By Ben Everard. Posted
You can run neural networks on your Raspberry Pi, but they’re demanding and can suck up a lot of your processor time – especially if you’re using a Raspberry Pi Zero. This HAT lets you offload some of the processing of neural networks to an additional processor, the Kendryte K210. This isn’t as powerful as some of the other neural network accelerators – running at about 0.5 TOPS. But it doesn’t suck up too much electricity and is convenient to use.
Pre-trained models will be available for object detection and face detection, but if you want to dig into TensorFlow, you can create your own. While TensorFlow development is quite an involved affair, you can run existing models without too much difficulty from Python. If you’ve already got a heavy workload on your Raspberry Pi, this is a great way to add a little extra compute power to enable AI.
$38, from crowdsupply.com
xTool Screen Printer review
Make T-shirts with a laser cutter
TOPS - Traverser of planar surfaces
Aaed Musa tells us how they made their robotic pet
Video Game Module for Flipper Zero review
A module that’s not really for video games
TypeWryter
Apparently, it’s easier to write when you’re not constantly distracted by funny videos of baby animals
Moving Pixel Clock
Erich credits Leoni Etter, Jan Rohrer, and Livio Stadelmann for their help in building this bonkers device
HOMS microprocessor
If the chip shortage strikes again, maybe we’ll end up using something like this
Print subscriptions to HackSpace magazine start from just £4, saving you a big chunk of money on the cover price. You’ll get exclusive subscriber offers, and more projects, ideas, and inspiration than you can shake a toolbox at. Upgrade your ideas — subscribe to HackSpace magazine!
