IS GOOGLE USING RECAPTCHA TO FIND B2 BOMBERS?

if you get a new captcha click on the squares that are not the plane to fuck it up

oy vey delet this

>reading my tabs
>not whoising the website to maximum spook me
Step it up, shillpai.

Google compares your results to those of others, so we all have to click in the same spot or else it gets filtered out. We need to organize like we did with operation nigger.

The difference is now that they're doing it with things that we could reasonably make a fuss about without looking like autists.

>mfw
Well then, time to get my passport and try to apply for US citizenship.

holy fug

THREAD HAS BEEN DELETED ON Cred Forums BY IT'S KEKOLD MODS. THE SHILLS ARE WINNING.

ive seen this captcha before, thought it was weird. why would google need captcha to identify B2s though im sure they can make some sort of shape recognition software without having to use captcha

>shape recognition software to identify planes specifically designed to not be recognized
>this is what tech illiterates actually believe
It's easier and cheaper to have normal people do it.

but..but how does clicking the b2 on a captcha help weapons research?

SOMETHING'S AFOOT!!!!

oh nevermind i think i might understand now, is it to help the ai learn to recognize b2s and shoot them/everything but them down?

SOMETHING'S REALLY AFOOT.

It will help the US military to censor any photos of it.
Any blog that houses said pictures will suddenly disappear from the search listings

Wouldn't you want pictures of enemy stealth planes if you were doing weapons research? I think it's kind of self explanatory.

I don't really know what motivation they have for it, but mechanical turking pictures of b2 bombers is pretty fucking spooky.

OY VEY THE GOYIMS KNOW

Thanks Poland. One shekel has been deposited into the mutual anti-Nazi defense fund.

This thread is only here because many just went back to using legacy captchas.
Nice try.

They've been doing this for a while. There was a bit of an op (for the lulz) on Cred Forums a few years ago to make google/recaptia falsely OCR words as "nigger". It was pretty fun. The first word was the reference word, and the second was the word that the machine couldn't identify. So so many words are "nigger" in google's book database now.

You think that's scary? Look at the frames it sends to your screen.

Happened to me yesterday, I thought that picture was a common picture from an aviation forum or something
>Using their recaptcha to train the Machine Learning models is kind of an ingenious move.
They been doing this since Recaptcha v2

...

...

how do we fuck new captcha??

The theory is something like the following:

-Nations want to develop non-radar-based methods of tracking planes in the sky, since radar waves can be detected by those planes and traced back to their source. They are also subject to various forms of jamming or spoofing.

-One such method would be high-resolution passive "visual" analysis using super-sensitive ground-based cameras. Like space telescopes, but designed to search the terrestrial zone rather than areas beyond the atmosphere itself.

-But object recognition in the form of "machine vision" is actually pretty hard; much harder than radar analysis. You need lots and lots of data from various angles to get to the point where the machine is good enough to indicate to you when a hostile plane might have entered your airspace.

-Farming out the machine vision work to Cred Forums NEETs and other captcha users speeds up the process.

In theory. But who knows. I'm not a machine vision specialist.

>tfw you have to select all

You dont. Cred Forums should stop using google services though.

designed to not be recognized by radar, it has a very distinct visual shape

but you can get high quality pictures and videos of b2 bombers anywhere, why would google need us pointing them out in shitty pictures? the realization i had here seems more likely but still a little far fetched

It's not so much about identifying B2 bombers specifically as much as it is about machine learning. Machine vision (or machines seeing things like you and I) is an extremely difficult thing to teach a computer. So far, we've only been able to teach machines the generalities of sight. We've moved on to using AI algorithms to extend the capabilities of machine vision. Basically, yes... we're participating in teaching machines how to "see". And yes, Google has openly stated that this what they're using your recaptia data for.

tl;dr: I hope you're having fun teaching Skynet how to identify humans, so that fucker can blast your asperger's-having ass.

...

The point isn't for high-res images, and you know this, as you literally mention the point of having a large number of images.

When you want to train a neural network, you do so with a massive image library, and if you want it done well, you do it with a massive library that closely resembles what you're looking for but just isn't, like birds with long wingspans. Google knows damn good and well about neural networks, since several of theirs are open source.

My brother worked on developing a program for his company that used machine sight to identify faulty car parts, i will have to bring this up to him sometime

I'm sure he already knows about neural networks and their business applications. They aren't esoteric knowledge. They're ted-talk tier common.

Really yeah of course they are!

...