Google Adds Vision to Raspberry Pi

PDF Print E-mail

Google provides means for customers to add its AI-powered vision to the Raspberry Pi-- the AIY Vision Kit, a package consisting of a circuit board, computer vision software and additional accessories.

AIY Vision kitThe AIY Vision Kit allows one to turn the Raspberry Pi into a more hackable version of Google's own Clips smart camera. The main component is the Vision Bonnet, a circuit board packing a Movidus MA2450 vision processing chip acting as a "neural network accelerator." It does not need to connect to the Google Cloud, meaning all images are processed locally on the device.

On the software side the kit offers 3 neural network models-- one able to recognise 1001 common objects, one able to recognise faces and expressions, and a "person, cat and dog detector." Google hopes bolder users will further improve on the neural models, for instance by adding the ability to recognise rabbits for a project used in a rabbit hutch.

The kit also includes a piezo speaker, macro/wide lens kit, tripod nut and other connecting components, and the box acts as an outer shell holding the final device together. What users need to add are a Raspberry Pi Zero W, Pi Camera, SD card and power supply. It should be available by end 2017.

Go Introducing AIY Vision Kit