欢送关注我的公众号 [极智视界],回复 001 获取 Google 编程标准
O_o
>_<
o_O
O_o
~_~
o_O
本教程具体记录了在 win10 环境中应用 qmake 构建 ncnn vs 工程的办法。
1、筹备 opencv
windows 上想要应用 opencv 很简略,opencv 官网提供了编译好的 windows 版本的库能够间接拿来用。
传送门:https://opencv.org/releases/
当然也能够抉择源码编译 opencv,能够参考《【教训分享】windows 源码编译 opencv》。
2、编译 protobuf
留神要用 vs2017 自带的终端 – 实用于 VS2017 的 X64 本机工具命令提醒。
下载 protobuf-3.4.0,传送:https://github.com/google/pro…
编译 protobuf:
cd <protobuf-root-dir>
mkdir build
cd build
call "C:\Program Files (x86)\Microsoft Visual Studio\2017\Professional\VC\Auxiliary\Build\vcvarsall.bat" amd64
cmake -G"NMake Makefiles" -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=%cd%/install -Dprotobuf_BUILD_TESTS=OFF -Dprotobuf_MSVC_STATIC_RUNTIME=OFF ../cmake
nmake
nmake install
编译完会在 build/install 下生成如下目录构造:
3、编译 ncnn
clone ncnn 工程:
git clone --recursive https://github.com/Tencent/ncnn.git
编译 ncnn,其中 <protobuf-root-dir> 为下面下载的 protobuf 源码包门路:
cd <ncnn-root-dir>
mkdir -p build
cd build
call "C:\Program Files (x86)\Microsoft Visual Studio\2017\Professional\VC\Auxiliary\Build\vcvarsall.bat" amd64
cmake -G"NMake Makefiles" -DCMAKE_BUILD_TYPE=Release -DCMAKE_INSTALL_PREFIX=%cd%/install -DProtobuf_INCLUDE_DIR=<protobuf-root-dir>/build/install/include -DProtobuf_LIBRARIES=<protobuf-root-dir>/build/install/lib/libprotobuf.lib -DProtobuf_PROTOC_EXECUTABLE=<protobuf-root-dir>/build/install/bin/protoc.exe -DNCNN_VULKAN=OFF ..
nmake
nmake install
编译实现后会在 build/install 下生成如下目录构造:
且在 build/src 下会生成一些 .h 和 .cpp,会用于前面构建工程。以上就实现了 vs2017 构建 ncnn windows 版本库依赖的编译,上面进入 vs2017 构建。
4、vs2017 构建 ncnn
我这里应用 qmake 来构建 vs 工程。qmake 的 ncnn.pro 如下,从 pro 里能够看到构建 ncnn 工程时,会将后面的 opencv、protobuf、ncnn 编译进去的文件作为三方库了。
# ncnn.pro
TEMPLATE = app
CONFIG += console debug_and_release warn_on c++11
QT -= gui
win32 {QMAKE_CXXFLAGS += /MP /utf-8}
QMAKE_CXXFLAGS_RELEASE += $$QMAKE_CFLAGS_RELEASE_WITH_DEBUGINFO
QMAKE_LFLAGS_RELEASE += $$QMAKE_LFLAGS_RELEASE_WITH_DEBUGINFO
Debug:{
TARGET = ncnn
DESTDIR = ./build/debug
TEMP_DESTDIR = ./build/intermediate/debug/$$TARGET
}
Release:{
TARGET = ncnn
DESTDIR = ./build/release
TEMP_DESTDIR = ./build/intermediate/release/$$TARGET
}
MOC_DIR = $$TEMP_DESTDIR/moc
RCC_DIR = $$TEMP_DESTDIR/rcc
UI_DIR = $$TEMP_DESTDIR/qui
OBJECTS_DIR = $$TEMP_DESTDIR/obj
win32 {
