Q made a post about how to get around AI image recognition by changing the corners of the image, don't remember the post number or the exact percentage of the image that had to deviate from the norm for it to work but I think it was at least 25%...
Yes, and some tests were done using twitter, fb and even tineye (tineye is actually the strongest one it seems).
The original method was to cover the corners and shrink the image, after some testing what was determined was that one could apply a filter over the image, as long as it covers 25% of the "features" (a jargon term) with an alpha of 6% or more (alpha is transparency where 0 = invisible and 1 = opaque). Then reuploading the image will show that it is a unique image.
Others have made scripts that can run through memes and apply that overlay to camouflage memes. It does work, and would require manually removing because the AI would recognize it as completely separate from the banned image.
I do not know what the AIs are doing exactly, I suppose no-one does :)
But I'd assume they run ie. OCRs on an image and process the extracted texts.
If there's gibberish, it just might not get blocked automatically. It sill might trigger a manual review, but that is a very resource consuming process.
FB definitely OCRs any image meme that I submit and looks for key words. If it finds any, it puts up the relevant warning flag to discourage or even prevent posting. So I have two strategies:-
If I put up an image meme with text, I smudge the key words so the OCR doesn't work but the human eye can figure it out. Or I use a typeface that OCR can't recognise. Or I use a word with similar meaning that won't trigger the A.I.
If I type actual text, I replace characters such as "i" and "a" and "v" and "o" and "u" with equivalent Greek characters. This makes the text unintelligible to A.I. but it remains perfectly readable to a human. Example:
Normal: The antivirus vaccination is dangerous.
With Greek: The αntινιrυs ναccιnαtιοn ιs dαngerουs.
(To get a "v" in Greek type "n".)
Misspelling might also work. Be inventive. For "vaccination" try "faksee nayshun" or similar.
I'm not convinced they can read text. If you have Google photos, you can search for photos. Like if you search "car", it will show all pictures in your Google photos cloud that have a car in it. I have thousands of memes stored, and many are just text images. When I search for a keyword in a meme, say "pandemic", Google is not able to display the meme with that word in it. I'm not convinced OCR is being used widespread.
I don't believe that any AI system can extract meaning from memes (at least not yet, or widely known).
Think about a common meme like the superhero button choice. You see 2 buttons with text in one frame and the stress drops and expression on the heroes face, and automatically make the connection of the choice to be made between 2 options and the implication that one button MUST be pressed.
For the computer; it can classify that the image matches one that was learned. It could possibly extract the text above the buttons. It could classify the buttons as buttons. It could classify the person. It could classify the drops as water. BUT, the currently available / known AI has no means of applying those traits to any specific meaning.
From what I hear from people at fakebook, not only Zuck has invested a lot of money and effort into this, he has made huge strides. It still requires human intervention, but the AI can filter out potential candidates very quickly and the humans can provide confirmation - which goes back to train the AI.
From what I gather outside of this insider insights is that, this technology is part of those military / CIA tech thats usually 20-30 years ahead of whats released publicly.
Putting the two together, I have to believe they have far more than we can imagine right now
Yes, they have probably put up billions in research and hardware for training.
As advanced as AI gets, it is still fundamentally based on digital representations of reality. They can get statistically strong scores but will individually produce some inexplicable results.
In other words, they can be hacked and manipulated.
For example; they did some tests on a visual AI system, you want to know what the computer believes the difference is between a wolf and a German shephard? The wolf has snow in the background.
Another example that fooled facial trackers was a miniature test image in the frame ( a known test image) this fooled the AI into believing it was in the testing phase.
And again, as with Tineye, adding a random transparent overlay on a known image fools the system into thinking it is unique.
While these systems are very good, none known are capable of taking a movie screenshot to determine the contextual meaning and then adding that meaning to a text in the image.
Sometimes as little as a singular pixel can make the difference between an image of a panda with 95% confidence to a boat with 99% confidence.
Statistically good, individually flawed results. That's why camouflage will continue to work for the foreseeable future.
OCR and getting meaning from text is one thing, but change the text to something a leftist would say and an image whose meaning would imply the text is stupid and it would require a human (unless it exists) OR apply camouflage to an existing meme and it can be reused.
Another option would be to make the text unreadable to OCR, jagged lines with a 6% alpha should work.
It absolutely can read text. Even screen clipping tools have the ability to "copy text from images", and that's not even AI. It just looks for pixels that make the shapes of letters, instead of Bridges or motorcycles. Easy stuff nowadays.
That's awesome. Now, kindly tell us all what blindtext is, how to put it in, and we will all be on the same page. Not all of us are tech savvy. Some of us have other occupations....
Its not on anyone here to spoon feed you solutions and tbh that mind set is the exact opposite of this entire site. Save this post and come back.. and read solutions.
You need to at least try or your no different then the left sitting there waiting for programming to be fed in by other people willing to do the work.
How do you know anyone here isn't leading you astray? You don't. Which is why you do the work.
Q made a post about how to get around AI image recognition by changing the corners of the image, don't remember the post number or the exact percentage of the image that had to deviate from the norm for it to work but I think it was at least 25%...
Yes, have used this technique with success on facebook to get around censorship
Yes, and some tests were done using twitter, fb and even tineye (tineye is actually the strongest one it seems).
The original method was to cover the corners and shrink the image, after some testing what was determined was that one could apply a filter over the image, as long as it covers 25% of the "features" (a jargon term) with an alpha of 6% or more (alpha is transparency where 0 = invisible and 1 = opaque). Then reuploading the image will show that it is a unique image.
Others have made scripts that can run through memes and apply that overlay to camouflage memes. It does work, and would require manually removing because the AI would recognize it as completely separate from the banned image.
