Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Let artificial intelligence guess your attractiveness and age

faces.ethz.ch

Post reply on HN
Print view View original

121–130 of 139 posts

Re: Let artificial intelligence guess your attractiveness and age

#121

Post by timofter »

fatjokes wrote: post #97
rrothe wrote: post #14

Hi, I'm Rasmus and worked on the algorithms behind faces.ethz.ch. If you have any technical questions, let me know! Sorry for having some stability issues, we got much more traffic than expected and are currently working hard to fix everything!

Hi Rasmus, Cool work! I'm curious regarding the training dataset. What is the distribution of faces by race/age? Also regarding the raters, what is their distribution? (Race/age/cultural background)? It's widely known that attractiveness is heavily dependent on cultural upbringing. Signed, Butthurt dev whose best pic only rated an "OK". EDIT: You also rated Yoona (a Korean pop star) as just "Nice": http://imgur.com/u…

Hi fatjokes,

I am Radu, one of the authors.

As mentioned in our article (http://arxiv.org/abs/1510.07867v1), the training dataset for attractiveness is from Blinq. The underage and above 37 years old face images were discarded. All people are heterosexual in our training dataset and mainly from Zurich area, Switzerland. Therefore, our model of attractiveness fits the cultural bias of Zurich, Switzerland.

We consider faces.ethz.ch a little fun tool. I hope that the fans of Yoona will understand :) With more training data from abroad Switzerland our algorithms will fit their expectations.

For age and gender we used much more and diverse data, therefore are more reliable for the majority of ethnic backgrounds.

Re: Let artificial intelligence guess your attractiveness and age

#122

Post by timofter »

bjourne wrote: post #74
rrothe wrote: post #14

Hi, I'm Rasmus and worked on the algorithms behind faces.ethz.ch. If you have any technical questions, let me know! Sorry for having some stability issues, we got much more traffic than expected and are currently working hard to fix everything!

Could this site be used to optimize your dating profile photo (if you have one)? I'm probably "Hmmm" on most photos but possibly "Ok" or even "Nice" on a few. What does the algorithm want?

Yes, it is a foreseen application.

Re: Let artificial intelligence guess your attractiveness and age

#123

Post by timofter »

rootedbox wrote: post #33

this cat is 31 and is hot https://s3.amazonaws.com/uploads.hipchat.com/50837/483192/6w...

The first processing step consist from (human) face detection. We use the standard OpenCV for our faces.ethz.ch demonstrator. A failure of this step is likely to propagate in the unreliable/wrong attractiveness prediction. For attractiveness, age, and gender prediction we start from a cropped image assumed to contain a (roughly aligned) face as found by the detector.

I hope that this helps to understand the aforementioned result.

Re: Let artificial intelligence guess your attractiveness and age

#124

Post by timofter »

tsumnia wrote: post #113

Oh hey, I worked on something similar two years ago: http://www.FaceMyAge.com (note, the age estimator has been taken offline - because, 2 years ago). A lot of the issues our estimator (just an age estimator) ran into were the standard face recognition problems: occlusion, lighting, and (the obvious) bogus images. Anyone involved, what data set was the attractiveness scale built from(Labelled Faces in the Wild Datase…

Hi tsumnia,

I am Radu, one of the authors.

After 2 years we face almost the same issues, but probably we cope differently with them. Note that our solutions are fully described in the two papers mentioned on our faces.ethz.ch page. For attractiveness we used data from Blinq.

Our apparent age estimation solution is the winner of the latest LAP challenge, ICCV 2015.

Re: Let artificial intelligence guess your attractiveness and age

#126

Post by timofter »

sethbannon wrote: post #34

This would make a great psychology experiment. Use an algorithm to detect someone's age, then randomly assign them an attractiveness score and see how their behavior differs based on the result. How does the random attractiveness result effect how likely they are to share their score? To retake the test? Does this vary based on the users age?

Hi sethbannon,

I am Radu, one of the authors.

We thought at similar experiments, however psychology is not our expertise. If you check our paper on hotness/attractiveness (http://www.vision.ee.ethz.ch/~timofter/publications/Rothe-ar...), I am sure that you'll find some interesting results on how different age-grouped people rate, a paradox, and more. And yes, there are many interesting experiments to do and questions to answer.

Re: Let artificial intelligence guess your attractiveness and age

#128

Post by nice_byte »

freddealmeida wrote: post #127

This fails for Asian faces. Tested it on a picture of Bing Bing (Chinese Actress) that is absolutely stunning. She got a "Nice" rating. This is a toy. Check out how Sensetime did a similar feature.

They said it themselves - "Our algorithm is trained on the pictures of the BLINQ community that is mainly based in Switzerland. In other parts of the world the perception might be very different."

Re: Let artificial intelligence guess your attractiveness and age

#129

Post by novaleaf »

Dr_tldr wrote: post #114
verroq wrote: post #112

Earlier quoted context omitted.

If people are not mature enough to understand that a machine learning algorithm cannot be racist, shouldn't you go educate them instead of telling people not to offend these idiots? Secondly what makes you think that every race is equally attractive universally? Studies have people find people of their own race more attractive. If whites are more attractive (either by their training set or user base ratings) doesn't…

You have very poor reading comprehension skills, or else you appended your comment to the wrong post. Almost no part of what you're saying corresponds to anything that I said. PS: Giving some solid advice to public-facing startups looking for funding. Definitely take the autistic high-ground on every issue. Investors don't care about bad PR, they only care about abstract principles of truth as you understand them.

he was reading between the lines. A little too much of course, but people get pretty race-sensitive when you throw around look comparisons of black/white/whatever people.

Re: Let artificial intelligence guess your attractiveness and age

#130

Post by phlandis »

Your AI is impressive. Its ability to guess age was on the mark within a year or two on all of my tests. I am a not very well known male model and scored in the hot to godlike range with a mode of stunning. When I tested the AI on top male models in the industry I got consistent godlike and stunning ratings. The one problem with the AI could be the sensitivity to after editing. I am now sure how one would get past this issue.
Post reply on HN