TensorFlow Lite for Microcontrollers Benchmarked on Linux SBCs

TensorFlow Lite microcontrollers benchmark linux SBC

Dimitris Tassopoulos (Dimtass) decided to learn more about machine learning for embedded systems now that the technology is more mature, and wrote a series of five posts documenting his experience with low-end hardware such as STM32 Bluepill board, Arduino UNO, or ESP8266-12E module starting with simple NN examples, before moving to TensorFlow Lite for microcontrollers. Dimitris recently followed up his latest “stupid project” (that’s the name of his blog, not being demeaning here :)) by running and benchmarking TensorFlow Lite for microcontrollers on various Linux SBC. But why? you might ask. Dimitris tried to build tflite C++ API designed for Linux, but found it was hard to build, and no pre-built binary are available except for x86_64. He had no such issues with tflite-micro API, even though it’s really meant for baremetal MCU platforms. Let’s get straight to the results which also include a Ryzen platform, probably a laptop, for reference: SBC Average for 1000 runs  (ms) Ryzen 2700X (this …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

Waveshare Jetbot AI Kit for NVIDIA Jetson Nano Board Sells for $100 and Up

NVIDIA Jetbot

NVIDIA Jetson Nano developer kit was launched for $99 last March with impressive specifications for the price including one module with four Arm Cortex-A57 cores, a 128-core Maxwell GPU, and 4GB LPDDR4 RAM. The company also introduced Jetbot robot based on the new board, with all instructions available on Github, but until now you had to put some efforts to build it up as the fairly long list of parts had to be purchased or 3D printed separately. It’s now become easier, as Waveshare has started selling their Jetbot AI kit on Amazon for $122.99 without Jetson Nano, and $259.99 with, meaning you may be better off purchasing NVIDIA board separately for around $100, or you may purchase the complete kit directly on Waveshare website for $215.99 plus shipping, or $99.99 without the SBC. Jetbot AI kit content: Optional NVIDIA Jetson Nano SBC 64GB micro SD Card Metal box Camera holder Acrylic piece Jetbot expansion board with battery holder, motor …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

Linux 5.2 Release – Main Changes, Arm, MIPS & RISC-V Architectures

Linux 5.2 Changelog

Linus Torvalds announced the release of Linux 5.2 last Sunday: So I was somewhat pre-disposed towards making an rc8, simply because of my travels and being entirely off the internet for a few days last week, and with spotty internet for a few days before that [*]. But there really doesn’t seem to be any reason for another rc, since it’s been very quiet. Yes, I had a few pull requests since rc7, but they were all small, and I had many more that are for the upcoming merge window. Part of it may be due to the July 4th week, of course, but whatever – I’ll take the quiet week as a good sign. So despite a fairly late core revert, I don’t see any real reason for another week of rc, and so we have a v5.2 with the normal release timing. There’s no particular area that stands out there – the changes are sosmall that the appended …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

AAEON Launches BOXER-8150AI Compact Embedded Box PC Features 8 USB 3.0 Ports

BOXER-8150AI

The new compact embedded solutions’ addition to the BOXER-8100AI family is specially designed for artificial intelligence (AI) support. According to its makers, AAEON, it has a configuration of eight USB 3.0 ports made available through four USB controllers chips. This makes it perfect for a wide range of computer vision applications, some of which are visual inspection, quality control, and facial recognition. Now, each USB chip controls a pair of USB 3.0 ports. By dedicating one chip to one pair of ports, each chip handles less data traffic overall.   This also improves bandwidth from connected devices and allows the BOXER-8150AI to maximize speeds and framerates of AI image processing. The product is currently in stock and available for purchase on AAEON online store for $1,142.00. Key Features of BOXER-8150AI Nvidia TX2 Module HMP Dual Denver 2 + Quad ARM A57 8GB LPDDR4 32GB eMMC 5.1 HDMI 2.0 8x USB 3.0 1x GbE LAN 1x COM Compact Size chassis The …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

