Hmm...

[Tervuren]

So lately I've been getting security log in that try to identify street signs, vehichles, etc...

Are these trying to get image recognition for self driving cars?

What is the purpose?

The down side is that any one that knows that only some of the images are known can fluster their way through clicking things that aren't really so.

It is using your time to solve problems rather than straight security, I can answer wrong to the ones that are problems to be solved and still get through.

I'm curious as to why the focus seems to be on street scenes, and if it relates to self driving cars the fact that I and others can create intentional errors is a serious problem to doing this.

[drkshdw]

Dunno what the purpose is of them other than to annoy the crap outta people. Another forum I used to frequent gave me one of those after signing in and after 20 pictures I quit. It was a never ending supply of pics. One or two? Sure. But 20+? Nope. Haven't been back to that forum for awhile.

[Tervuren]

Originally Posted by drkshdw

Dunno what the purpose is of them other than to annoy the crap outta people. Another forum I used to frequent gave me one of those after signing in and after 20 pictures I quit. It was a never ending supply of pics. One or two? Sure. But 20+? Nope. Haven't been back to that forum for awhile.

A few control images are used, those control images have to be passed correctly.

Filtered in with the control images are images that are "unknown", and they make money based on solving the unknown images for whoever their customer is.

Your time is being used to make judgements on the photo's for the company that runs the place's profit.

If they gave me a cut I might not have a problem with it.

The problem arises when a large amount of people realize they aren't being judged by the accurateness of every frame, just some verified control frames that are inserted with the unknowns.

This means you can half glute it and not spend the time to get it all right and still make it through. They try to layer it to make it harder to do this, but it can still happen. If enough people do this their results end up wrong.

I was curious why the focus is on vehicles and street signs, who wants to know?

Some one is paying to have these images put through an analysis to identify which ones match the question.

I to will avoid re-using sites that use this, it is a waste of my time that I do not get payed for that someone else is getting paid to waste my time.