5 ms·

>The weird thing being claimed here is that it actually doesn't matter which features you pick as long as 1) feature space is huge 2) you pick enough features 3

>The weird thing being claimed here is that it actually doesn't matter which features you pick as long as 1) feature space is huge 2) you pick enough features 3) you use a "correct" sparse representation of the features I haven't read the paper yet, but from your description this part is indeed the least convincing. It still matters which features you use! What if I pick the average color of the image (and millions of similar "features")? A feature like that won't help me in face recognition.

iamwil 19y ago
You're in computer vision, so you'd know more than me. Like you, I am a bit skeptical about it, but maybe this particular technique happens to work well enough for faces, due to the nature of face image data. Just like Bayesian classifiers assume independence when there isn't any, it works "pretty damn well". If you do get around to reading it, lemme know if you picked anything up from it that also smell fishy.
- pixcavator 19y ago
  What they say "features" they mean simply pixels! I have no idea why. So, if you have a collection 100x100 images, they are all points in a 10,000 space. This seems to be a common approach in pattern recognition. Now for face identification, the approach could only work under the assumption that people are "linearly independent": no face in the collection can be represented as a linear combination of the rest of the collection. It’s a bold assumption. If it is true, then the challenge is to make the algorithm efficient enough. The idea is that you don’t need all of those pixels/features and they in fact could be random. That must be the point of the paper.
  - pixcavator 19y ago
    The testing was done on two collections with several thousand images each. But the number of individuals in these collections was 38 and 114!
  - iamwil 19y ago
    Yeah, they simply mean pixels. I failed to mention that in my previous post. It seems odd to me that it works because doesn't representing it as a linear combo of pixels mean they throw out structural information of the image? I don't know how they came to make that assumptions, but could it be that the linear independence comes about because of the high dimensionality? The higher the dimensions, the harder it is to find a combination of sparse representations (linear comb) that gives you the sparse representation matching the test image. And if you do, and it all comes from the same person, then they say it matches. There must be some shape or property of the face subspace that allows for this, since potentially, you can match a duck to someone's face.
    - pixcavator 19y ago
      I don’t think they throw out structural information (like adjacency of pixels). Since each pixel corresponds to an independent dimension, the adjacency is still contained in those coordinates: (a,b,...) is not the same as (b,a,...). So, as long as all images have the same dimension, it’s OK. If you had both 100x100 images and 1x10,000 images in the collection, that would mess up everything! As we know from linear algebra, if a collection is linearly dependent, you can find a subcollection that adds up to 0! In other words, everything cancels out and you end up with a blank photo. Is it possible? If the dimension is too low (the images are small), yes. If the collection is large (a lot of images), maybe. What if it is small? (It is in fact!) In a very extreme case you may need the negative for each image to cancel it: same shape with dark vs. light hair, skin, eyes, teeth (?). Interesting stuff...

iamwil 19y ago

You're in computer vision, so you'd know more than me. Like you, I am a bit skeptical about it, but maybe this particular technique happens to work well enough for faces, due to the nature of face image data. Just like Bayesian classifiers assume independence when there isn't any, it works "pretty damn well". If you do get around to reading it, lemme know if you picked anything up from it that also smell fishy.

pixcavator 19y ago
What they say "features" they mean simply pixels! I have no idea why. So, if you have a collection 100x100 images, they are all points in a 10,000 space. This seems to be a common approach in pattern recognition. Now for face identification, the approach could only work under the assumption that people are "linearly independent": no face in the collection can be represented as a linear combination of the rest of the collection. It’s a bold assumption. If it is true, then the challenge is to make the algorithm efficient enough. The idea is that you don’t need all of those pixels/features and they in fact could be random. That must be the point of the paper.
- pixcavator 19y ago
  The testing was done on two collections with several thousand images each. But the number of individuals in these collections was 38 and 114!
- iamwil 19y ago
  Yeah, they simply mean pixels. I failed to mention that in my previous post. It seems odd to me that it works because doesn't representing it as a linear combo of pixels mean they throw out structural information of the image? I don't know how they came to make that assumptions, but could it be that the linear independence comes about because of the high dimensionality? The higher the dimensions, the harder it is to find a combination of sparse representations (linear comb) that gives you the sparse representation matching the test image. And if you do, and it all comes from the same person, then they say it matches. There must be some shape or property of the face subspace that allows for this, since potentially, you can match a duck to someone's face.
  - pixcavator 19y ago
    I don’t think they throw out structural information (like adjacency of pixels). Since each pixel corresponds to an independent dimension, the adjacency is still contained in those coordinates: (a,b,...) is not the same as (b,a,...). So, as long as all images have the same dimension, it’s OK. If you had both 100x100 images and 1x10,000 images in the collection, that would mess up everything! As we know from linear algebra, if a collection is linearly dependent, you can find a subcollection that adds up to 0! In other words, everything cancels out and you end up with a blank photo. Is it possible? If the dimension is too low (the images are small), yes. If the collection is large (a lot of images), maybe. What if it is small? (It is in fact!) In a very extreme case you may need the negative for each image to cancel it: same shape with dark vs. light hair, skin, eyes, teeth (?). Interesting stuff...