The Raspberry Pi AI Camera is a compact camera module based on the Sony IMX500 Intelligent Vision Sensor. The IMX500 combines a 12 MP CMOS image sensor with on-board inferencing acceleration for various common neural network models, allowing users to develop sophisticated vision-based AI applications without requiring a separate accelerator.
The AI Camera enhances captured still images or video with tensor metadata, while keeping the Raspberry Pi's processor free for other tasks. Support for tensor metadata in the libcamera and Picamera2 libraries, as well as the rpicam-apps application suite, ensures ease of use for beginners while providing unparalleled power and flexibility for advanced users.
The Raspberry Pi AI Camera is compatible with all Raspberry Pi models.
Features
12 MP Sony IMX500 Intelligent Vision Sensor
Sensor modes: 4056x3040 (@ 10fps), 2028x1520 (@ 30fps)
1.55 x 1.55 µm cell size
78-degree field of view with manually adjustable focus
Integrated RP2040 for neural network and firmware management
Specifications
Sensor
Sony IMX500
Resolution
12.3 MP (4056 x 3040 pixels)
Sensor size
7.857 mm (type 1/2.3)
Pixel size
1.55 x 1.55 μm
IR cut filter
Integrated
Autofocus
Manual adjustable focus
Focus range
20 cm – ∞
Focal length
4.74 mm
Horizontal FOV
66 ±3°
Vertical FOV
52.3 ±3°
Focal ratio (F-stop)
F1.79
Output
Image (Bayer RAW10), ISP output (YUV/RGB), ROI, metadata
Input tensor maximum size
640 x 640 (H x V)
Framerate
• 2x2 binned: 2028x1520 10-bit 30fps• Full resolution: 4056x3040 10-bit 10fps
Ribbon cable length
20 cm
Cable connector
15 x 1 mm FPC or 22 x 0.5 mm FPC
Dimensions
25 x 24 x 11.9 mm
Downloads
Datasheet
Documentation
The Raspberry Pi AI HAT+ is an expansion board designed for the Raspberry Pi 5, featuring an integrated Hailo AI accelerator. This add-on offers a cost-effective, efficient, and accessible approach to incorporating high-performance AI capabilities, with applications spanning process control, security, home automation, and robotics.
Available in models offering 13 or 26 tera-operations per second (TOPS), the AI HAT+ is based on the Hailo-8L and Hailo-8 neural network accelerators. The 13 TOPS model efficiently supports neural networks for tasks like object detection, semantic and instance segmentation, pose estimation, and more. This 26 TOPS variant accommodates larger networks, enables faster processing, and is optimized for running multiple networks simultaneously.
The AI HAT+ connects via the Raspberry Pi 5’s PCIe Gen3 interface. When the Raspberry Pi 5 is running a current version of the Raspberry Pi OS, it automatically detects the onboard Hailo accelerator, making the neural processing unit (NPU) available for AI tasks. Additionally, the rpicam-apps camera applications included in Raspberry Pi OS seamlessly support the AI module, automatically using the NPU for compatible post-processing functions.
Included
Raspberry Pi AI HAT+ (26 TOPS)
Mounting hardware kit (spacers, screws)
16 mm GPIO stacking header
Downloads
Datasheet
This collection features the best of Elektor Magazine's articles on embedded systems and artificial intelligence. From hands-on programming guides to innovative AI experiments, these pieces offer valuable insights and practical knowledge for engineers, developers, and enthusiasts exploring the evolving intersection of hardware design, software innovation, and intelligent technology.
Contents
Programming PICs from the Ground UpAssembler routine to output a sine wave
Object-Oriented ProgrammingA Short Primer Using C++
Programming an FPGA
Tracking Down Microcontroller Buffer Overflows with 0xDEADBEEF
Too Quick to Code and Too Slow to Test?
