26 replies to this topic

[mariusj]

I am just going to generally address the response. I just don't feel that an exercise of this sort for lack of better words constitutes much of being informative even as approximation. Can you imagine if you were to want to find the average of 20 rear ends photographed them and stiched them together it would still simply look like just yet another unique rear end! On many levels you have said that it even serves to overturn people's stereotypes that they have already formed in their minds using true to scale randomness since it is reality which informs real stereotypes and photos on magazines and such do so much to contort the issue of a person's innate subjectivity by masquerading it as some form of objectivity. The critical point you bring up is that even in pursuing stereotypes people do far better in judging for themselves but actually score inefficiently when they are presented with the average faces. Unless you can take a person's dna and create a map which shows just exactly which part of the dna will construct which part of the anatomy and then proceed to tally up the most common ones there is absolutely no way that this kind of theorization about the average of faces can actually work in producing a clear analysis of who from where looks like what which I believe you brought up repeatedly in saying that the algorithms used to process the faces are in question. The most obvious sorts of errors in the data sampling are implied by the way the photo was taken at what angle and with what lighting involved in the set up of the photo the cosmetics involved in relation to the individuals their hair blocking their forehead etc etc. I just think this sort of thing is better left to the antics of showing what kind of babies you and someone else could possibly make.

Actually, Central Limit Theorem states that if you do an random event [in your example, rear ends] numerous times it approximate the normal distribution, and that you will find that a mean exists, in this case, the 'average' looking one. So if you in fact does do an event many times, it will provide you with the more 'average' looking item, in Long Ma's case, faces.

When you mix many faces, even if they are not purely random [or fair] they do approximate even if some what biased.

[LongMa]

Actually, Central Limit Theorem states that if you do an random event [in your example, rear ends] numerous times it approximate the normal distribution, and that you will find that a mean exists, in this case, the 'average' looking one. So if you in fact does do an event many times, it will provide you with the more 'average' looking item, in Long Ma's case, faces.

When you mix many faces, even if they are not purely random [or fair] they do approximate even if some what biased.

Thank you, this is absolutely correct and the basis of a statistical representation of a population.

As I have already stated though, I do not have enough pictures of any population to do this exactly, but I have spoke to experts and my images are darn close. LOL So 100 images for 1 million people, actually I think is not needed. I would think 50 images would not show much of a differences, that is just my opinion after doing over 80 of these images and comparing them to the work of others.

so despite what mohistManiac, real research will continue on this subject, as it is a statistically legitimate exercise.

Edited by LongMa, 04 March 2011 - 09:05 AM.

[mohistManiac]

So it is now explaned that there is some quantifiable measuring going on. Still I have the doubt that the average face produced is any accurate representation of actual facial types that one comes across in common every day life because after all it is just a blending of the range of features of the population and artificially producing a result which doesn't particularly reflect what is actually seen. I may choose a photograph of a Chiwuawua and a Siberian Husky and any other that I choose to involve in the sample and produce an "average" but it won't and can't represent the fact that Siberian Husky are the most common amongst dog owners and should probably be what the guage is for actually knowing what is really present and available in some remote Siberian region unless you want to say a 2:1 ratio of Husky and Chihuahua imaging is somehow going to reveal the look of the new mutagenic dog breed.

Edited by mohistManiac, 04 March 2011 - 11:34 AM.

[LongMa]

So it is now explaned that there is some quantifiable measuring going on. Still I have the doubt that the average face produced is any accurate representation of actual facial types that one comes across in common every day life because after all it is just a blending of the range of features of the population and artificially producing a result which doesn't particularly reflect what is actually seen. I may choose a photograph of a Chiwuawua and a Siberian Husky and any other that I choose to involve in the sample and produce an "average" but it won't and can't represent the fact that Siberian Husky are the most common amongst dog owners and should probably be what the guage is for actually knowing what is really present and available in some remote Siberian region unless you want to say a 2:1 ratio of Husky and Chihuahua imaging is somehow going to reveal the look of the new mutagenic dog breed.

Two extremely phenotypically different dog breeds? Come on, be a bit intellectually honest in your opposition.

