Featured Open Source AI Hardware You Should Consider

Artificial intelligence (AI) is a computer system or other machine able to perform tasks that ordinarily require human intelligence like planning, learning, reasoning, problem solving, knowledge representation, perception, motion, manipulation and even social intelligence and creativity.

To extend the reach of AI hardware, Seeed provides a wide selection of AI hardware focusing on Machine Learning and Computer Vision, Edge Computing, Speech Recognition & NLP and Neural Networks Acceleration to allow anyone to build AIOT projects. Here is our pick of pieces of kit worth your attention based on function:

Machine Learning

There is a broad body of research in AI, much of which feeds into and complements each other.

Currently enjoying something of a resurgence, machine learning is where a computer system is fed large amounts of data, which it then uses to learn how to carry out a specific task, such as understanding speech or captioning a photograph.

By Nick Heath | February 12, 2018 — 11:23 GMT (19:23 GMT+08:00)
Topic: Managing AI and ML in the Enterprise

Machine learning has been widely used in data processing, computer vision, natural language processing, biometrics, search engines, medical diagnostics, detection of credit card fraud, securities market analysis, and robotics. But what is it exactly?

• Machine learning is a part of the artificial intelligence research field, more specifically, in how to improve the performance of specific algorithms in empirical learning.
• Machine learning is the study of computer algorithms that can be automatically improved through experience.
• Machine learning is the use of data or past experience to optimize the performance standards of computer programs.

With the explosive growth of connected devices, combined with a demand for privacy/confidentiality, low latency and bandwidth constraints, AI models trained in the cloud increasingly need to be run at the edge.
MAIX is Sipeed’s purpose-built module designed to run AI at the edge. We call it AIoT. It delivers high performance in a small physical and power footprint, enabling the deployment of high-accuracy AI at the edge with a competitive price that makes it possible to embed into any IoT device. Sipeed MAIX is quite like Coral board, but it is lower cost and lower power.

Seeed is developing a Grove HAT for Edge Computing based on the Sipeed MAIX-I module, aiming at enabling more possibilities in areas such as predictive maintenance, anomaly detection, robotics and many more. 

It is an ideal way to start Machine Learning studies with NVIDIA Jetson Nano.

The NVIDIA® Jetson Nano™ Developer Kit delivers the computing performance to run modern AI workloads at an unprecedented size, power, and cost. Developers, learners, and makers can now run AI frameworks and models for applications like image classification, object detection, segmentation, and speech processing. Find more information at the NVIDIA Jetson Nano’s official page. 

Check our Unboxing of Jetson Nano & Quick Start-Up of Two Vision Demos

Coral from Google – Another hot piece of kit tailored to AI machine learning

The Coral Dev Board is a single-board computer with a removable system-on-module (SOM) that contains eMMC, SOC, wireless radios, and Google’s Edge TPU. It’s perfect for IoT devices and other embedded systems that demand fast on-device machine learning inferencing.

You can use the Dev Board as a single-board computer for accelerated machine learning processing in a small form factor, or as an evaluation kit for the onboard SOM. The 40 mm × 48 mm SOM on the Dev Board is available at volume and it can be combined with custom PCB hardware using board-to-board connectors to integrate into products.

The SOM is based on NXP’s iMX8M system-on-chip (SOC), but its unique power comes from the Edge TPU coprocessor. The Edge TPU is a small ASIC designed by Google that provides high-performance machine learning inferencing with a low power cost. For example, it can execute state-of-the-art mobile vision models such as MobileNet v2 at 100+ fps, in a power efficient manner. 

Edge Computing

Edge computing brings memory and computing power closer to the location where it is needed.

Edge computing
From Wikipedia, the free encyclopedia

Select the development board for your specific edge computing application:

• Processing and analyzing data at the edge of the network.
• More secure, real-time data analysis and intelligent processing
• Reducing energy consumption of the entire system
• Reducing the time for integrating and migrating data.

Hikey970 is the super edge AI computing platform powered by Kirin970 SOC with 4 x Cortex A73,4 x Cortex A53. Hikey970 has 6GB LPDDR4 RAM, 64GB UFS storage, Gigabit Ethernet, GPS, PCIE Gen2 and CAN on board. As the world’s first dedicated NPU AI platform, Hikey970 integrates Huawei HiAI computing architecture and popular neural network frameworks which supports CPU, GPU AI and Neural Processing Units dedicated to AI acceleration. It also comes with Huawei’s HiAI SDK. Hikey 970 can be used in most applications in deep learning, Robotics, automobiles and smart cities.