I think it was 20%, these drops?
https://qalerts.app/?n=4660
https://qalerts.app/?n=4663
Camo thread, talks about the 20% corners, layering on patterns,
https://8kun.top/qrmemes/res/606.html
https://greatawakening.win/p/12hkd4PY0K/for-those-pedes-who-are-making-m/c/ https://derpa-block.azurewebsites.net/
http://overlay.imageonline.co/
https://www.gimp.org/downloads/ free image editor, does fonts, layers, transparency
Say more please
I do not know what the AIs are doing exactly, I suppose no-one does :)
But I'd assume they run ie. OCRs on an image and process the extracted texts. If there's gibberish, it just might not get blocked automatically. It sill might trigger a manual review, but that is a very resource consuming process.
Can you explain it like I'm 5 and don't know what a computer is?
What is is?
FB definitely OCRs any image meme that I submit and looks for key words. If it finds any, it puts up the relevant warning flag to discourage or even prevent posting. So I have two strategies:-
If I put up an image meme with text, I smudge the key words so the OCR doesn't work but the human eye can figure it out. Or I use a typeface that OCR can't recognise. Or I use a word with similar meaning that won't trigger the A.I.
If I type actual text, I replace characters such as "i" and "a" and "v" and "o" and "u" with equivalent Greek characters. This makes the text unintelligible to A.I. but it remains perfectly readable to a human. Example:
Normal: The antivirus vaccination is dangerous.
With Greek: The αntινιrυs ναccιnαtιοn ιs dαngerουs.
(To get a "v" in Greek type "n".)
Misspelling might also work. Be inventive. For "vaccination" try "faksee nayshun" or similar.
OCR = Optical Character Recognition
Putting random gibberish as in the OP's picture can prevent it being flagged.
Oh, I see. Thank you very much. Makes a lot more sense now. I appreciate your help.
I'm not convinced they can read text. If you have Google photos, you can search for photos. Like if you search "car", it will show all pictures in your Google photos cloud that have a car in it. I have thousands of memes stored, and many are just text images. When I search for a keyword in a meme, say "pandemic", Google is not able to display the meme with that word in it. I'm not convinced OCR is being used widespread.
Ah the AI that runs on your Google Photos is very different than the AI that runs on their server detecting meaning from memes.
I don't believe that any AI system can extract meaning from memes (at least not yet, or widely known).
Think about a common meme like the superhero button choice. You see 2 buttons with text in one frame and the stress drops and expression on the heroes face, and automatically make the connection of the choice to be made between 2 options and the implication that one button MUST be pressed.
For the computer; it can classify that the image matches one that was learned. It could possibly extract the text above the buttons. It could classify the buttons as buttons. It could classify the person. It could classify the drops as water. BUT, the currently available / known AI has no means of applying those traits to any specific meaning.
From what I hear from people at fakebook, not only Zuck has invested a lot of money and effort into this, he has made huge strides. It still requires human intervention, but the AI can filter out potential candidates very quickly and the humans can provide confirmation - which goes back to train the AI.
From what I gather outside of this insider insights is that, this technology is part of those military / CIA tech thats usually 20-30 years ahead of whats released publicly.
Putting the two together, I have to believe they have far more than we can imagine right now
Yes, they have probably put up billions in research and hardware for training.
As advanced as AI gets, it is still fundamentally based on digital representations of reality. They can get statistically strong scores but will individually produce some inexplicable results.
In other words, they can be hacked and manipulated.
For example; they did some tests on a visual AI system, you want to know what the computer believes the difference is between a wolf and a German shephard? The wolf has snow in the background.
Another example that fooled facial trackers was a miniature test image in the frame ( a known test image) this fooled the AI into believing it was in the testing phase.
And again, as with Tineye, adding a random transparent overlay on a known image fools the system into thinking it is unique.
While these systems are very good, none known are capable of taking a movie screenshot to determine the contextual meaning and then adding that meaning to a text in the image.
Sometimes as little as a singular pixel can make the difference between an image of a panda with 95% confidence to a boat with 99% confidence.
Statistically good, individually flawed results. That's why camouflage will continue to work for the foreseeable future.
Not true. I created my own custom meme (based on a more poorly done Biden meme) and Facebook scanned it with OCR in order to apply a fact check on it.
I believe you misunderstood the point.
OCR and getting meaning from text is one thing, but change the text to something a leftist would say and an image whose meaning would imply the text is stupid and it would require a human (unless it exists) OR apply camouflage to an existing meme and it can be reused.
Another option would be to make the text unreadable to OCR, jagged lines with a 6% alpha should work.
Weren't there also Q or FBIanon posts about this?
It absolutely can read text. Even screen clipping tools have the ability to "copy text from images", and that's not even AI. It just looks for pixels that make the shapes of letters, instead of Bridges or motorcycles. Easy stuff nowadays.
yea can confirm. check out google translate.
you can take an foggy photo of text and it will be able to copy it down.
How do you know that image doesn't have meta text being read?
Any distraction is good, I think, whether it's text or other pictures.
This dicussion originated from https://greatawakening.win/p/12iNZU99ns/x/c/4E0w7fDR5Ik - if you have one or multiple FB accounts, give it a real-time try and report here.
this exact image (without modification) got me banned today for misinformation LOL, I'll have to adjust my methods
That's awesome. Now, kindly tell us all what blindtext is, how to put it in, and we will all be on the same page. Not all of us are tech savvy. Some of us have other occupations....
Its not on anyone here to spoon feed you solutions and tbh that mind set is the exact opposite of this entire site. Save this post and come back.. and read solutions.
You need to at least try or your no different then the left sitting there waiting for programming to be fed in by other people willing to do the work.
How do you know anyone here isn't leading you astray? You don't. Which is why you do the work.
Then what?