
Machine learning tool ported to Raspberry Pi 4
Edge Impulse in the US now supports a wider range of processors for embedded machine learning applications with the Raspberry Pi 4 board. Developers can use the existing support for low-power MCUs or venture into processor classes that run embedded Linux if highest performance is the objective.
The Raspberry Pi 4 board, developed in the UK, uses the 1.5GHz Broadcom BCM2711 multicore processor with four ARM A72 cores and supports embedded Linux with portable container technology that gives access to AI applications that also run in the data centre.
“We’ve brought the same great user experience our developers are already familiar with into the Linux domain (using full hardware acceleration on the Pi 4), with a refreshed set of tools and capabilities that makes deploying embedded machine learning models on Linux as easy as … Pi,” says Zin Thein Kyaw, Lead User Success Engineer at Edge Impulse.
“[This] announcement from Edge Impulse is a big step, and makes machine learning at the edge that much more accessible. With full support for Raspberry Pi, you now have the ability to take data, train against your own data in the cloud on the Edge Impulse platform, and then deploy the newly trained model back to your Raspberry Pi,” said Alasdair Allan, Technical Documentation Manager at Raspberry Pi.
Edge Impulse also announced the launch of support for true object detection as part of its computer vision ML pipeline. Users can use a Raspberry Pi camera or plug in a standard USB web camera into one of the available USB slots on the Pi, and harness the raw power of higher performance compute and more sophisticated frameworks and libraries to facilitate computer vision applications.
For audio applications, says the company, users can plug a standard USB microphone into one of the available USB slots on the Pi. For sensor fusion, the 40-pin GPIO header on the Pi can be employed to connect to their favorite sensors as well.
To get started, the company offers a Raspberry Pi 4 guide. In addition, an object detection tutorial explains how to easily train an object detection model. SDKs for Python, Node.js, Go, and C++ are provided so users can easily build their own custom apps for inferencing.
Related articles:
- Edge cloud boost for Raspberry Pi 4
- Raspberry Pi 4 Compute simplifies power pinout
- Edge AI partnership combines neural sensor SoC, TinyML platform
- Microchip and ML startups partner to simplify AI-at-the-edge design
Other articles on eeNews Europe
- Europe falls behind in chip design market
- Free custom AI boards for Google Coral chip
- Mipsology signs Taiwan AI deal
- Vulkan Video adds hardware extensions for portable multicore codecs
