Dev Kit Weekly: BeagleBone AI
This week we jump right into a descendant of one of the original open hardware maker pro development kits that has gotten a facelift for artificial intelligence applications – BeagleBone AI.
The BeagleBone AI is a performance-packed platform is based on Texas Instruments' Sitara AM5729 SoC, which integrates so much heterogeneous compute power it's actually tucked away underneath a heatsink on the board.
Dual TI C66x floating-point DSPs, four Embedded Vision Engines, a dual-core PowerVR 3D GPU, a Vivante 2D graphics accelerator, dual Arm Cortex-A15 cores, dual Arm Cortex-M4 cores, an H.264 video encode/decode subsystem, and two programmable real-time units (RTUs).Beyond just a long list of processing engines, all of that diverse compute makes the BeagleBone AI a perfect match for a wide range of neural networking applications.
For instance, high-speed video interfaces on the board can be used to pipe in video streams through the encode/decode subsystem and pass them on to a CNN running on one of the DSPs or GPU. Meanwhile, code or control logic running on one of the Arm cores or PRUs could trigger an action. So a BeagleBone AI-based home automation system could use facial recognition algorithms to detect "Brandon," the proceed to unlock the front door.
And because there's such a large, diverse array of compute on the TI device, BeagleBone AI can handle such use cases without really breaking a sweat. In addition to 2.5 MB of integrated on-chip RAM, the kit also includes an additional 1 GB of RAM onboard and 16 GB of high-speed eMMC flash. That's plenty of memory headroom for the frequent memory accesses required by even the largest, most demanding AI inferencing algorithms.
And a special treat that comes with the latest Beagle is that it is pin- and mechanically compatible with the original BeagleBone, meaning that any "oldie but goodie" capes (shields) and code should be able to make the leap to higher performance without issue. Only now you have access to tons of new high-speed I/O and wired/wireless connectivity.
From a software perspective, there's a Debian Linux image on the board as Beagle users have come to know and love and can be programmed using IDEs like Cloud9. But this member of the pack is also supported by TI's Deep Learning Library, which includes a set of OpenCL custom models that you can easily API into to accelerate AI workloads on the EVE processing engines mentioned earlier. The Deep Learning Library also provides Caffe models and an SDK for AI development.
All in all, the BeagleBone AI is a very good dog. It's a great opportunity for hardware and software engineers of all skill levels to immerse themselves in AI from whatever angle is most comfortable to them.
If you're interested in learning more about the BeagleBone AI, visit the BeagleBoard.org website at https://beagleboard.org/ai. There you can find a list of authorized distributors selling the dev kit for $125.
But if you want to save a few bucks, you have the opportunity to win this kit by entering the raffle here: https://opensysmedia.formstack.com/forms/dev_kit_weekly_kit_raffle
Questions for @techielew about the kit? Leave a comment.
Please like, subscribe, or follow us online.
See you next week...
Follow Brandon:
Twitter: https://twitter.com/techielew
LinkedIn: https://www.linkedin.com/in/techielew
Follow Embedded Computing Design:
Twitter: https://twitter.com/embedded_comp
LinkedIn: https://www.linkedin.com/groups/1802681/
Facebook: https://www.facebook.com/Embedded.Computing.Design/
Instagram: https://www.instagram.com/embedded_computing/
Видео Dev Kit Weekly: BeagleBone AI канала Embedded Computing Design
The BeagleBone AI is a performance-packed platform is based on Texas Instruments' Sitara AM5729 SoC, which integrates so much heterogeneous compute power it's actually tucked away underneath a heatsink on the board.
Dual TI C66x floating-point DSPs, four Embedded Vision Engines, a dual-core PowerVR 3D GPU, a Vivante 2D graphics accelerator, dual Arm Cortex-A15 cores, dual Arm Cortex-M4 cores, an H.264 video encode/decode subsystem, and two programmable real-time units (RTUs).Beyond just a long list of processing engines, all of that diverse compute makes the BeagleBone AI a perfect match for a wide range of neural networking applications.
