3 ms·

No, they use Magic Number among a bunch of various tricks. You can't really call it image recognition, more like 'data-hash comparison.' To train true image rec

No, they use Magic Number among a bunch of various tricks. You can't really call it image recognition, more like 'data-hash comparison.' To train true image recognition would require a LARGE subset of stored illegal pornography.

tomp 9y ago
Why? Just use two separate datasets, one of images of adults and children, to learn age, the other of porn and non-porn, to learn what's "nudity" or "sex".
- freeone3000 9y ago
  Knowledge transfer training is really hard. If you can do this (with different categories, for instance, determining what is nudity, then determining black people, to automatically classify porn with black actors automatically), it would be quite impressive. Results have nearly 5% error on CIFAR-10, which is full of clearly distinct categories - planes versus boats.
  - cr0sh 9y ago
    > it would be quite impressive Not only impressive, but likely very profitable. > Results have nearly 5% error on CIFAR-10 Any idea what the error rate is for human classification on CIFAR-10 images?
  - tomp 9y ago
    Can you explain (just a broad overview) why that wouldn't work? I mean, it's not like black porn actors are categorically different from black people in general... in particular, if you have an algorithm that can locate the face, you could just pass the face of the porn actor to the other algorithm to determine age/race/whatever...
salmonfamine 9y ago
What do you mean by "Magic Number" in this context?
- rarepostinlurkr 9y ago
  It's in the file format https://en.m.wikipedia.org/wiki/List_of_file_signatures https://en.m.wikipedia.org/wiki/List_of_file_signatures Also check out "libmagic"
obmelvin 9y ago
I mean if companies like Google train ML models on millions/billions of user's private data without looking at it, then surely it would be possible to train this in an ethical manner for ethical purposes. And via some Gov collaboration on the legal side
dairus 9y ago
You are confusing image recognition with classification. They use recognition, similar to reverse image search services. Not really 'data-hash comparison', but rather 'image-hash', since it may see through re-compression and other minor modifications.

tomp 9y ago

Why? Just use two separate datasets, one of images of adults and children, to learn age, the other of porn and non-porn, to learn what's "nudity" or "sex".

freeone3000 9y ago
Knowledge transfer training is really hard. If you can do this (with different categories, for instance, determining what is nudity, then determining black people, to automatically classify porn with black actors automatically), it would be quite impressive. Results have nearly 5% error on CIFAR-10, which is full of clearly distinct categories - planes versus boats.
- cr0sh 9y ago
  > it would be quite impressive Not only impressive, but likely very profitable. > Results have nearly 5% error on CIFAR-10 Any idea what the error rate is for human classification on CIFAR-10 images?
- tomp 9y ago
  Can you explain (just a broad overview) why that wouldn't work? I mean, it's not like black porn actors are categorically different from black people in general... in particular, if you have an algorithm that can locate the face, you could just pass the face of the porn actor to the other algorithm to determine age/race/whatever...

salmonfamine 9y ago

What do you mean by "Magic Number" in this context?

rarepostinlurkr 9y ago
It's in the file format https://en.m.wikipedia.org/wiki/List_of_file_signatures https://en.m.wikipedia.org/wiki/List_of_file_signatures Also check out "libmagic"

obmelvin 9y ago

I mean if companies like Google train ML models on millions/billions of user's private data without looking at it, then surely it would be possible to train this in an ethical manner for ethical purposes. And via some Gov collaboration on the legal side

dairus 9y ago

You are confusing image recognition with classification. They use recognition, similar to reverse image search services. Not really 'data-hash comparison', but rather 'image-hash', since it may see through re-compression and other minor modifications.