On Wednesday, Google released AutoDraw, a web-based drawing tool, which pairs machine learning with drawings from artists to help everyone create comparatively more legible drawings. AutoDraw uses the same AI technology as Google’s QuickDraw. Doodles and sketches are matched with Google’s existing database of clipart, which are then given as suggestions to users to choose from, to help decide the closest image to what they were trying to draw.
It’s likely that choices users make will help improve Google AI with image recognition, in case of clipart. Cliparts are commonly used in sign boards. AutoDraw is a part of Google’s AI experiments. Its earlier experiments include, Giorgio Cam, which turns photographs into songs; QuickDraw, which tries to guess what the user is drawing; The Infinite Drum Machine, which organises everyday sounds and can be used to make beats; Thing Translator, which helps users identify an object in a different language using photographs; etc.
The app is free and works across desktops, tablets and smartphones and images are saved on the devices in Portable Network Graphic (.PNG) format. There is also a free draw tool that lets users create images without the AI intervening.
Google is pitching AutoDraw as a free and easier alternative to existing image creation and manipulation tools. Right now, AutoDraw can recognise hundreds of images and Google claims it will add more to its database in the future. Artists who’d like to contribute to the project, can do that here.