A face recognition algorithm using QR based PCA.
The solution is created with Visual Studio 2012. You are more than welcome to use other IDEs. All you need to do is to copy all the codes from QR-PCA-FaceRec.cpp and link the dependencies correctly.
I'm using OpenCV 3.1, while other version of OpenCV may work as well, since I only use its basic I/O and Mat functionalities.
A high quality linear algebra library (matrix maths) for the C++ language.
Please note that Armadillo requires LAPACK for matrix decompositions and boost. I'm using Intel MKL through Student Program. You may choose any LAPACK replacement you want.
Take these 5 steps:
-
Download the "AT&T Database of Faces" from here and place it anywhere you want.
-
Change the
orl_pathat line 22 to your AT&T Database folder. -
Set number of dimensions you're aim for by changing the
component_numat line 20. -
Choose which measure of distance you want to use at line 164 and 169.
-
Build and run!
HanyNet is released under the MIT license. Considering the dependencies are both open-source, you are granted with the same rights as they offer.