When you think in your mind of the "average" [insert ethnicity] it is a very general stereotype. Maybe focused a set of 3 or 4 specific facial features that stood out to you that you have seen quite often, that you associate with this group. This does not mean you don't see these features in some other combination or in total, elsewhere, it just means you see it very often in this group and associate it with that group. It also does not mean it is an "exact" match for a specific human being.

For example, for many Americans, they might say a stereotypical Italian is Robert DiNero...




Al Pacino




Sylvester Stallone



The reason they might hold this stereotype is because most Italian immigrants to American can trace their ancestry to poor Southern Italian farmers, who do indeed tend to look like the people above.

These three people have certain features in common, eye color, hair, nose shape, maybe a bit of a more narrow face that set them apart from the typical WASP American.


However in Switzerland, if you say what does an Italian look like, they might think immediately...of people from Lombard (where many Italian immigrants come from, and the people who share a dialect family with Italian speaking Swiss)...who look like this:

Alessandra Torresani



Anna Rita FIORONI





Frank Sinatra




The point is simple. The software is generating a stereotype based on common features that keep appearing...the same thing the human mind does. So it is generit, it is not supposed to be a "specific individual" bit it is supposed to represent the general stereotypical features of a group.

[mohistManiac]

Two extremely phenotypically different dog breeds? Come on, be a bit intellectually honest in your opposition.

When you think in your mind of the "average" [insert ethnicity] it is a very general stereotype. Maybe focused a set of 3 or 4 specific facial features that stood out to you that you have seen quite often, that you associate with this group. This does not mean you don't see these features in some other combination or in total, elsewhere, it just means you see it very often in this group and associate it with that group. It also does not mean it is an "exact" match for a specific human being.

For example, for many Americans, they might say a stereotypical Italian is Robert DiNero...




Al Pacino




Sylvester Stallone



The reason they might hold this stereotype is because most Italian immigrants to American can trace their ancestry to poor Southern Italian farmers, who do indeed tend to look like the people above.

These three people have certain features in common, eye color, hair, nose shape, maybe a bit of a more narrow face that set them apart from the typical WASP American.


However in Switzerland, if you say what does an Italian look like, they might think immediately...of people from Lombard (where many Italian immigrants come from, and the people who share a dialect family with Italian speaking Swiss)...who look like this:

Alessandra Torresani



Anna Rita FIORONI





Frank Sinatra




The point is simple. The software is generating a stereotype based on common features that keep appearing...the same thing the human mind does. So it is generit, it is not supposed to be a "specific individual" bit it is supposed to represent the general stereotypical features of a group.

I guess I can agree with the analysis that is being presented here but allow me to add just one thing on why it may be a little sketchy. If we are able to examine a lot of photographs of many Italian Americans and then examine a lot of photographs of many Italians living in the Tuscany region we may begin to get an understanding of the different types of looks that are featured in the corresponding areas. It is as you said things that brought attention towards association, something like a softer jaw or darker hair color etc. However what this averaging of faces does is it simply goes beyond that and gives you a sort of over arching scheme which displaces the validity of certain prevailing looks like a more prominent brow ridge versus none at all and gives you this absolutely in between blended look so that you assume that the constituents are to be found within this one master image and that is what I consider to be the false average. It is like saying the half man half woman image of Shiva is what a metrosexual man is going to derive his looks from as the best guesstimate or most people look androgenous anyways to begin with.

[William O'Chee]

I guess I can agree with the analysis that is being presented here but allow me to add just one thing on why it may be a little sketchy. If we are able to examine a lot of photographs of many Italian Americans and then examine a lot of photographs of many Italians living in the Tuscany region we may begin to get an understanding of the different types of looks that are featured in the corresponding areas. It is as you said things that brought attention towards association, something like a softer jaw or darker hair color etc. However what this averaging of faces does is it simply goes beyond that and gives you a sort of over arching scheme which displaces the validity of certain prevailing looks like a more prominent brow ridge versus none at all and gives you this absolutely in between blended look so that you assume that the constituents are to be found within this one master image and that is what I consider to be the false average. It is like saying the half man half woman image of Shiva is what a metrosexual man is going to derive his looks from as the best guesstimate or most people look androgenous anyways to begin with.

