The Raspberry Pi AI HAT+ is an add-on board with a built-in Hailo AI accelerator forRaspberry Pi 5. It provides an accessible, cost-effective, and power-efficient way tointegrate high-performance AI. Explore applications including process control, security,home automation, and robotics.
The AI HAT+ is available in 13 and 26 tera-operations per second (TOPS) variants,built around the Hailo-8L and Hailo-8 neural network inference accelerators. The 13TOPS variant capably runs neural networks for applications including object detection,semantic and instance segmentation, pose estimation, and more. The 26 TOPS variantcan handle larger networks, can run them more quickly, and can more effectively runmultiple networks simultaneously
The AI HAT+ communicates using Raspberry Pi 5’s PCIe Gen 3 interface. When the hostRaspberry Pi 5 is running an up-to-date Raspberry Pi OS image, it automatically detectsthe on-board Hailo accelerator and makes the NPU available for AI computing tasks. Thebuilt-in rpicam-apps camera applications in Raspberry Pi OS natively support the AImodule, automatically using the NPU to run compatible post-processing tasks.
Feature
Note
For optimal performance with reComputer, we recommend using 128GB / 256GB / 512GB / 1TB SSDs from Seeed. Since some SSDs on the market may only be compatible with specific JetPack versions which may cause the device to fail to function properly, and this issue is also present with official NVIDIA dev kits.
If you are looking for the version without power adapter, please check out reComputer J2021 without power adapter.
Specification
Jetson Xavier NX System on Module
AI Performance
reComputer J2022, Xavier NX 16GB
reComputer J2021, Xavier NX 8GB
GPU
384-core NVIDIA Volta™ architecture GPU with 48 Tensor Cores
CPU
6-core NVIDIA Carmel Arm®v8.2 64-bit CPU 6MB L2 + 4MB L3
Memory
16 GB 128-bit LPDDR4x 59.7GB/s
8GB 128-bit LPDDR4x 59.7GB/s
Storage
16 GB eMMC 5.1
Video Encoder
2x 4K60 | 4x 4K30 | 10x 1080p60 | 22x 1080p30 (H.265)
Video Decoder
2x 8K30 | 6x 4K60 | 12x 4K30 | 22x 1080p60 | 44x 1080p30 (H.265)
Carrier Board
Networking
Ethernet
1x RJ45 Gigabit Ethernet Connector (10/100/1000)
M.2 KEY E
1x M.2 Key E
I/O
USB
4x USB 3.1 Type A Connector；
1x USB Type-C (Device mode)
CSI Camera
2x CSI Camera (15 pos, 1mm pitch, MIPI CSI-2 )
Display
1x HDMI Type A; 1x DP
Fan
1x FAN connector(5V PWM)
CAN
1x CAN
Multifunctional Port
1x 40-Pin Expansion header
RTC
RTC 2-pin
RTC socket
Power
9-19V DC
Mechanical
Dimensions (W x D x H)
130mm x 120mm x 58.5mm (with case)
Installation
Desk, wall-mounting
Operating Temperature
-10℃~60℃
Warranty
1 Year
Hardware Overview
reComputer J202 carrier board, included in the full system - reComputer J2021
Desktop, Wall Mount, Expandable, or Fit in Anywhere
Application
Autonomous Mobile Robot (AMR)
AI Video Analytics
Machine Vision
Documents
Datasheet
Schematic
3D File
Seeed Nvidia Jetson Product Catalog
Nvidia Jetson Comparison
Nvidia Jetson Product Comparison
Seeed Nvidia Jetson Success Cases
Seeed Jetson One Pager
ECCN/HTS
HSCODE
8471419000
USHSCODE
8543708800
UPC
EUHSCODE
8471800000
COO
CHINA
Part List
Acrylic Cover
1x
Aluminium Frame
1x
Jetson Xavier NX module
1x
Aluminum heatsink with fan
1x
Carrier board
1x
12V/5A(Barrel Jack 5.5/2.5mm) Power Adapter (Power cable not included)
1x
The Coral USB Accelerator adds a Coral Edge TPU to your Linux, Mac, or Windows computer so you can accelerate your machine learning models. This page is your guide to get started.
All you need to do is download the Edge TPU runtime and PyCoral library. Then we'll show you how to run a TensorFlow Lite model on the Edge TPU.
To learn more about the hardware, see the USB Accelerator datasheet.
Requirements
A computer with one of the following operating systems:
Linux Debian 10, or a derivative thereof (such as Ubuntu 18.04), and a system architecture of either x86-64, Armv7 (32-bit), or Armv8 (64-bit) (includes support for Raspberry Pi 3 Model B+, Raspberry Pi 4, and Raspberry Pi Zero 2)
macOS 10.15 (Catalina) or 11 (Big Sur), with either MacPorts or Homebrew installed
Windows 10
One available USB port (for the best performance, use a USB 3.0 port)
Python 3.6 - 3.9
The Raspberry Pi AI Kit bundles the Raspberry Pi M.2 HAT+ with a Hailo AI accelerationmodule for use with Raspberry Pi 5. It provides an accessible, cost-effective, and powerefficient way to integrate high-performance AI. Explore applications including processcontrol, security, home automation, and robotics. Contains:
• a Hailo AI module containing a Neural Processing Unit (NPU)• a Raspberry Pi M.2 HAT+, to connect the AI module to your Raspberry Pi 5• a thermal pad pre-fitted between the module and the M.2 HAT+• a mounting hardware kit• a 16mm stacking GPIO header
The AI module is a 13 tera-operations per second (TOPS) neural network inferenceaccelerator built around the Hailo-8L chip. The module uses the M.2 2242 form factor,and comes pre-installed in the M.2 HAT+, to which it connects through an M key edgeconnector. The M.2 HAT+ communicates between the AI module’s M.2 interface and theRaspberry Pi 5’s PCIe 2.0 interface.
