New Google AIY kit

Vision Bonnet board. © Google.

AIY Vision Kit. © Google.

Last May, Google released its first AIY Project kit stemming from its partnership with the Raspberry Pi Foundation, dubbed Voice HAT. Yesterday, Billy Rutledge, Google’s AIY Project Director, showed off Google’s new image recognition kit. The “Vision Kit” is a do-it-yourself build that creates an image recognition device able to see as well as identify objects. It relies on TensorFlow, Google’s open-source tools for automatic learning. In addition to the kit, you’ll need a Raspberry Pi Zero W board, a Camera Module V2, a power source, and an SD card. The main component of the kit is a Vision Bonnet accessory board for Raspberry Pi. The board comes equipped with an Intel Movidius MA2450 processor—a unit for low-power vision processing that can run neural network models. The software includes three neural network models: one that can recognize up to a thousand everyday items, one that can recognize faces and expressions and one that is a “person, cat and dog detector.” Users can even train and retrain their models with TensorFlow. It brings to mind many possible applications, like a device that could recognize your kitty meowing to be let in, and send you a notification. The kit will be available in select stores in early December, retailing for $45 USD.

⇨ Google, The Keyword, “Introducing AIY Vision Kit: Make devices that see.”