Frequently asked questions
What is image annotation?
Image annotation is the act of selecting a part of an image to highlight some "thing" in the image.
For example, if you draw a box around a car in an image we can use that to collect lots of nice examples of what cars look like.
From these boxes we can make a dataset.
What is a datset?
Let's make a cars dataset. First we need some images of cars, actually quite a lot of images.
Usually a customer will have these images already, for example they might take them from their website, app or social network.
Then we annotate these images by drawing boxes around any cars to highlight them.
A cars dataset is a whole bunch of images, with the cars highlighted, and no mistakes. It's really important not to have any mistakes.
Why is it important to not have mistakes ?
Once we have the cars dataset we show it to a machine, one that tries to learn to recognise things like us, it's called a neural network and we tell it:
"Look!, all these images are cars, this is what cars look like!"
And, so long as we dont have any mistakes, the machine with get it! It will be able to recognise cars in images now, just like we would teach a child.
If we make a mistake, like if we show it a bus instead of a car, it can get confused. We certainly dont want confused machines driving us around.
What kind of person are you looking for?
Honestly, we need all kinds of people to create our network. Just by being different you can give different point of views and this alone is valuable.
At this stage in our growth we are looking for some specific kinds of people. There is so many thimgs to do in a startup company like ours that we prefer to find annotators to have some other skills and passions.
You can find your own personal level here, you can be involved, you can contribute. Most people here started as annotators. You can totally go as far as you want too.
However, even if thats not the case, we always need dependable, productive and hardworking people.
Who uses this data?
Compnaies that have or are making artificial intelligences or machine intelligences.
For example, IBM Watson, who once won the US gameshow Jeopardy, uses this data to be able to recognise brand logos in images.
So, if you show it the Adidas logo it will be able to recognise it.