Re: Face size

[Georges Quenot <Georges.Quenot@imag.fr>]

Paul Over wrote:
> 
> Huiping Li wrote:
> 
> > I saw in the website that the face required to be detected for the
> > evaluation  is at least a quarter of frame width and height.  This is a
> > really big face. I browse some video clips and did not  find one clip
> > contains such a big face. maybe we can change it to something like 30x30
> > pixels ?
> >
> 
> Hi,
> 
> It may be the minimum size should be smaller. Some guess as to the expected
> frequency of occurrence in the feature test set is useful. (It would of
> course not be appropriate to study the feature extraction test set. I hope
> everyone has segregated that already). I'd be interested in  hearing input
> from others who plan to do the feature extraction task.
> 
> One thing to keep in mind is that the shots will be evaluated by a human
> who is given the feature definition as a guide. So the definition must
> be in terms any college-educated adult can understand. The judges will
> not be counting pixels.

Hi,

We also find the minimum face size quite large. Also, considering
human face and NTSC image aspects ratios, it does not make much
sense to specify "at least one quarter of the image height" once
"at least one quarter of the image width" is satisfied. Also
again, the definition of the size of a face is not obvious
(what precise portions of the face/head are counted in or out,
what about the hairs, beards, ... ?).

A definition in terms any college-educated adult can understand
and that would remove part of these difficulties could be given
in the form of a set of examples of faces of comparable sizes
that will be representative of the targeted minimum size. Systems
could easily get this minimum size information directly from the
samples.

Regards.

Georges Quénot.

Email: Georges.Quenot@imag.fr
CLIPS-IMAG, 385, rue de la Bibliothèque, B.P. 53, 38041 Grenoble Cedex 9
Tel: (33-4) 76 63 58 55, Fax: (33-4) 76 44 66 75