Linux 5.1 Release – Main Changes, Arm, MIPS & RISC-V Architectures

Linux 5.1 Changelog

Linus Torvalds has just announced the release of Linux 5.1: So it’s a bit later in the day than I usually do this, just because I was waffling about the release. Partly because I got some small pull requests today, but mostly just because I wasn’t looking forward to the timing of this upcoming 5.2 merge window. But the last-minute pull requests really weren’t big enough to justify delaying things over, and hopefully the merge window timing won’t be all that painful either. I just happen to have the college graduation of my oldest happen right smack dab in the middle of the upcoming merge window, so I might be effectively offline for a few days there. If worst comes to worst, I’ll extend it to make it all work, but I don’t think it will be needed. Anyway, on to 5.1 itself. The past week has been pretty calm, and the final patch from rc6 is not all that …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

ZED Depth and Motion Tracking Camera Supports NVIDIA Jetson Nano Board

ZED depth camera Jetson Nano

When NVIDIA launched their low cost Jetson Nano development board earlier this week, one reader asked whether it would support binocular depth mapping. It turns out Stereo Labs has updated the SDK (Software Development Kit) for the ZED depth and motion tracking camera in order to support the latest NVIDIA developer kit. Jetson Nano can manage depth and positional tracking at 30 fps in PERFORMANCE mode with 720p resolution, and while the more powerful and expensive Jetson TX2 achieves doubles the performance at 60 fps, it does so at a much higher cost. ZED depth and motion tracking camera specifications: Video 2.2K @ 15 fps (4416×1242 resolution) 1080p @ 30 fps (3840×1080 resolution) 720p @ 60 fps (2560×720 resolution) WVGA @ 100 fps (1344×376 resolution) Depth Resolution – Same as selected video resolution Range – 0.5 to 20 m Format – 32-bits Stereo Baseline – 120 mm Motion 6-axis Pose Accuracy Position – +/- 1mm Orientation – 0.1° Frequency – …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

NVIDIA Introduces $99 Jetson Nano Developer Kit

Jetson Nano

NVIDIA Tegra X1 octa-core Arm processor with a 256-core Maxwell GPU was introduced in 2015. The processor powers the popular NVIDIA Shield Android TV box, and is found in Jetson TX1 development board which still costs $500 and is approaching end-of-life. The company has now introduced a much cheaper board with Jetson Nano Developer Kit offered for just $99. It’s not exactly powered by Tegra X1 however, but instead what appears to be a cost-down version of the processor with four Arm Cortex-A57 cores clocked at 1.43 GHz and a 128-core Maxwell GPU. Jetson Nano developer kit specifications: Jetson Nano CPU Module 128-core Maxwell GPU Quad-core Arm A57 processor @ 1.43 GHz System Memory  – 4GB 64-bit LPDDR4 @ 25.6 GB/s Storage  – microSD card slot (devkit) or 16GB eMMC flash (production) Video Encode – 4K @ 30 | 4x 1080p @ 30 | 9x 720p @ 30 (H.264/H.265) Video Decode – 4K @ 60 | 2x 4K @ 30 …

Support CNX Software – Donate via PayPal or become a Patron on Patreon

BOXER-8120AI is a Compact Jetson TX2 Mini PC for Drones, Robots and Surveillance Applications

Jetson TX2 Mini PC

AAEON has just launched BOXER-8120AI compact mini PC based on NVIDIA Jetson TX2 processor module with 8GB RAM, 32GB storage, and four Gigabit Ethernet ports. The fanless mini PC targets smart surveillance/ security/ parking, unmanned stores, drones and robotic controllers, or any applications that can leverage Jetson TX2’s 256 CUDA cores for A.I. workloads. BOXER-8120AI Jetson TX2 Mini PC specifications: Processor Module – NVIDIA Jetson TX2 with HMP Dual Denver 2 + Quad Arm A57, NVIDIA Pascal GPU with 256 CUDA cores, 4K (HEVC) video encoder, 4K 12-bit video decoder System Memory – 8GB LPDDR4 @ 59.7 GB/s Storage – 32GB eMMC 5.1 flash, MicroSD slot Display Interface – HDMI 2.0 type A Networking – 4x 10/100/1000Base-TX Ethernet USB – 2x USB 3.0 ports, 1x micro USB OTG port Serial – 2x COM ports (DB9) Misc – Power Button, Power LED, 2x SMA Holes, Remote Power On/Off Power Supply – 9~24V DC-in power input via lockable DC jack Dimensions – …

Support CNX Software – Donate via PayPal or become a Patron on Patreon