Aren't you really glad now you uploaded all these photos to facebook?
I didn't. I joined to keep up with a few friends and family, and they uploaded photos with me in them, nicely tagged.
DeepFace: Closing the Gap to Human-Level Performance in Face Verification
21–30 of 46 posts
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#22Important to note: - they still need 1000 labeled samples per identity - their network can only handle 4000 distinct identities (at 97.25% accuracy) at a time It's still a very worrying development for online and offline privacy.
Actually, based on my reading of the paper, it seems that they learn a representation using one data set (the one with 1000 labeled samples per identity), then use that representation to classify on other training sets (like the Labeled Faces in the Wild, which has 13,323 photos of 5,749 celebrities). In fact, from what I can tell from section 5.1, they seemed to use face pairs (and so trained on 1 sample per person,…
I think you guys are reading different parts of the article.
They present both results, supervised and unsupervised (where unsupervised uses the SFC dataset to train). They achieved 95.92% accuracy LFW with unsupervised (section 5.3) - so they can train on SFC and then classify a single image in a different domain with 95.92% accuracy.
They achieved the 97.25% accuracy level was achieved as you say, when they let the pairs into the training set. But they overfit with LFW alone, and has to add an additional 100k identities with more samples (30) per identity. A very impressive measure, but not quite as good as being able to generalize with 97% accuracy from a single training photo.
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#23Important to note: - they still need 1000 labeled samples per identity - their network can only handle 4000 distinct identities (at 97.25% accuracy) at a time It's still a very worrying development for online and offline privacy.
Actually no. For one thing, this isn't exactly a stealthy or cheap thing to do. It involves datacenters full of computing resources even for 4000 identies. I also don't believe it's so much a privacy issue. If I upload pictures to facebook I actually want them to be seen by human beings. The face-recognition only helps with that. If facebook recognizes me on a picture someplace else, I actually rather want to know ab…
> For one thing, this isn't exactly a stealthy or cheap thing to do.
That's actually the more disturbing part about this: it require computational resources that only governments and large corporations can afford, so they are the ones who gain the most from it; and it gives them more leverage over the population.
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#24The key innovation is an accurate, reliable method for rotating faces so they're 'looking straight at the camera' before feeding them to a deep neural network. They call this 3D photo rotation process "frontalization." Figure 1 on page 2 of the paper shows at a very high level how this is being done. Very nice!
Yeah, generally speaking for object detection - Scale, position, rotation: typically you can identify something based on 2 of these really efficiently (for example, wavelets give you a transform that is sufficient for object identification under scale and position transformations in O(N) operations).
If you'd rather have orientation identification (ie. rotation angle) and scale in that mix but don't care about position, the radon transform is nice and easy to work with.
But beyond inverting 1-2 key transformations, one usually has to pay a pretty hefty computational cost which often precludes online (near real-time) use.
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#25Earlier quoted context omitted.
Privacy issue is not about what you want, it's about what can be done (and often is without you knowing). Automated facial recognition is a serious privacy concern, it's not just about the slimy despicable thing facebook is. For example, in the uk where there are more cctv video feed than people to watch them, an automated facial recognition can track you around constantly. Their current automatic number plate recogn…
This can already be done by someone who truly wants to keep an eye on you, only it takes more manpower (well, detective power). This is simply the same trend as ATMs replacing (some) bank tellers many years ago. Will it be easier to perform and abuse mass surveillance? Sure. Will people with something important to hide still wear disguises? I'd bet yes. My stance is that we can't fight progress, but we can start figh…
>Will people with something important to hide still wear disguises? I'd bet yes.
Facial recognition systems can be (and are) trained to not be fooled by things like facial hair, hair changing, eyeglasses and sunglasses, hats, etc. Although it can obscure your features for sure.
If someone does something really odd to try to avoid the facial recognition (I've people posting things like making your hair non-symmetrical, or just avoiding the camera in the first place) then they just train on those things too. And that person gets flagged as being really suspicious. And things like full face masks are banned in some places already.
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#26Reducing error by 25% from a 96.5% baseline gives you their stated 97.25% accuracy. About 0.75% fewer errors. Still amazing, but less impressive than the abstract makes it sound.
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#27It's surprising to me that there isn't a single positive comment in this thread considering how amazing this is. Sure, it has other implications, but were we really hoping to prevent computers from recognizing faces permanently?
No, we were just hoping to prevent creeps like facebook from using it to tailor advertising.
Ah, there's how you get to minority report type stores/advertising being used in practice...
1. You upload photos to Facebook. Facebook detects various commercial products or even--assuming deep learning can take things to "higher level" representations--"style preferences" of people recognized in the photo. It will also develop style preferences based on context, location, etc. (dive bar or classy lounge?) Because it knows who is in the picture, etc. it can correlate that with an identity.
2. Facebook then sells this "identity" to the Gap -- no actual information about the user, just these raw vectored "style preferences" which contain all knowable brand information about each user. Facebook can provide massive coverage here.
3. You walk into a Gap store. Gap has installed software provided by Facebook to detect your face/person/style preferences (but no personally identifiable information, just that "the person with this face has these preferences and probably makes this much money so you might offer X,Y,and Z at these prices") and you then get an offer via facebook message (or "facebook offer"?) on your phone to buy what you think is actually a really cool jacket at an admittedly reasonable price (based on what you're used to paying for jackets).
This probably has massive ramifications for outfits like Costco or Target or Walmart where individual consumer preference/taste/whatever can really make a difference in choosing effective lineups of products... Maybe they manage to offer deals that sort of price themselves based on what they know the user will pay?
Almost not a bad idea...
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#28Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#29Earlier quoted context omitted.
I didn't. I joined to keep up with a few friends and family, and they uploaded photos with me in them, nicely tagged.
you can disable their ability to do so.
No, you can't stop them from uploading photos and mentioning your name, even if they didn't specifically put a rectangle on your face.
Re: DeepFace: Closing the Gap to Human-Level Performance in Face Verification
#30Earlier quoted context omitted.
Actually, based on my reading of the paper, it seems that they learn a representation using one data set (the one with 1000 labeled samples per identity), then use that representation to classify on other training sets (like the Labeled Faces in the Wild, which has 13,323 photos of 5,749 celebrities). In fact, from what I can tell from section 5.1, they seemed to use face pairs (and so trained on 1 sample per person,…
I think you guys are reading different parts of the article. They present both results, supervised and unsupervised (where unsupervised uses the SFC dataset to train). They achieved 95.92% accuracy LFW with unsupervised (section 5.3) - so they can train on SFC and then classify a single image in a different domain with 95.92% accuracy. They achieved the 97.25% accuracy level was achieved as you say, when they let the…
I looked hard, and could not prove they can identify a single image with that accuracy. I couldn't tell WHAT the trial protocol was. Where is that in the document?