just from my reading of papers, optimized patterns make use of the fact that detectors (even modern transformer based architectures) are capital S Sensitive to the sillouette of an object.

in my other comment i linked a paper where they fool modern detectors with 80+% success rate. the clothing looks like a tiedye shirt to me. not fashionable, but not drawing attention either. if there is a way that you can make this fashionable is not a question i can answer :p

from my reading combining detectors (ensembling) does not help. this leaves spies with a problem, how to detect these noisy people without losing performance on normal nonnoisy people. there are tricks to this, but they are limited at best, for the same reason the noise worked in the first place. here is the reason.

assume noise makers create noise with methods A,B,C. you train on a dataset with images from the different noise makers (you don’t know which method they used). each of these noisy groups will have a distict sillouette you can detect, but together, you are building a function like:

pineapple: fruit bannana: fruit apple: fruit

this is a solvable problem for ml, so how do the noise makers win ultimately? randomization in the process. if each noise is distinct there is no training, you only have one image per type of fruit, you might as well use traditional CV to detect these noisy people (good luck).

My opinion is i want a product where the customer provides a random seed which generates the process of noise generation for their shirt. the product is the transparency into how the seed effected the noise making, and an evaluation of the noise on modern off the shelf detectors.

obviously we don’t see that here, but it seems doable, maybe a business idea for me? haha

this is the most recent arxiv paper on adversarial clothing (since a lot of people are speculating about if it’s possible): [2511.16020] Physically Realistic Sequence-Level Adversarial Clothing for Robust Human-Detection Evasion https://share.google/VMhtGB8P2kTMorGx6

they boast an 80%+ success rate at evading the instance level (each time you walk past an ai enabled camera) detection with sequential models (more difficult to fool as they see more of your sillouette).

not saying they are fashionable and obviously aren’t commercially available yet, but it definitely seems possible.

my main complaint is i wish these companies could make a good baseline so we could compare their efficacy. evaluating the trade off with the level of drip.

look at Mister 7 minutes over here. the Olympics called they want their stamina back

i keep putting books in front of my girlfriend like a moonstone, but it never works.

military contractors go brrr

I do get that this is the real differentiator to businesses in a chat app. how much it spies on employees. we need to unionize yesterday in tech

yeah dude, fuck you OP

it looped back atleast

maybe they were all shot down due to sampling bias? would have been B XVII .

a fulfilled life

maybe they shouldn’t have profited off and funded the war by buying discounted oil then

similar game play to goose game, but your a penguin, trying to break into the research station and act like you belong for as long as possible. as you always wanted to be an academic.

your penguin might do things like choose recordings from a tape recorder to convince your “colleuges” in a conversation (they sound like the peanuts teacher though) that you are one of them. wearing cute costumes to blend in (lab coat, twead sweater, glasses, etc). if you are caught doing penguin things (waddling, eating raw fish, etc) you are kicked out as the lab has a strict no penguins policy.

if your duration in the lab exceeds the top x % on a weekly leader board you get accepted in to a school and the game moves to a school setting. except your penguin is competing with a mix of other penguins (other players) and actual students (npcs) to stay in the program. you can collect recordings in your recorder and change costumes. you can understand other peguins, but the actual students again sound garbled. if your likeability score gets low enough it is discovered youre a penguin and you get thrown out, as the school has a strict no peguin policy. you are released in the icy wasteland and you have to start your academic career over. if you last long enough in school you graduate the game gives you an nft which can be submitted on a website to verify, which you can brag about in the real world! you have an nft!

if you log in after this your Penguin is stuck in an office with students saying garbled things, you play your recorder, they always accept what you say as you have tenure. you can look out the windows at children playing. raising the question did your penguin really want to get in to academia, to be accepted, or something else…

what do they think the force is that the military is projecting?

i declare security!

this is what i imagine if happening in those kinder coding classes.

“today class, we will be delving into the wonderful world of ternary operations”

“at my old school we never transformed ourselves into anthropamorphic logic gates”

“carlos!”

“/shrug”

hear me out, a show like picard but for ds9. this is what odos like?

is there any reason to think they’ve stopped?

A Related family has 600k take home. their kids go to athletic camps with trainers from professional teams in the off season. then go back to school and expectedly run circles around their peers. the difference in skill is always attributed to their work ethic or genetics…

Flamio, hotman.

this strategy seems to have worked well for Greenland and Canada. Grey rocking at an International scale.