4 ms·
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 g
by rrothe 11y ago
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!
- frik 11y agoYour website is offline, do you have a paper about your technology?
- mkuhn 11y agoIt is back right now and they list the research.
- sawwit 11y agoFrom a data privacy point of view I would very much appreciate if you added a prominent "delete upload" button to the result page.
- rndmind 11y agoYes a "delete" button would nice.
- rrothe 11y agoWe do not store any photo on the server, so no worries!
- tlrobinson 11y agoWell, you could still add a delete button to make people feel better (I'm only half joking)
- TeMPOraL 11y agoAnd as a potential user I'm only half-insulted. But seriously, what purpose would it serve over a promise in TOS? A normal user will trust both (except the button would be bullshitting them). A privacy-obsessed one will trust neither.
- the-dude 11y agoAKA https://en.wikipedia.org/wiki/Placebo_button https://en.wikipedia.org/wiki/Placebo_button
- jmorrow977 11y agoYou could make a delete button which merely opens a popup to explain that they do not save it in the first place. That way if they're looking for a delete button, they'll find it, but it will give them the right information.
- Jach 11y ago"We do not save the uploaded image." from the take-this-seriously? popup.
- anotheryou 11y agoWhat does it analyze? Picture or Extracted Face?
- bjourne 11y agoCould 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?
- timofter 11y agoYes, it is a foreseen application.
- MrQuincle 11y agoHi Rasmus, you described your paper on age estimation on the site. Do you also have something written on the hotness scale? :-) Edit: Sorry, missed it! http://arxiv.org/pdf/1510.07867v1.pdf http://arxiv.org/pdf/1510.07867v1.pdf
- Madmallard 11y agoNone of the images I am uploading are working. It is saying it cannot detect a face on any of them. Is this a masking of the stability error or some other issue?
- otakucode 11y agoIt is trying to let you know that you are not quite human.
- fatjokes 11y agoHi 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/uJVnQ9S http://imgur.com/uJVnQ9S. I guess that makes me feel better about my "OK". I'd stay out of Korea if I were you---I hear their fans aren't very forgiving.
- timofter 11y agoHi fatjokes, I am Radu, one of the authors. As mentioned in our article (http://arxiv.org/abs/1510.07867v1 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.