ubuntu16.04编译seetaface

重装系统又编译了一遍seetface，发现果然忘得差不多了，不过重新上手轻车熟路

转载地址https://blog.csdn.net/qq_32231743/article/details/53607221?locationNum=13&fps=1

本文大部分内容来自： 
http://www.cnblogs.com/makefile/p/6059833.html 
和 
http://blog.csdn.net/u013718481/article/details/52994346?locationNum=15&fps=1 
侵删。 
该编译测试过程作为记录备忘。

1，下载seetaface源码

git clone https://github.com/seetaface/SeetaFaceEngine.git

• 2、安装opencv

sudo apt-get install libopencv-dev

• 3、升级cmake

sudo add-apt-repository ppa:george-edison55/cmake-3.x
sudo apt-get update
sudo apt-get install cmake

• 4、安装解压工具

sudo apt-get install unrar

• 5，编译facedetection

mkdir ~/seeta/seetaface
将下载文档中FaceDetection,FaceAlignment,FaceiDentification三个文件夹负责到~/seeta/seetaface
cd ~/seeta/seetaface/FaceDetection/

#新建一个build文件夹
mkdir build
cd build
cmake ..
sudo make -j4

• 6、测试facedetection 

在当前目录下有一个facedet_test文件

#命令原型是./facedet_test image_path model_path

./facedet_test ../data/0_1_1.jpg ../model/seeta_fd_frontal_v1.0.bin

• 7、编译facealignment

cd ~/seeta/seetaface/FaceAlignment
mkdir build
cd build
cmake ..
#将FaceDetection/include/face_detection.h和
#FaceDetection/build/libseeta_facedet_lib_so分别拷贝到
#FaceAlignment/include和FaceAlignment/build目录下

#将FaceAlignment/src/cfan.cpp里面所有的isnan修改成std::isnan
sudo make -j4

• 8、测试facealignment 

编译成功，但测试提示‘段错误’，应该是模型或者图片路径出错，因该模块不需单独使用，暂时不管了。

• 9、编译faceidentification

cd ~/seeta/seetaface/FaceIdentification
mkdir build
cd build

将libseeta_facedet_lib.so,libseeta_fa_lib.so拷贝到~/seeta/seetaface/FaceIdentification/build目录下

将face_alignment.h,face_detection.h拷贝到~/seeta/seetaface/FaceIdentification/include目录下

在src/test/CMakeLists.txt文件循环前添加link_directories(${PROJECT_BINARY_DIR})

修改src/test/CMakeLists.txt，将
  add_executable(${BIN} ${f})
  target_link_libraries(${BIN} viplnet ${OpenCV_LIBS} )
修改为：
  add_executable(${BIN} ${f})
  target_link_libraries(${BIN} viplnet ${OpenCV_LIBS} seeta_facedet_lib seeta_fa_lib)

将model目录中的模型解压
unrar  x  seeta_fr_v1.0.part1.rar

cmake .. 
sudo make -j4