LIBS += -L./3rdparty/protobuf/lib/ libprotobuf.lib libprotobuf-lite.lib libprotoc.lib
CONFIG(debug,debug|release){
LIBS += \
-L./3rdparty/opencv/x64/vc15/lib opencv_imgprocd.lib opencv_imgcodecsd.lib opencv_cored.lib opencv_dnnd.lib
}
CONFIG(release,debug|release){
LIBS += \
-L./3rdparty/opencv/x64/vc15/lib opencv_imgproc.lib opencv_imgcodecs.lib opencv_core.lib opencv_dnn.lib
}
}
win32 {
INCLUDEPATH += \
./3rdparty/opencv/include \
./3rdparty/protobuf/include/google/protobuf \
./3rdparty/ncnn/include/ncnn \
./3rdparty/ncnn/src \
./3rdparty/ncnn/src/layer/x86
}
HEADERS += ./src/xxx.cpp \
SOURCES += ./src/xxx.h \
#CUDA_SOURCES += \
win32 {
SYSTEM_NAME = x64
SYSTEM_TYPE = 64
CUDA_ARCH = compute_35
CUDA_CODE = sm_35
CUDA_INC = $$join(INCLUDEPATH,'"-I"','-I"','"')
MSVCRT_LINK_FLAG_DEBUG = "/MDd"
MSVCRT_LINK_FLAG_RELEASE = "/MD"
# Configuration of the Cuda compiler
CONFIG(debug, debug|release) {
# Debug mode
cuda.input = CUDA_SOURCES
cuda.output = $$OBJECTS_DIR/${QMAKE_FILE_BASE}_cuda.obj
cuda.commands = $$PWD/./3rdparty/cuda/bin/nvcc.exe -D_DEBUG -Xcompiler $$MSVCRT_LINK_FLAG_DEBUG -c -Xcompiler $$join(QMAKE_CXXFLAGS,",") $$join(INCLUDEPATH,'"-I"','-I"','"') ${QMAKE_FILE_NAME} -o ${QMAKE_FILE_OUT}
QMAKE_EXTRA_COMPILERS += cuda
} else {
# Release mode
cuda.input = CUDA_SOURCES
cuda.output = $$OBJECTS_DIR/${QMAKE_FILE_BASE}_cuda.obj
cuda.commands = $$PWD/./3rdparty/cuda/bin/nvcc.exe -Xcompiler $$MSVCRT_LINK_FLAG_RELEASE -c -Xcompiler $$join(QMAKE_CXXFLAGS,",") $$join(INCLUDEPATH,'"-I"','-I"','"') ${QMAKE_FILE_NAME} -o ${QMAKE_FILE_OUT}
QMAKE_EXTRA_COMPILERS += cuda
}
}
再构建一个 create_vs_pro.bat:
# create_vs_pro.bat
set PRO_FILE=ncnn.pro
set VC_DIR=C:\Program Files (x86)\Microsoft Visual Studio\2017\Professional\VC\Auxiliary\Build
call "%VC_DIR%\vcvars64.bat"
set QT_5_9_8=C:\Qt\Qt5.9.8\5.9.8\msvc2015_64\bin
set OPENCV_4_3_0=.\3rdparty\opencv\x64\vc15\bin
set CUDA_11_0=C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v11.0\bin
SET "PATH=%OPENCV_4_3_0%;%CUDA_11_0%;%PATH%"
qmake -r -tp vc %PRO_FILE%
pause
双击 create_vs_pro.bat 就能生成 ncnn.sln,而后就能够进 vs studio 进行编译调试了。
须要留神的是,我这里是没有把 examples、tools、tests、benchmark 里大量的实用工具做成独立工程的模式,这些用到的时候再独自加到工程的 source 里就能够了。
其实 qmake 编译那块还有些细坑没写,推敲一下应该能解决,若有问题欢送分割我,心愿我的分享能对你有一点帮忙。
【公众号传送】
《【教训分享】win10 qmake 构建 ncnn vs 工程》