Ask Your Question

How is the LBPH predict value in FaceRec calculated

asked 2014-11-21 11:47:36 -0500

MaxMuc gravatar image

updated 2020-10-27 05:36:52 -0500

Hi everyone, my question concerns the faceRec class and the predict function within. I know that the predict value for Fisher and Eigenface is the distance in the subspace between the input image and the closest training image.

But how is the predict value calculated for LBPH? Is it the number of different values in the histograms? Or the number of different part-histograms? Or something totally different?

Greets Max

edit retag flag offensive close merge delete

2 answers

Sort by ยป oldest newest most voted

answered 2014-11-26 19:40:01 -0500

updated 2014-11-27 22:28:31 -0500

The principle is the same as in Eigenfaces and Fisherfaces: finding the image in the gallery/reference set whose distance to the test image is the closest. But with LBPH, the distance function is compareHist (see function void LBPH::predict(InputArray _src, int &minClass, double &minDist) at this file for more details).

edit flag offensive delete link more

answered 2014-11-26 09:22:20 -0500

berak gravatar image

it is the distance between the histograms, calculated by compareHist(a,b,HIST_CMP_CHISQR)

edit flag offensive delete link more

Question Tools


Asked: 2014-11-21 11:47:36 -0500

Seen: 496 times

Last updated: Nov 27 '14