If you’re crap at drawing, artificial intelligence (AI) technology may be able to turn you into Artgem… one day. For now, it can convert your crude cat doodles into amusing monstrosities. Try it out for yourself.
There’s unlimited potential for machine learning and neural networks for the development of smarter artificial intelligence (AI). For example, you can use them to make computers recognise objects (which is harder than it sounds) for a range of different applications.
Google decided to open source its machine learning system TensorFlow in 2015 and since then a number of people have played around with it to create their own AI experiments.
One of these people is software engineer and cofounder of Pushbullet Chris Hesse, who ported an image-to-image translation model called pix2pix to Tensorflow and made it work in a browser.
He created a few image translation demos for several objects. One of them is called edges2cats, which is trained on around 2000 stock cat photos and can convert a line drawing of a cat into something that somewhat resembles the real thing.
It’s a great demonstration of what is possible with machine learning, but it has its limitations.
“Some of the pictures look especially creepy, I think because it’s easier to notice when an animal looks wrong, especially around the eyes”, Hesse admits on his blog. “The auto-detected edges are not very good and in many cases didn’t detect the cat’s eyes, making it a bit worse for training the image translation model.”
With that said, edge2cats will spit out images that looks somewhat decent if you feed it more details.
If you’re only giving it crude lines to work with though, you get something like this:
And if muck around with it, you get something that looks like it’s from a HP Lovecraft novel:
Hesse also created similar demos for shoes, buildings and handbags. You can try it out for yourself here. It’s recommended that you use the Google Chrome browser to make the most of these demos.