Why is Google so paranoid about identifying crosswalks? Can't they just GPS that shit?

Why is Google so paranoid about identifying crosswalks? Can't they just GPS that shit?

Attached: jal.jpg (405x720, 29K)

>he doesn't have a pass

>he has a pass

Attached: 1535395650954.jpg (674x674, 102K)

>Why is Google so paranoid about identifying crosswalks?
Because they believe that they can detect the differences between how machines spot crosswalks and humans spot crosswalks.

We are training their automated cars. It is important that they are able to recognise crosswalks

>he doesn't monaterily support his face Webby site

>he has a pass

Attached: Autism_levels.gif (332x187, 1007K)

who's this waifu?

Because if their automated cars can't detect crosswalks properly and hit a single person then it would make self-driving cars look really bad.
It is vitally important that their cars do not accidentally hit someone, hence the extreme crosswalk detection.

Is that the roflcopter?

Captcha is a good study on mob rule. They're never gonna tell the difference between a traffic sign and a sign that just so happens to be near the street.

Get a job, NEET.

They already know that humans, unlike machines, are assholes, user. They're not using your inputs to train any production systems; they're checking to see if you do the human thing or the right thing.

its so that when the AlphaBET Chromium androids are set loose to kill everyone and replace them with new robot copies they will know where they can cross the road safely.

But they'll still struggle with storefronts...

>It is vitally important that their cars do not accidentally hit someone, hence the extreme crosswalk detection.

B-but i've been missing at least one crosswalk on each captcha and it passes me anyway. Will this kill people?

Attached: nerv.png (320x295, 119K)

Big mistake. Every captcha you solve is linked to your IP address. If a google car runs over a person and the captcha which caused it to incorrectly verify the situation is linked to your IP you will be charged with murder.

T. Brainlet attempting to become proto-intellect.

You do realize that Google collects millions of solved captchas and cross-verifys them to delineate the erroneous answers. In other words "ONLY" a "SINGLE" solved captcha isn't used to train their neural networks.

Autism

>that heli entering the nose of the plane
Comfy

Kill youre self

i don't understand how solving captcha provides them with training data. It's not like it resolves even when you click on the wrong boxes so there is obviously something that knows what the correct boxes are. So then why do they need humans input for data when they could just generate their own.

Attached: 99134ff6c071ad88dad24f4e426b80cefa6d1425ae451fd53eccdc1d1b10f905.jpg (640x360, 91K)

i want to bury my face between her breasts

Retard

My guess is that they give people a couple ones with verified answers, to see how well a person performs and then a couple captchas they don't know the answer to.

I also noticed not all captcha's require a 100% correct answer. So I assume google knows some squares are correct and is uncertain about some other squares.

>literally paying to shitpost

Say the 3x3 grid contains 4 "correct" images and Google knows for sure that 3 of them are correct. Having enough people (who don't actively disrupt their captchas) do captchas will eventually lead to the majority of people picking the "uncertain" image, giving the system confidence that this image indeed contains what was requested.

sidewalks never require a correct answer

mommy!