For instance, high-speed video interfaces on the board can be used to pipe in video streams through the encode/decode subsystem and pass them on to a CNN running on one of the DSPs or GPU. Meanwhile, code or control logic running on one of the Arm cores or PRUs could trigger an action. So a BeagleBone AI-based home automation system could use facial recognition algorithms to detect "Brandon," the proceed to unlock the front door.
And because there's such a large, diverse array of compute on the TI device, BeagleBone AI can handle such use cases without really breaking a sweat. In addition to 2.5 MB of integrated on-chip RAM, the kit also includes an additional 1 GB of RAM onboard and 16 GB of high-speed eMMC flash. That's plenty of memory headroom for the frequent memory accesses required by even the largest, most demanding AI inferencing algorithms.
And a special treat that comes with the latest Beagle is that it is pin- and mechanically compatible with the original BeagleBone, meaning that any "oldie but goodie" capes (shields) and code should be able to make the leap to higher performance without issue. Only now you have access to tons of new high-speed I/O and wired/wireless connectivity.
From a software perspective, there's a Debian Linux image on the board as Beagle users have come to know and love and can be programmed using IDEs like Cloud9. But this member of the pack is also supported by TI's Deep Learning Library, which includes a set of OpenCL custom models that you can easily API into to accelerate AI workloads on the EVE processing engines mentioned earlier. The Deep Learning Library also provides Caffe models and an SDK for AI development.
All in all, the BeagleBone AI is a very good dog. It's a great opportunity for hardware and software engineers of all skill levels to immerse themselves in AI from whatever angle is most comfortable to them.
If you're interested in learning more about the BeagleBone AI, visit the BeagleBoard.org website at https://beagleboard.org/ai. There you can find a list of authorized distributors selling the dev kit for $125.
But if you want to save a few bucks, you have the opportunity to win this kit by entering the raffle here: https://opensysmedia.formstack.com/forms/dev_kit_weekly_kit_raffle
Questions for @techielew about the kit? Leave a comment.
Please like, subscribe, or follow us online.
See you next week...
Follow Brandon:
Twitter: https://twitter.com/techielew
LinkedIn: https://www.linkedin.com/in/techielew
Follow Embedded Computing Design:
Twitter: https://twitter.com/embedded_comp
LinkedIn: https://www.linkedin.com/groups/1802681/
Facebook: https://www.facebook.com/Embedded.Computing.Design/
Instagram: https://www.instagram.com/embedded_computing/
Видео Dev Kit Weekly: BeagleBone AI канала Embedded Computing Design
Показать
Комментарии отсутствуют
Информация о видео
1 ноября 2019 г. 2:56:49
00:06:57
Другие видео канала
BeagleBone AI Review - Setup and First ImpressionsMeet the dazzling flying machines of the future | Raffaello D'AndreaDev Kit Weekly: Power Integrations' BridgeSwitch RDK-852How computers learn to recognize objects instantly | Joseph RedmonBeagleBoard Comparison | Arrow.comBeagleBone AI Hands-on - T2sday Updates #25010 Steps To Self Learn Embedded Systems Episode #1 - Embedded System Consultant ExplainsI Built an A.I. Voice Assistant using PyTorch - part 1, Wake Word DetectionTop 5 New SBCs 2020High Torque Cycloidal Drive (NEMA 23 Motor)The Best Single Board Computers Of 2020 Top 5 ARM SBC’sSmall Brain, Big Think: AI on the EdgeMaking a Digital Dashboard! (w/ Google Calendar integration)AI Plays Hide and Seek...Surprises Creators (Bonus Scenes)Animatronic #Terminator Skull with #BeagleBone® AIRaspberry Pi 8GB Review - Should you upgrade?Raspberry Pi 4B vs Jetson NanoFrom MCU to MPU - Beaglebone Black Wireless - Thoroughbred or Heinz 57 (Introduction)Building a Car Interface Using Raspberry Pi 4 | Bluetooth Sync With OBD II AdapterBeaglebone: Adding USB Wi-Fi & Building a Linux Kernel