Top 5 Development Boards for Deep Learning and AI projects

Spread the love

Deep learning and AI are the future of technology. We are seeing how it is changing the entire consumer product industry in a way we had not seen before or imagined before. Netflix knows which movie or shows you will watch next, amazon knows what products you are going to buy in the next 3 months and more. The hardware and electronics industry is trying to keep pace with the ever-evolving need for processing power to fulfill the requirements of consumers. In this post, we are going to list down the top 5 development boards that are available in the market for building deep learning and AI projects. 

In our earlier blog post we have covered NVIDIA Jetson Nano and Google Coral, we have also touched upon Microsoft Azure Sphere. These remain at the top of the chart. We have some newcomers from well known DIY and community-based boards into the list.

NVIDIA Jetson Nano


Driven by its own very strong philosophy as a hardware company. NVIDIA hardware design always outperforms anyone in the market for that matter. It also has a special place in the development community, since it was already in a favorable spot when machine learning started to make use of application-specific hardware. Nvidia was within arm’s reach with its GPUs when crypto heads and machine learners needed something to multiply and add, fast.

NVIDIA Jetson Nano is a small, powerful computer that delivers 472 GFLOPs for running modern AI algorithms fast. It lets you run multiple neural networks in parallel for applications like image classification, object detection, segmentation, and speech processing. All in an easy-to-use platform that runs in as little as 5 watts. It is an ideal board for applications like entry-level Network Video Recorders (NVRs), home robots, and intelligent gateways with full analytics capabilities.

Google Coral

Google Coral, a powerful TPU (tensor processing unit) were originally designed for Google’s own internal data center. The launch of Google Coral Edge TPU to embedded developers is a vision for a bigger ambition that Google has now and the debut of their ambition is recently made possible with the recent beta program for Google Coral.

Coral covers a wide application domain with its peripherals:

  • MIPI camera/display interfaces,
  • 2 onboard MEMS microphones,
  • a 3W speaker output,
  • an onboard MIMO WiFi connection
  • a Quad-core ARM A53 application processor (NXP iMX8M, which hits the mark for its intended audience, but remains slow when compared to the competition),
  • and a Microchip ATECC608A for cryptographic co-processor.

Google Coral has everything needed for an IoT device to execute machine learning or deep learning programs for audiovisual applications. Google simply states that both machine learning inference and IoT security is a serious issue, and should both be addressed with their dedicated hardware. I also believe that it is a complete set to span as many applications as possible in the domain.  

Intel Neural Compute Stick 2


Intel Neural Compute Stick is a plug and play AI module that can be added to a Raspberry pi board, for instance, to provide an AI interface for audiovisual processing needs. It includes 16 powerful processing cores (called SHAVE cores) and a dedicated deep neural network hardware accelerator for high-performance vision and AI inference applications—all at low power.

Intel Neural Compute Stick 2 is a power device with specs as –


  • Processor: Intel Movidius Myriad X Vision Processing Unit (VPU)
  • Supported frameworks: TensorFlow*, Caffe*, Apache MXNet*, Open Neural Network Exchange (ONNX*), PyTorch*, and PaddlePaddle* via an ONNX conversion
  • Connectivity: USB 3.0 Type-A
  • Dimensions: 2.85 in. x 1.06 in. x 0.55 in. (72.5 mm x 27 mm x 14 mm)
  • Operating temperature: 0° C to 40° C


  • OpenVINO™ toolkit (both Intel® Distribution of OpenVINO™ toolkit and open-sourced distribution of OpenVINO™ toolkit)
  • Supported operating systems: Ubuntu 16.04.3 LTS (64 bit), CentOS* 7.4 (64 bit), Windows 10 (64 bit), Raspbian (target only), Other (via the open-source distribution of OpenVINO™ toolkit)

BeagleBone® AI Board


BeagleBone AI board is a small Single Board Computer that is based on Texas Instruments AM5729. BeagleBone AI board makes it easy to explore how artificial intelligence (AI) can be used in everyday life via the TI C66x digital-signal-processor (DSP) cores and embedded-vision-engine (EVE) cores supported through an optimized TIDL machine learning OpenCL API with pre-installed tools. Focused on everyday automation in industrial, commercial, and home applications.

BeagleBone AI board comes with below specs – 

  • Dual Arm® Cortex®-A15 microprocessor subsystem
  • 2 C66x floating-point VLIW DSPs
  • 2.5MB of on-chip L3 RAM
  • 2x dual Arm® Cortex®-M4 co-processors
  • 4x Embedded Vision Engines (EVEs)
  • 2x dual-core Programmable Real-Time Unit and Industrial Communication SubSystem (PRU-ICSS)
  • 2D-graphics accelerator (BB2D) subsystem
  • Dual-core PowerVR® SGX544™ 3D GPU
  • IVA-HD subsystem (4K @ 15fps encode and decode support for H.264, 1080p60 for others)



Taking a bet here by putting Arduino Portenta H7 in the list of the top development board for AI but I am very sure that Arduino Portenta H7 will justify this after the launch. Portenta H7 has 2 parallel cores that can simultaneously run high-level code along with real-time tasks. 

Portenta H7’s has a dual-core STM32H747 that includes a Cortex® M7 running at 480 MHz and a Cortex® M4 running at 240 MHz. The two cores communicate via a Remote Procedure Call mechanism that allows calling functions on the other processor seamlessly.

By default the Arduino Portenta H7 comes with:

  • STM32H747 dual-core processor with a graphics engine
  • 16MB NOR Flash
  • 10/100 Ethernet Phy
  • USB HS
  • NXP SE050C2 Crypto
  • WiFi/BT Module
  • UFL Connector (Antenna)
  • DisplayPort over USB-C

and can be extended to host a 64 MByte of SDRAM, and 128 MByte of QSPI Flash. It also comes with an external UFL connector for adding a high-gain antenna to the board. Decide between crypto-chips from Microchip and NXP.

Both processors share all the on-chip peripherals and can run:

  • Arduino sketches on top of the Arm® Mbed™ OS
  • Native Mbed™ applications
  • MicroPython / JavaScript via an interpreter
  • TensorFlow™ Lite

The onboard wireless module on Portenta H7 allows simultaneous management of WiFi and Bluetooth connectivity. The WiFi interface can be operated as an Access Point, as a Station or as a dual-mode simultaneous AP/STA, and can handle up to 65 Mbps transfer rate. Bluetooth® interface supports Bluetooth Classic and BLE.

The Portenta H7 follows the Arduino MKR form factor but enhanced with the Portenta family 80 pin high-density connector. Learn more about the board’s pinout by reading the board’s pinout documentation.


Spread the love

You may also like...