AICAN Does Not Require Human Interference To Paint Like Picasso
AI has boomed onto the art sector over the last few years, with everyone from tech giants to artists testing with the new tools that tech offers. While the GANs (generative adversarial networks) that fuel the likes of Google’s BigGAN are able to create outstandingly odd images, they need a huge amount of human guidance and interaction. Not so with the AICAN (Artificial Intelligence Creative Adversarial Network) system designed at Rutgers University’s AI & Art Lab by Professor Ahmed Elgammal and his associates. It is an almost autonomous system skilled on 500 Years worth of Western arts that makes its own readings of these classic strokes.
The AICAN uses the same adversarial network framework as GANs, but it operates them differently. Adversarial networks work with 2 sets of nodes: one set creates pictures on the basis of the visual training data set that it was offered while the second set evaluates how closely the created picture resembles the real pictures from the training info.
On a related note, if you have ever required to swiftly take out the background of a picture you know it can be tiresome, even with authorization to tools such as Photoshop. Well, Remove.bg is a sole-purpose site that employs AI to do the tedious job for you. Just upload any picture and the website will automatically verify any individual in it, remove the background, and allow you get a PNG with a transparent foreground.
It is the latest instance of how ML methods that were once radical are being converted into easy consumer equipments. In the case of eliminating the background of an image, there are already a couple of open-source algorithms that can manage this specific job. Remove.bg has just converted them into an online free utility. Other akin products comprise LetsEnhance.io that employs AI to automatically upscale images and Deepart.io that applies the style of one picture to another.