.cpp image_dataset_metadata.cpp numpy_returns.cpp line.cpp gui.cpp Creating library D:/dlib/build/temp.win-amd64-3.6/Release/Release/dlib.lib and object D:/dlib/build/temp.win-amd64-3.6/Release/Release/dlib.exp dlib_python.vcxproj -> D:\dlib\build\lib.win-amd64-3.6\dlib.cp36-win_amd64.pyd Building Custom Rule D:/dlib/tools/python/CMakeLists.txt CMake does not need to re-run because D:/dlib/build/temp.win-amd64-3.6/Release/CMakeFiles/generate.stamp is up-to-date. creating build\bdist.win-amd64 creating build\bdist.win-amd64\egg copying build\lib.win-amd64-3.6\dlib.cp36-win_amd64.pyd -> build\bdist.win-amd64\egg creating stub loader for dlib.cp36-win_amd64.pyd byte-compiling build\bdist.win-amd64\egg\dlib.py to dlib.cpython-36.pyc creating build\bdist.win-amd64\egg\EGG-INFO copying dlib.egg-info\PKG-INFO -> build\bdist.win-amd64\egg\EGG-INFO copying dlib.egg-info\SOURCES.txt -> build\bdist.win-amd64\egg\EGG-INFO copying dlib.egg-info\dependency_links.txt -> build\bdist.win-amd64\egg\EGG-INFO copying dlib.egg-info\not-zip-safe -> build\bdist.win-amd64\egg\EGG-INFO copying dlib.egg-info\top_level.txt -> build\bdist.win-amd64\egg\EGG-INFO writing build\bdist.win-amd64\egg\EGG-INFO\native_libs.txt creating dist creating 'dist\dlib-19.15.0-py3.6-win-amd64.egg' and adding 'build\bdist.win-amd64\egg' to it removing 'build\bdist.win-amd64\egg' (and everything under it) Processing dlib-19.15.0-py3.6-win-amd64.egg creating c:\program files\python36\lib\site-packages\dlib-19.15.0-py3.6-win-amd64.egg Extracting dlib-19.15.0-py3.6-win-amd64.egg to c:\program files\python36\lib\site-packages Adding dlib 19.15.0 to easy-install.pth file

Installed c:\program files\python36\lib\site-packages\dlib-19.15.0-py3.6-win-amd64.egg Processing dependencies for dlib==19.15.0 Finished processing dependencies for dlib==19.15.0

Steps to Reproduce

1. download dlib from github
2. go to dlib folder and run python setup.py install -G "Visual Studio 15 2017" --clean
3. import dlib
4. get error

import dlib Traceback (most recent call last): File "", line 1, in ImportError: DLL load failed: The specified module could not be found.

• Version: 19.15
• Where did you get dlib: https://github.com/davisking/dlib
• Platform: Window 10 64 bit python 3.6
• Compiler: C++ on visual studio 2017

[davisking]

What happens with the newest dlib?

[thwestz]

Same error.If possible, In my environment have 2 visual studio ( 2017 , 2019 ) ?

[davisking]

Try building with cuda. Give --clean --no DLIB_USE_CUDA

[davisking]

And run pip uninstall dlib a bunch to make sure you don't have several versions of dlib installed first.

[thwestz]

Now,i typing

1. pip uninstall dlib
2. python setup.py install -G "Visual Studio 15 2017" --clean --no DLIB_USE_CUDA i can import dlib but my dlib.DLIB_USE_CUDA return false this mean my dlib not use gpu ? and how to turn it to true ?

[davisking]

Something is wrong with your CUDA install, this is not a dlib issue. You were getting this DLL error because windows can't find the CUDA runtime DLLs. You probably don't have the CUDA runtime in your PATH.

On Thu, Dec 26, 2019 at 11:42 PM THWest notifications@github.com wrote:

Now,i typing

1. pip uninstall dlib
2. python setup.py install -G "Visual Studio 15 2017" --clean --no DLIB_USE_CUDA i can import dlib but my dlib.DLIB_USE_CUDA return false this mean my dlib not use gpu ? and how to turn it to true ?

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/davisking/dlib/issues/1954?email_source=notifications&email_token=ABPYFR2XO4ZOPXZYB3JQIO3Q2WBUBA5CNFSM4J6QULZ2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEHWRPAA#issuecomment-569186176, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABPYFR4UJNUABUCFSTDBF23Q2WBUBANCNFSM4J6QULZQ .

[thwestz]

Thank you, Now i can solved this problem via set CUDA runtime in my PATH

[yanglixiao1994]

@thwestz Could you explain more specific?I have the same problem.How to set CUDA runtime PATH?My PATH already have the path to CUDA when i install it.Just like the picture below: Thanks for any advise.

[Itach1Uchixa]

Getting same error with CUDA v10.2 in Windows 10. Can you explain step-by-step how to add CUDA runtime to PATH? I have added all paths that shown here but error persisted after compilation. I am using: Visual Studio 2019 compiler, CMake 3.17.0-rc3, CUDA 10.2, CUDnn 7.6, python 3.5. For information I can compile and run CUDA sample projects with Visual Studio 2019 Please help! Thanks in advance.

[cashlo]

@yanglixiao1994 @Itach1Uchixa Are you using Python 3.8? I have the same error, and after digging into this until 5am, it turns out Python 3.8 no longer load DLLs from PATH, the same issue and a fix is documented here in a different project. I have created another issue https://github.com/davisking/dlib/issues/2039

[Itach1Uchixa]

@cashlo I actually got it compiled by adding CUDNN\bin dir to my PATH. Somehow you need to add CUDNN\bin dir too not just CUDNN root dir. In my case (and I think in many cases too) cmake couldn't find dll file inside CUDNN\bin dir and compiled dlib with CPU support only. Below is my working path configuration

[ahsanfayaz52]

I had the same issue i was using python 3.6 and dlib 19.19.0 so i just downgraded my dlib version to 19.8.1 and booommm it workd for me

[ahsanfayaz52]

I also uninstalled cuda 10.1 and installed cuda 8.0 as cuda 10.1 was giving me runtime error i suggest to downgrad your cuda version too

[siddh2]

when i try to import dlib then I get error My error

---

---

ImportError: DLL load while importing dlib: A dynamic link library (DLL) initialization routing failed.

plz help me