Digital cameras are old news. Instant film cameras are back in vogue, but hardly inspiring. The market needs something fresh, and that something may be doodles. A newly detailed camera powered by artificial intelligence brings a twist to the photography industry, snapping “doodles” instead of actual images. The camera is a DIY project involving a Raspberry Pi and some Google magic.
The camera project was created by Dan Macnish, who detailed the creation on his website. He describes the device as a polaroid camera, one re-imagined with a fun purpose: creating crude doodles instead of printing images. The device is powered by a Raspberry Pi coupled with a thermal printer, which produces instant prints on thermal paper.
At the heart of the project is Google’s Quick, Draw! project, which the company revealed a couple years ago. As part of it, the company had the public draw items as basic sketches, contributing data that helped train its AI. That game is still available for anyone to try here. According to the site, more than 50 million drawings have been created by users.
An object recognition system is used to enable the camera to determine what is present within any given image. After determining the object in the image, the Google doodle dataset is used to translate the photo into a child-like doodle. That doodle is then printed onto thermal paper as an instant photo.
Instructions on creating the camera are provided on GitHub.
SOURCE: Dan Macnish