FaceRecognizer - prediction value [closed]

asked 2019-12-13 13:49:30 -0600

amigo421 gravatar image

hi

just newbie question, I've built facerec_video sample from contrib package to try recognition from camera, drop a couple photos of Arnold Schwarzenegger into txt database as described, with different labels , just to fit Fisher algorithm which requires at least two different labels, right?

and I always get prediction value = 1, as I understand this means that person from first photo recognized , what I'm doing wrong?

opencv and contrib built by mingw 7, for win10

P.S. and- yes, I'm not Arnold )))

edit retag flag offensive reopen merge delete

Closed for the following reason not a real question by amigo421
close date 2019-12-13 14:16:37.077174

Comments

what I'm doing wrong?

you seem to expect magic (w/o any own effort understanding it) ;(

to scare you: please take a look here

berak gravatar imageberak ( 2019-12-13 14:12:05 -0600 )edit

you are right , I was wrong. minimal reading the doc gives the understanding . as I got from doc - it always return some index from the database of photos, just need to estimate a correctness with confidence value

amigo421 gravatar imageamigo421 ( 2019-12-13 14:15:21 -0600 )edit

@amigo421. I used FisherFaceRecognize using picamera and USB webcam for Raspberry pi 3/4.

supra56 gravatar imagesupra56 ( 2019-12-13 14:49:46 -0600 )edit
add a comment