Problem Training

[ozgurgundogan]

I think there is a problem in testing. I tested different number of classifiers, different size of features and different number of training image. However, it either find all faces correctly (0 correct in non-faces) or find all non-faces correctly (0 correct in faces).
By the way I am using Caltech web faces dataset. Why does it happen ?

Can you share your dataset which you used in notebook ?

[Simon-Hohberg]

Hey,
that sounds strange. The data set that I used came from the university, which is why I am not sure whether I can make it public.
However, I found that the data set here is exactly the data set I used.

Regards,

Simon