Computer Vision

With Plug and AI in mind, Horned Sungem is dedicated to being the simplest AI device allowing developers, students, AI hobbyist and enthusiasts to create their own AI applications with ease. Without any dependency on deep learning frameworks or complex libraries, your device will be ready to see and understand the world after you plug HS into the USB port and run a short installation script.

Speech Recognition & NLP

Natural Language Processing (NLP) refers to AI method of building systems that can understand language.

NLP is required when building intelligent systems to ask robots to perform per given instructions.

ReSpeaker is an open modular voice interface to energize the world around you just using your voice. Interact with home appliances, plants, the office, internet-connected devices and other things in day-to-day life, with the power of speech. The ReSpeaker project provides hardware components and software libraries to build fully voice-enabled devices.

The below table we focused on the hardware specification and features to help you choose the most suitable development boards, modules and kits for you specific area projects.

	Key Parameters				Products
	SoM/Soc/CPU	NPU/TPU/VPN/KPUperformance	Key HardwareFeatures	Supported OS/Framework/SDK/Languages
AI Modules / Dev. Boards / Kits	RISC-V Dual Core 64-bit @400MHz, independent FPU, KPU,APU	independent KPU 0.5TOPS 0.5TOPS@800MHz 0.25TOPS@400MHz,300mW	3-channel DC-DC power8MB/16MB/128MB FlashWiFi chip ESP8285	OS: -Framework: Tiny YOLO / MobileNet v1 / TensorFlow LiteLanguages: Micropython / C++SDK: Maixpy / Standalone SDK / FreeRTOS SDK	Sipeed MAix-I Module
					Sipeed MAix BiT
					Sipeed M1w Dock Suit
					Sipeed MAix GO Suit
	Edge TPU Module / NXP i.MX 8M SOC	–	Integrated GC7000 Lite Graphics1 GB LPDDR48 GB eMMC802.11b/g/n/acGigabit Ethernet port3.5mm audio jack HDMI 2.0a (full size) 39-pin FFC connector	OS: Debian LinuxFramework: TensorFlow LiteLanguages: Python(C++ Coming soon)	Google Coral Dev Board with a removable SOM
	Rockchip RK3399	Lightspeeur 2801S5.6 TOPS5.6 [email protected] TOPS@300mW	ARM Mali-T860 MP4 quad core GPUVPU Support 4K VP9 and 4K 10bits H265/H264 video decoding, up to 60 fpsGigabit Ethernet	Supported OS: Ubuntu18.04 / Debian9 / Linux+QT / xserver / wayland / Android 8.1Supported Framework: Pytorch / Caffe / TensorFlow LiteSupported Languages: Supported SDK: GTI SDK / Linux SDK / Android SDK / Firefly API	NCC S1+ ROC-RK3399-PC AI Package
	GAP8	8 GOPS200 MOPS@1mW	RISC-V, Hardware Convolution Engine, ARDUINO form factor	Supported Framework: TensorFlowSupported Languages: C/ C++ / OpenMPSupported SDK: CMSIS API	GAPUINO GAP8 Developer Kit
	HiSilicon Kirin 970	FP16: 1.92TFLOPS	6GB LPDDR4X 1866MHz1080p@60Hz HDMIBluetooth/WIFI/GPS	OS: Android / LinuxFramework: Caffe / TensorFlow / OpenCVLanguages: PythonSDK: HiAI SDK / HiAI API	HiKey 970 Development Board
	XC7Z020-1CLG400C	–	TBD	TBD	PYNQ™ Z2 board – based on Xilinx Zynq C7Z020 SoC
	RK3308	–	Support DDR3/NandFlash/eMMC/MicroSD/802.11 b/g/n/Bluetooth 4.2, built-in audio CODEC	OS: Amazon AVS / DuerOS / AliOS Things / ROSFramework: -Languages: C / PythonSDK: Firefly SDK / Buildroot / DuerOS / Aispeech / iflytek / Amazon Alexa	ROC-RK3308-CC Quad-Core 64-Bit AIOT Main Board

AI Enablement for the Grove system is ongoing!

We have now updated the Grove.py Python library to support the Coral Dev Board and NVIDIA Jetson Nano. Connect over 200 Grove modules simply and easily with the new libraries.

Here is a blinking button demo with the Coral Dev board. Code can be found here.

import time from grove.gpio 
import GPIO 
led = GPIO(12, GPIO.OUT) 
button = GPIO(22, GPIO.IN) 
while True:     
     if button.read():         
           led.write(1)     
     else:         
           led.write(0)     
     time.sleep(0.1)

In the near future, Seeed will also expand the Grove system to include Scenario kits with the community for real projects using Google’s Edge TPU! Keep in touch with us! Let us know what you want to see in the forum, and we will do our best to listen and take action!

About Author