face-api

因face-api工程不用CUDA 所以编译的opencv和caffe都需要不带CUDA版本。

caffe 编译：

修改默认的makeconfigure文件 ,使用nocuda

cmake .. -DCUDA_USE_STATIC_CUDA_RUNTIME=false
make -j16

opencv编译：

unzip opencv2.4.13.zip
mkdir ./release/installed
cmake .. -DWITH_CUDA=OFF -DCMAKE_INSTALL_PREFIX=/srv/luoyang/face/opencv/opencv-2.4.13/release/installed/
make -j16
make install

编译face-api程序：

修改makelist
修改MTMain.h 中的#define CPU_ONLY

cmake .. -DWITH_CUDA=OFF
make -j16

如果 fatal error: caffe/proto/caffe.pb.h: No such file or directory（使用caffe时编译自己工程出错）:
参考https://blog.csdn.net/xiongmingkang/article/details/83421856?utm_medium=distribute.pc_relevant.none-task-blog-title-2&spm=1001.2101.3001.4242

mkdir /srv/luoyang/face/caffe-master/include/caffe/proto
cd /srv/luoyang/face/caffe-master/src/caffe/proto/
protoc caffe.proto  --cpp_out=.
mv caffe.pb.* /srv/luoyang/face/caffe-master/include/caffe/proto/

找不到opencv_dep 用sudo添加软连接

sudo ln -s /usr/local/cuda/lib64/libcudart.so /usr/lib/libopencv_dep_cudart.so

以下本次工程用不着，仅作记录

以下软连接操作是使用CUDA9的时候与opencv2.4冲突 文件命名更改时的解决方法

sudo ln -s /usr/local/cuda-9.0/lib64/libnppial.so /usr/local/lib/libopencv_dep_nppial.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppicc.so /usr/local/lib/libopencv_dep_nppicc.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppicom.so /usr/local/lib/libopencv_dep_nppicom.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppidei.so /usr/local/lib/libopencv_dep_nppidei.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppif.so /usr/local/lib/libopencv_dep_nppif.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppig.so /usr/local/lib/libopencv_dep_nppig.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppim.so /usr/local/lib/libopencv_dep_nppim.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppist.so /usr/local/lib/libopencv_dep_nppist.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppisu.so /usr/local/lib/libopencv_dep_nppisu.so
sudo ln -s /usr/local/cuda-9.0/lib64/libnppitc.so /usr/local/lib/libopencv_dep_nppitc.so

opencv2.4.13与 cuda9.0不兼容
编译出现err opencv_dep_CUDA_nppi_LIBRARY 或者最后调用时找不到相关so文件
解压源文件，代替两个.cmake 再用下面代码编译

cmake -D CMAKE_BUILD_TYPE=RELEASE -D CMAKE_INSTALL_PREFIX=/usr/local -D CUDA_GENERATION=Kepler ..
make -j16

https://blog.csdn.net/qq_41598072/article/details/100663873
https://blog.csdn.net/u014613745/article/details/78310916
https://blog.csdn.net/renhanchi/article/details/80493232

编译失败建议删除全部代码 再次解压重来