Following the release of its voice control kit earlier this year, Google has produced a new DIY kit for the Raspberry Pi which lets hobbyists turn the compact board into a computer vision system – or in layman’s terms, a smart camera capable of recognizing stuff.
The AIY (do-it-yourself AI) Vision Kit consists of a VisionBonnet circuit board, a lens kit, a button and speaker, along with a cardboard outer case and various connecting bits and pieces (including a tripod mounting nut).
When combined with a Raspberry Pi Zero W, a Raspberry Pi Camera, plus an SD card (and power supply), the user can build their own powerful computer vision system which has no need to hook up to the cloud for extra processing power. It’s a fully self-contained cardboard box of tricks.
The time for assembling the kit is estimated at around an hour.
Recognizing the benefits
It comes with software that can pull off feats like successfully recognizing common objects from a database, and recognizing human faces and their expressions (happy, sad, confused-why-are-you-pointing-that-camera-at-me, that sort of thing).
There’s a third piece of software (all of these utilize TensorFlow-based neural network models) which can be used as a person, cat or dog detector.
So, for example, you could set up a system that detects when a person enters the room, or when your pet appears at the back door, wanting to be let in because it has started raining.
The AIY Vision Kit will be unleashed at the end of this month, retailing at $45 (around £33, AU$60) in the US, with pre-orders now live over at Micro Center.
As we mentioned previously, Google also released an AIY Voice Kit back in May, which has many uses including potentially driving a voice-powered robot, or making your own digital assistant. That’ll set you back £25 (around $34, AU$45) over at Pimoroni.
The Raspberry Pi is nothing if not versatile, and has even been used – in very large quantities – to build a ‘supercomputer’ of sorts.
Via: The Verge
- Wondering what else you can do with your Raspberry Pi? Check out our projects