I agree with the work that LongMa is doing, and think it is valid.

As for your concern that by blending you take out certain prevailing looks, like a prominent brow ridge or whatever, I suspect this will not be an issue. If the feature is predominant in the majority of the population, then it will appear more strongly in the composite.

[mohistManiac]

I agree with the work that LongMa is doing, and think it is valid.

As for your concern that by blending you take out certain prevailing looks, like a prominent brow ridge or whatever, I suspect this will not be an issue. If the feature is predominant in the majority of the population, then it will appear more strongly in the composite.

In the case of some area which may be smaller and have less population relatively to another place which have larger overall I may have more confidence as you would to agree that the composite would be more finely tuned since you would be sampling a greater proportion of the population but the result where there are different trajectory looks in a relatively larger place and ratio of population might indicate considerable inaccuracy. The question is how cautious should one confine the limits of the area to be sampled and what would be the reasons. I don't believe weighting can be applied directly to represent a uniformity in many cases but only reflects the looks on the level of hypothetical.

[bloodmerchant]

From what I know, most Southern Italians and Sicilians are largely descended from Greek colonists (Sicilians are also descended from Siculo-Arabs and related to Maltese), due to their historical presence for about a millennium. Once the majority population in certain parts of Southern Italy, they have only dwindled to about 80,000 individuals (due to assimilation/Italianization). Though some speakers of Greek still remain, but are divided between Calabria and Salento, which meant that Greek speakers were once spread throughout much of Southern Italy and Sicily as late as the Norman period. (Though I do believe that Sicilian Greeks has been extinct as a distinct community since the 19th century, assimilated into Sicilian people- The last mention of the Sicilian Greeks was in the Catholic Encyclopedia, as a reference regarding the use of the Byzantine Greek rite in their Catholic churches:

In 1784 the Greeks of Sicily obtained from Pius VI an episcopus ordinans, resident at Piana dei Greci.

http://books.google....epage&q&f=false)



Greek-speakers in Italy

Edited by bloodmerchant, 05 March 2011 - 03:17 PM.

[LongMa]

From what I know, most Southern Italians and Sicilians are largely descended from Greek colonists (Sicilians are also descended from Siculo-Arabs and related to Maltese), due to their historical presence for about a millennium. Once the majority population in certain parts of Southern Italy, they have only dwindled to about 80,000 individuals (due to assimilation/Italianization). Though some speakers of Greek still remain, but are divided between Calabria and Salento, which meant that Greek speakers were once spread throughout much of Southern Italy and Sicily as late as the Norman period. (Though I do believe that Sicilian Greeks has been extinct as a distinct community since the 19th century, assimilated into Sicilian people- The last mention of the Sicilian Greeks was in the Catholic Encyclopedia, as a reference regarding the use of the Byzantine Greek rite in their Catholic churches:


http://books.google....epage&q&f=false)


Greek-speakers in Italy

This is true, there were also smaller amounts of Normans and Arabs (North African ones) who also controlled and mixed with them for a time.

For Northern Italians, the greatest concentration of people who lived there originally were Celtic tribes, who lived from South of the Alps all the way to Southern Germany and up the Atlantic coast to Brittan and Ireland. Then the area was Romanized, then came a large amount of Lombard Germans, who never left, they simply mixed into the population, and today the region of Lombardy is named after them.

This does help to explain the difference in average appearance of far Northern and far Southern Italians. In the North I believe there was also numerous Hungarian, Slavic, and Avars invasions too, but none of those groups stayed, besides rapes, I am not sure they left any genetic imprint.

That's kind of off topic though.

[Jeff_R]

is it possible that by making composited average faces for different groups some of the information is lost which otherwise would be important to discern between groups?

[adilima]

Is there a composite for Cantonese or HK Chinese as opposed to broad Han Chinese?

Just have a good look at my photo here, I have Cantonese blood from my mother's side

[xng]

Vietnamese look like southern chinese.

Thai looks like a mixture of southern chinese and cambodians (khmer)

Cambodians look like Malays/Indonesians.


http://i46.tinypic.com/o9fp7p.jpg

Edited by xng, 19 August 2012 - 09:15 PM.