Ask Your Question

Rank N for OpenCV Face Recognition

asked 2018-04-05 11:58:42 -0500

Lucas Amparo Barbosa gravatar image

Hi everyone,

Someone already does a Rank N for the algorithms from OpenCV faces?

Have some way to take only the feature set from the eigen/fisher/lbph processing to build my onw matching process?

Thanks guys.

edit retag flag offensive close merge delete

1 answer

Sort by ยป oldest newest most voted

answered 2018-04-06 03:38:40 -0500

berak gravatar image

updated 2018-04-06 03:40:01 -0500

by default, a 1 nearest neighbour distance is used (that would be Rank 1)

but you can implement your own ranking scheme by accessing the complete results from the prediction

something like:

Ptr<PredictCollector> collector = StandardCollector::create();
recognizer->predict( img, collector );
std::vector< std::pair<int, double> >  = collector->getResults();
edit flag offensive delete link more


and.... ofc. you can grab the feature extraction code from the face module, and build your own scheme !

also note, that latest opencv alsohasapretrained openface dnn, which can beused easily for this.

berak gravatar imageberak ( 2018-04-06 04:10:08 -0500 )edit

i think this code willl works for me. Thanks berak.

Lucas Amparo Barbosa gravatar imageLucas Amparo Barbosa ( 2018-04-06 04:19:55 -0500 )edit

Question Tools

1 follower


Asked: 2018-04-05 11:58:42 -0500

Seen: 111 times

Last updated: Apr 06 '18