search

409544320 / face_recog.src

face recognition via Sparse Representation-based Classification (SRC)
27 stars 21 forks source link

face recognition via sparse representation #2

Open mohanrajmit opened 9 years ago

mohanrajmit commented 9 years ago

Dear sir,

I am doing research on face recognition, i saw your work and its very nice. What is the SCI value i supposed to give while testing . please help me.

chaotien commented 9 years ago

SCI is used for blocking person who is not in your training data base(or user data base).

If you want to test YaleB database for face recognition accuracy, set SCI = 0.

In my opinion,

If you want to build a strict face login system, give a high SCI threshold like 0.7~0.9.

Otherwise, your can set it about 0.5.

mohanrajmit commented 9 years ago

thanks for your reply, but your code takes longer time to recognize the person, now i am trying to run in gpu. If you are implemented already in gpu please share your code.

chaotien commented 9 years ago

You're welcome. This is not my code, ha :D

mohanrajmit commented 9 years ago

k