When the host Raspberry Pi 5 is running an up-to-date Raspberry Pi OS image,it automatically detects the Hailo module and makes the NPU available for AI computingtasks. The built-in rpicam-apps camera applications in Raspberry Pi OS natively supportthe AI module, automatically using the NPU to run compatible post-processing tasks.
Feature
Note
The reComputer J3011 is equipped with the same NVIDIA Jetson Orin Nano 8GB production module. You can perform a self - upgrade to Jetpack 6.2. Once upgraded, you'll instantly experience a significant boost in computing power, with the performance leaping from 40 Tops to 67 Tops, offering capabilities comparable to those of the NVIDIA Jetson Orin Nano Super Developer Kit.
For optimal performance with reComputer, we recommend using 128GB / 256GB / 512GB / 1TB SSDs from Seeed. Since some SSDs on the market may only be compatible with specific JetPack versions which may cause the device to fail to function properly, and this issue is also present with official NVIDIA dev kits.
If you are looking for the version without power adapter, please check out reComputer J3011 without power adapter
reComputer J3011 (original with JP5.1.3)
reComputer J3011 (super with JP6.2)
GPU
NVIDIA Ampere architecture
1,024 CUDA Cores
32 Tensor Cores
635 MHz
NVIDIA Ampere
architecture
1,024 CUDA Cores
32 Tensor Cores
1,020MHz
AI PERF
40 INT8 TOPS (Sparse)
20 INT8 TOPS (Dense)
67 TOPS (Sparse)
33 TOPS (Dense)
CPU
6-core Arm Cortex-
A78AE v8.264-bit CPU
1.5 GHz
6-core Arm Cortex-
A78AE v8.264-bit CPU
1.7 GHz
Memory
8GB 128-bit LPDDR5
68GB/s
8GB 128-bit LPDDR5
102GB/s
MODULE POWER
7W|15W
10W|15W|25W
Specification
Jetson Orin Nano System on Module
AI Performance
reComputer J3011 - Orin Nano 8GB, up to 67 TOPS
reComputer J3010- Orin Nano 4GB – up to 34 TOPS
GPU
1024-core NVIDIA Ampere architecture GPU with 32 Tensor Cores (Orin Nano 8GB)
512-core NVIDIA Ampere architecture GPU with 16 Tensor Cores (Orin Nano 4GB)
CPU
6-core Arm® Cortex®-A78AE v8.2 64-bit CPU 1.5MB L2 + 4MB L3
Memory
8GB 128-bit LPDDR5 68 GB/s (Orin Nano 8GB)
4GB 64-bit LPDDR5 34 GB/s (Orin Nano 4GB)
Video Encoder
1080p30 supported by 1-2 CPU cores
Video Decoder
1x 4K60 (H.265) | 2x 4K30 (H.265) | 5x 1080p60 (H.265) | 11x 1080p30 (H.265)
Carrier Board
Storage
1x M.2 Key M PCIe (M.2 NVMe 2280 SSD 128G included)
Networking
Ethernet
1x RJ-45 Gigabit Ethernet (10/100/1000M)
M.2 KEY E
1x M.2 Key E(pre-installed 1x Wi-Fi/Bluetooth combo module)
I/O
USB
4x USB 3.2 Type-A (10Gbps), 1x USB2.0 Type-C (Device Mode)
Camera
2x CSI (2-lane 15pin)
Display
1x HDMI 2.1
Fan
1x 4 pin Fan Connector (5V PWM)
CAN
1x CAN
Multifunctional Port
1x 40-Pin Expansion header
1x 12-Pin Control and UART header
RTC
1x RTC 2-pin, supports CR1220 but not included
Power
9-19V DC
Mechanical
Dimensions (W x D x H)
130mm x 120mm x 58.5mm (with case)
Installation
Desk, wall-mounting
Operating Temperature
-10℃~60℃
Warranty
1 Year
Hardware Overview
reComputer J401 carrier board, included in the full system - reComputer J3011
● Desktop, Wall Mount, Expandable, or Fit in Anywhere
Application
Application Fields:
AI Video Analytics
Machine Vision
Autonomous Mobile Robot (AMR)
Generative AI
● Capable of Bringing Generatie AI to the Edge
Now you can build AI agents capable of processing large amounts of live or archived videos and images with Vision-Language Models (VLM) such as LLaVA, this new AI agents help nearly every industry summarize, search, and extract actionable insights from video using natural language.
● Build Multi-streams AI Video Analytics
For building video analytics, Jetson Orin Nano incorporates the NVIDIA Multi-Standard Video Decoder. This video decoder accelerates video decode, supporting low resolution mobile content, Standard Definition (SD), High Definition (HD) and UltraHD (8K, 4K, etc.) reComputer J3011 can take 13x1080p30 streams. See our tested YOLOv8 performance using NVIDIA Deepstream of single model and multi model on multi streams.
● Fastest Way to Deploy Generative AI and Computer Vision Models
We have jetson-example prepared for you! It offers one-line deploy projects edge AI applications of generative AI including Ollama, Llama3; computer vision including YOLOv8; and others. We have configured all environment for you to provides single command deployment of projects.
