Thread Subject: Gender detection

After I detect the face of a person which I stored in an axes component(guide matlab) and compute the haussdorf distance using that formula between the eyes for example(which represent the distance between two points)

Formula for distance Hausdorff:

distance_Hausdorff=sqrt( (x1-x2)*(x1-x2) + (y1-y2)*(y1-y2) ) ; for instance is the generalisde Pytagora theorem in analytical geometry. That works!

My problem is now how can I determine if the person which I load the picture in axes component is a man or female.

Any ideas of links or topics?

Thanks!

On Jul 3, 9:57 am, "Sprinceana " <mihai...@yahoo.com> wrote:
> After I detect the face of a person which I stored in an axes component(guide matlab) and compute the haussdorf distance using that formula between the eyes for example(which represent the distance between two points)
>
> Formula for distance Hausdorff:
>
> distance_Hausdorff=sqrt( (x1-x2)*(x1-x2) + (y1-y2)*(y1-y2) ) ; for instance is the generalisde Pytagora theorem in analytical geometry. That works!
>
> My problem is now how can I determine if the person which I load the picture in axes component is a man or female.
>
> Any ideas of links or topics?
>
> Thanks!

IMHO, the distance between the eyes does not determine gender.
What determines gender is located somewhat south of the eyes.

On Jul 2, 5:57 pm, "Sprinceana " <mihai...@yahoo.com> wrote:
> After I detect the face of a person which I stored in an axes component(guide matlab) and compute the haussdorf distance using that formula between the eyes for example(which represent the distance between two points)
>
> Formula for distance Hausdorff:
>
> distance_Hausdorff=sqrt( (x1-x2)*(x1-x2) + (y1-y2)*(y1-y2) ) ; for instance is the generalisde Pytagora theorem in analytical geometry. That works!
>
> My problem is now how can I determine if the person which I load the picture in axes component is a man or female.
>
> Any ideas of links or topics?
>
> Thanks!
----------------------------------------------------------------------
I like Tidman's response. But this has been a well studied topic. A
Google search on "male female face differences" turns up over 4
million hits. There was even a famous article about it about 3-5
years ago. I forget which - Nature or Scientific American or
something like that. I don't know who are the most prominent
researchers in the field. I saw one site where they could use sliders
to change a person's gender, ethnicity, and something else (age?) on a
continuous scale - very cool! A theory was put forth many years ago
that the most beautiful face is the average of all faces, and that set
forth a flurry of experiments to see if that's true. It's a
complicated subject but perhaps you'll find something.

But how can you say that is the Hausdorf distance? You gave the
formula for a simple Euclidean distance - square root of the sum of
the squares. Here's a link on Hausdorf distance that you might use to
better understand it:
http://cgm.cs.mcgill.ca/~godfried/teaching/cg-projects/98/normand/main.html
If you have the very special case of only two isolated points, you'd
be right, but the Hausdorf distance was made for shapes (collections
of points) and in that more general situation (which it was designed
for), your formula is wrong.


http://www.virtualffs.co.uk/difference%20examples.htm
http://www.theage.com.au/news/national/beautiful-computer-says-yes/2007/03/17/1174080223528.html
http://www.cs.cmu.edu/afs/andrew/scs/cs/15-463/f07/proj3/www/lisachan/
http://morph.cs.st-andrews.ac.uk//fof/index.html

Ok.

Thanks for your responses!

I'll study and read this and if I have questions I'll continue this topic.


Yes that formula is for euclidian distance you're wright mea culpa!


Thanks again!

I think this link give me an idea on how to achieve that (in the way also it looks my matlab gui).

http://morph.cs.st-andrews.ac.uk//fof/index.html

Thanks again!

On 2 Jul, 23:57, "Sprinceana " <mihai...@yahoo.com> wrote:

> My problem is now how can I determine if the person which I load the picture in axes component is a man or female.
>
> Any ideas of links or topics?

Pick up a book on art, on drawing or painting portraits.

Artists need to be aware of the characteristics that
separate male and female faces, like the proportions
of certain features, thickness of bone structures and
so on.

Just be aware that you are playing with deceptive
concepts. You might think that "heck, *I* can see
if that person is a male or a female, it should be
easy to have the computer do the discrimination".

What happens in reality is that you base the discrimunation
on the whole evolutionary history of Homo Sapiens.
The human brain is fine-tuned to size up every single
person you see:

- Male or female?
- Age?
- Physical health?
- Social status?
- Ethnical genealogy?
- State of mind?

From an evolutionary perspective, it is vital for everybody
to size other people up quickly. Those who don't tend to end
up in trouble. Or miss a potential spouse. So you will *always*
come up with characteristics like those above, if you have to.
They will not always be correct, but you will come up with
them nonetheless.

In other words, you might be trying to replicate decisions
that in real life are made deep down in the human brains,
based on very subtle clues.

Rune

In article <588e84fa-295b-44a1-b53d-63b87911735e@v15g2000prn.googlegroups.com>,
TideMan <mulgor@gmail.com> wrote:
>IMHO, the distance between the eyes does not determine gender.
>What determines gender is located somewhat south of the eyes.

That is sex, surely? However, gender may be located behind the eyes.
;-)

Some discussion of gender vs sex here:

http://en.wikipedia.org/wiki/Gender#The_word_gender_in_English

"Among the reasons that working scientists have given me for choosing
gender rather than sex in biological contexts are desires to signal
sympathy with feminist goals, to use a more academic term, or to avoid
the connotation of copulation."
- David Haig, The Inexorable Rise of Gender and the Decline of Sex.

Francis

Very true Rune, as borne out by this study where people were able to
detect homosexuality in men based on very small portions of their
face:

http://www.scientificamerican.com/article.cfm?id=something-queer-about-that-face

".....Furthermore, in an even more rigorously controlled series of
experiments published in the Journal of Personality and Social
Psychology, Rule and his colleagues replicated their discovery that
people are able to accurately guess male sexual orientation. This
time, the researchers demonstrated that perceivers were able to do
this even when they were shown only individual features of the
target's face. For example, when shown only the eye region ("without
brows and cropped to the outer canthi so that not even "crow's-feet"
were visible"), perceivers were amazingly still able to accurately
identify a man as being gay. The same happened when shown the mouth
region alone. Curiously, most of the participants underestimated their
ability to identify gay faces from these features alone. That is to
say, people seem to have honed and calibrated their gaydar without
knowing they've done so......"

"Rune Allnor" <allnor@tele.ntnu.no> wrote in message
news:d30b2db0-96a6-453e-80b5-c3d572253d05@a36g2000yqc.googlegroups.com...

*snip*

> What happens in reality is that you base the discrimunation
> on the whole evolutionary history of Homo Sapiens.
> The human brain is fine-tuned to size up every single
> person you see:

And even at that, it's not always easy for humans to make that
determination.

http://en.wikipedia.org/wiki/Pat_(Saturday_Night_Live)

--
Steve Lord
slord@mathworks.com