01 环境
win10x64
vs2015 update3(其他vs2010以上应该都可以)
cmake3.7.2
git
opencv3.3.0,参考http://blog.csdn.net/longji/article/details/77885076
源码:https://github.com/5455945/SeetaFaceEngine.git
02 源码下载
cd D:\git\DeepLearning
git clone https://github.com/5455945/SeetaFaceEngine.git
cd SeetaFaceEngine
03 获取opencv3.3.0版本
参考:http://blog.csdn.net/longji/article/details/77885076
04 配置cmake-gui
源码路径:D:/git/DeepLearning/SeetaFaceEngine
目标路径:D:/git/DeepLearning/SeetaFaceEngine/build/x64
点击[Configure]后,设置OpenCV_DIR=C:/Program Files/OpenCV 3.3.0-dirty
设置好OpenCV_DIR后,点击[Generate],完成后,点击[Open Project]打开vs2015
05 编译
05.01 人脸检测test_face_detection
vs2015打开D:\git\DeepLearning\SeetaFaceEngine\build\x64\Project.sln
后,直接build debug版本后,F5。默认运行test_face_detection人脸检测工程。
或者设置test_face_detection工程为启动项。
05.02 人脸对齐test_face_alignment
设置test_face_alignment工程为启动项。设置适当的断点。运行后,生成D:\git\DeepLearning\SeetaFaceEngine\build\x64\bin\Debug\test_face_alignment.jpg
05.03 人脸识别测试test_face_recognizer
Test successful!
Average crop face time: 13.74ms
Test successful!
Average extract feature time: 167.22ms
Test successful!
Average extract feature time: 178.52ms
05.04 人脸识别相似度test_face_verification
相似度:0.691146
生成对齐图片:
D:\git\DeepLearning\SeetaFaceEngine\build\x64\bin\Debug\gallery_point_result.jpg
D:\git\DeepLearning\SeetaFaceEngine\build\x64\bin\Debug\probe_point_result.jpg
06 MFC demo
参考:http://blog.csdn.net/wuzuyu365/article/details/52574365
集成进来一个MFC 相似度对比的demo。
cmake的时候,可以勾选SFMFCTEST复选框,默认勾选状态。
测试效果如下: