EigenFaceRecognizer usage

asked 2020-04-27 02:52:29 -0500

amigo421 gravatar image

updated 2020-04-27 02:53:48 -0500

I'm trying to use Eigen recognizer in C++ code and it looks oddly for me, please correct my dev logic if it's wrong : example scenario:

  • there is a database of the photos with user ids (more than one photo per user id)
  • I'm training the recognizer with e.g. user id = 29 (5 photos), and user id = 2 (4 photos), so input vectors are (first column is a labels vector, second one is grayscaled Mats):

29 - 1th user's photo1

29 - 1th user's photo2

29 - 1th user's photo3

29 - 1th user's photo4

29 - 1th user's photo5

2 - 2th user's photo1

2 - 2th user's photo2

2 - 2th user's photo3

2 - 2th user's photo4

  • for the testing only , I'm using first photo of user id = 2 as input

  • the results looks oddly : confidence = ~40.000, but label = 29 (supposed 2 as user_id during the training in labels vector)

what is wrong in such logic?

thank you

edit retag flag offensive close merge delete

Comments

I used python.

supra56 gravatar imagesupra56 ( 2020-04-27 21:24:25 -0500 )edit