4 ms·
Engineers Test Highly Accurate Face Recognition
- jakewolf 19y agoThat would make tagging photos of people in social networks a cinch as you could check against photos of a user's friends. Facebook app?
- maximilian 19y agoThe only problem, as iamwil pointed out is that the algorithm doesn't work if it isn't given perfectly cropped face on photos. Its a "different problem" as they say, which i'd say is pretty valid.
- ed 19y agoHere's the paper: http://www.scribd.com/full/2359066?access_key=key-2ntxn9tb3uh1s4v7ie2t http://www.scribd.com/full/2359066?access_key=key-2ntxn9tb3u...
- iamwil 19y agoThis is rather neat, though the actual paper is less breathless than the wired article. I skimmed through it, and here's what I got: The algorithm assumes face images have been cropped and normalized, and only has been run on frontal views. Pose variation, detection, and alignment aren't handled by this algorithm--claimed to be a different problem. But it does handle different illumination and expressions, as well as occlusion and noise. According to No-free-lunch theorem, there is no generalized classifier that performs better than any other classifier on all classification tasks. Therefore, in any classification task, the solution is often tailored towards the problem, and picking features is one way to tailor it. Often times, you can't use all the features available to you, because they could be noisy or using them all can be computationally expensive. This is why people use SVD, PCA, or other techniques to filter out features that affect the classification the least, so you only do classification on features that affect classification the most. That's why people use some heuristics like picking eyes, mouth, noses as features, as it would seem like that's where most of the information is. 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 Apparently, they do recognition by trying to represent the new face as a linear combination of all known faces in the database. So if the new face belongs to a person in the database, it can be represented mostly by faces in the database known to belong to one person. If the new face belongs to someone not in the database, then we can only represent it as a combination of different peoples' faces in our database. In order for the surveillance mentioned in Wired to work, you'd still need to have an annotated database of people's faces built from somewhere. Same with auto-tagging people's photos on facebook. And you'd need more than one photo of a person too. However, given Louis Van Ahn's method of using games to entice people to do recognition tasks, and facebook tagging, you can create database of people's faces fairly easily.
- pixcavator 19y ago>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 agoYou'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 agoWhat 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 agoThe testing was done on two collections with several thousand images each. But the number of individuals in these collections was 38 and 114!
- 19y ago
- kurtosis 19y agohaven't read the paper but it reminds me of the work that's been done on single-pixel cameras and compressed sensing
- mynameishere 19y agoThis should obviate the need for RFID implants in our skulls.
- TheTarquin 19y agoAnyone have a guess as to whether or not this will pass the "bar fight test"? I go to the bar and start shit with Big McLargeHuge and he busts up my face. My nose is broken, one or both eyes are swollen, my jaw line is resting differently because of busted teeth or possibly a broken jaw. Can I still be recognized? People who know me will still be able to see who I am. A person comparing my face to a photo of me in an unbeaten condition would be able to recognize me. Will this software?
- iamwil 19y agoI think only if your beatup face retained the same ratio of colors as your normal face, if I'm not mistaken. It represents faces as a linear combination of pixels, so it throws away structural information.
- TheTarquin 19y agoInteresting. So it'd probably work alright if not too much of my face was bloody/bruised?