1. project description
I use lbp_hist describe the face,
then mark it with label 1,2,3...n,
i means the data is describe the i_th person...
2. wrong point
a. describe data, not the lbp_hist, but the sub of hist...
that is |Va-Vb|...
this can test all person...
b. label, not the 1,2,3...n, but the 0,1...
0 is not the same person, 1 is the same person...
this can improve the positive rate...