News for hackers. Stuff that matters.

A comment on Using Principal Component Analysis on Images.

Re:Using Principal Component Analysis on Images by got2surf on Monday August 17, 2015 @05:47PM (#10076369)
Really interesting, does anyone know more information on how the initial eigendresses were computed? Was there image processing to distill the dress into major components, hand tagging of variables (like short/long, color, etc), or just raw pixel data processed?
[ Parent | on HN ]
Re:Using Principal Component Analysis on Images by blackle on Monday August 17, 2015 @05:49PM (#10076376)
I'm thinking it's just raw pixel data, and the description of each eigendress's meaning was author's interpretation
[ Parent | on HN ]
Re:Using Principal Component Analysis on Images by geb on Tuesday August 18, 2015 @12:10AM (#10077563)
The only manual classification used was separating dresses into "like" and "dislike" categories. The rest is just pixel data.
[ Parent | on HN ]
Re:Using Principal Component Analysis on Images by thisjepisje on Monday August 17, 2015 @06:10PM (#10076471)
You might wanna do a search for "eigenfaces" (as opposed to "eigendresses" :P)

https://en.wikipedia.org/wiki/Eigenface

[ Parent | on HN ]
Re:Using Principal Component Analysis on Images by afarrell on Monday August 17, 2015 @06:32PM (#10076570)
Reading that, I can't help but wonder how similar a machine-learned eigenface is to an infant-learned one. http://lawcomic.net/guide/?p=3273

I wonder if there is some way to simulate the effect of growing up among people of different races on eyewitness identification and test if it matches what this lawyer is saying at http://lawcomic.net/guide/?p=3282

[ Parent | on HN ]
Re:Using Principal Component Analysis on Images by dspeyer on Monday August 17, 2015 @11:45PM (#10077508)
As I understand it, eigenfaces require preprocessing to orient and align the images. That doesn't seem to have come up here. Maybe Amazon is really consistent about centering models in dress photos?
[ Parent | on HN ]
Re:Using Principal Component Analysis on Images by dthal on Monday August 17, 2015 @11:02PM (#10077391)
>> Was there image processing to distill the dress into major components, hand tagging of variables (like short/long, color, etc), or just raw pixel data processed?

Its pretty clear that this is PCA on raw pixels of the images, so no, there is no hand-tagging or anything else. The eigendresses are just the eigenvectors from PCA projected back into the original, image-scale, basis.

[ Parent | on HN ]