4 ms·
>> “Please upload a photo of yourself that clearly shows your face. We’ll check it and then permanently delete it from our servers.” >> To determine if the acc
by YeGoblynQueenne 9y ago
>> “Please upload a photo of yourself that clearly shows your face. We’ll check it and then permanently delete it from our servers.”
>> To determine if the account is authentic, Facebook looks at whether the photo is unique.
I assume this means that the photo itself is deleted but a one-way hash of it created to test against later.
However, if I change one pixel of a picture of my face, A, to produce a new picture of my face, A', the hash of A' will not be the same as the hash of A, correct? And I can repeat this process n x m times, where n, m the dimension of the image, ja?
Additionally- when they say "unique", do they mean "known as unique" to Facebook, or "unique in the entire world"? If I take a picture of myself and put it on, dunno, my blog on Blogger, what's stopping someone copying it and uploading it to Facebook to pretend it's me? Will Facebook search the entire web for images potentially matching an uploaded image?
For the record, I don't use Facebook. And there are no pictures of myself anywhere on the internets. As if.
- czardoz 9y agoIt doesn't have to be a hash really. They probably use some image processing techniques to process facial features and store the signature. I'm not sure how it identifies a human from a bot though.
- oh_sigh 9y agoHashes don't necessarily work on the raw bit representation of something. You can have a hash which works on higher level constructs, which for images one solution would be called a perceptual hash.
- thisacctforreal 9y agoI thought the accepted term for this was "fingerprint" not "hash".
- oh_sigh 9y agoHashes produce fingerprints(including perceptual hashes). Most people are talking about cryptographic hash functions when talking about 'hashing' but that is just one type. There is a wikipedia page for 'perceptual hashing' as well as a number of libraries available which claim to do perceptual hashing, so it may just come down to what kind of crowd you roll with :)
- YeGoblynQueenne 9y agoOK, thanks- I didn't know that. It looks like perceptual hashing can even identify a heavily modified version of an original (e.g. with noise added etc). So what I say about changing a single pixel would definitely not work. I'm sure it's still possible to steal a picture from someone else's online presence outside of facebook though and upload it as "unique" and facebook won't have any way to know that, unless they scan every other social site (and the rest of the internet, possibly). And someone could always take a picture of someone else's face, with a bit of social engineering, and use it to get access to their facebok- although I don't think that's the kind of thing facebook want to detect here.
- moduspol 9y agoTineye has been able to do this black magic for a while [1]. I understand what they're saying they do, but it's too high level for me to grasp what algorithms are being used. [1] https://www.tineye.com/faq#how https://www.tineye.com/faq#how EDIT: Part of the black magic. This doesn't explain how they'd stop someone from using a random picture of you on the net.