Google Releases Mendel Linux 4.0 Distribution for Coral Boards

Google company presented distribution update Mendel Linuxdesigned for use on boards CoralSuch as Developer Boards ΠΈ SoM. The Dev Board is a platform for rapid prototyping of hardware based systems. Google Edge TPU (Tensor Processing Unit) to accelerate operations related to machine learning and neural networks. SoM (System-on-Module) is one of the turnkey solutions for running machine learning related applications.

Mendel Linux distribution based based on the Debian package base and fully compatible with the repositories of this project (unmodified binary packages and updates from the main Debian repositories are used). The changes come down to building an image booted from eMMC cards and including components to support the hardware components of the Coral platform. Coral-Specific Components extend licensed under Apache 2.0.

MendelLinux 4.0 was the first release upgraded to Debian 10 ("buster"). The build is optimized for embedded systems and does not contain anything extra, including Debian 10 innovations related to SecureBoot and AppArmor support. Among the innovations, there is support for OpenCV and OpenCL, the use of Device Tree overlays, as well as an update to GStreamer, Python 3.7, the Linux 4.14 kernel and the U-Boot bootloader 2017.03.3.

Of the specific innovations, the ability to use the Coral GPU installed on boards (Vivante GC7000) is mentioned to accelerate the conversion of pixel data from the YUV to RGB color model with a performance of up to 130 frames per second for 1080p video, which can be useful when using boards for processing video from cameras , generating a stream in YUV format. To apply machine learning to process streaming video and audio on the fly, it is proposed to use an open framework MediaPipe. For example, based on to realize a system for recognizing and tracking objects or faces in the video transmitted from the surveillance camera.

Pre-built and pre-trained generic machine learning models compiled for the Edge TPU processors used on Coral boards continue to ship on project site, but are gradually being transferred to the general catalog of public models TensorFlow Hub. To simplify the development of your own solutions based on Coral and Mendel Linux boards, prepared guide, showing how to assemble a smart sorter from scrap materials that distributes colored and white balls into different baskets using a Raspberry Pi and a Coral USB Accelerator.

Source: opennet.ru

Add a comment