Understanding the Neurons in Neural NetworksEmbedded Neurons
MAUI Programming for PC, Tablet, and SmartphoneThe New Framework in Theory and Practice
USB Killer DetectorBetter Safe Than Sorry
Understanding the Neurons in Neural NetworksArtificial Neurons
A Bare-Metal Programming Guide
Part 1: For STM32 and Other Controllers
Part 2: Accurate Timing, the UART, and Debugging
Part 3: CMSIS Headers, Automatic Testing, and a Web Server
Introduction to TinyMLBig Is Not Always Better
Microprocessors for Embedded SystemsPeculiar Parts, the Series
FPGAs for BeginnersThe Path From MCU to FPGA Programming
AI in Electronics DevelopmentAn Update After Only One Year
AI in the Electronics LabGoogle Bard and Flux Copilot Put to the Test
ESP32 and ChatGPTOn the Way to a Self-Programming System…
Audio DSP FX Processor Board
Part 1: Features and Design
Part 2: Creating Applications
Rust + EmbeddedA Development Power Duo
A Smart Object CounterImage Recognition Made Easy with Edge Impulse
Universal Garden LoggerA Step Towards AI Gardening
A VHDL ClockMade with ChatGPT
TensorFlow Lite on Small MicrocontrollersA (Very) Beginner’s Point of View
Mosquito DetectionUsing Open Datasets and Arduino Nicla Vision
Artificial Intelligence Timeline
Intro to AI AlgorithmsPrompt: Which Algorithms Implement Each AI Tool?
Bringing AI to the Edgewith ESP32-P4
The Growing Role of Edge AIA Trend Shaping the Future
The Raspberry Pi AI HAT+ is an expansion board designed for the Raspberry Pi 5, featuring an integrated Hailo AI accelerator. This add-on offers a cost-effective, efficient, and accessible approach to incorporating high-performance AI capabilities, with applications spanning process control, security, home automation, and robotics.
Available in models offering 13 or 26 tera-operations per second (TOPS), the AI HAT+ is based on the Hailo-8L and Hailo-8 neural network accelerators. This 13 TOPS model efficiently supports neural networks for tasks like object detection, semantic and instance segmentation, pose estimation, and more. The 26 TOPS variant accommodates larger networks, enables faster processing, and is optimized for running multiple networks simultaneously.
The AI HAT+ connects via the Raspberry Pi 5’s PCIe Gen3 interface. When the Raspberry Pi 5 is running a current version of the Raspberry Pi OS, it automatically detects the onboard Hailo accelerator, making the neural processing unit (NPU) available for AI tasks. Additionally, the rpicam-apps camera applications included in Raspberry Pi OS seamlessly support the AI module, automatically using the NPU for compatible post-processing functions.
Included
Raspberry Pi AI HAT+ (13 TOPS)
Mounting hardware kit (spacers, screws)
16 mm GPIO stacking header
Downloads
Datasheet
The reComputer J1020 v2 is a compact edge AI device powered by the NVIDIA Jetson Nano 4 GB production module, delivering 0.5 TFLOPs of AI performance. It features a robust aluminum case with a passive heatsink and comes pre-installed with JetPack 4.6.1. The device includes 16 GB of onboard eMMC storage and offers 2x SCI, 4x USB 3.0, M.2 Key M, HDMI, and DP.
Applications
Computer Vision
Machine Learning
Autonomous Mobile Robot (AMR)
Specifications
Jetson Nano 4 GB System-on-Module
AI Performance
Jetson Nano 4 GB (0.5 TOPS)
GPU
NVIDIA Maxwel architecture with 128 NVIDIA CUDA cores
CPU
Quad-core ARM Cortex-A57 MPCore processor
Memory
4 GB 64-bit LPDDR4 25.6 GB/s
Video Encoder
1x 4K30 | 2x 1080p60 | 4x 1080p30 | 4x 720p60 | 9x 720p30 (H.265 & H.264)
Video Decoder
1x 4K60 | 2x 4K30 | 4x 1080p60 | 8x 1080p30 | 9x 720p60 (H.265 & H.264)
Carrier Board
Storage
1x M.2 Key M PCIe
Networking
Ethernet
1x RJ-45 Gigabit Ethernet (10/100/1000M)
I/O
USB
4x USB 3.0 Type-A1x Micro-USB port for device mode
CSI Camera
2x CSI (2-lane 15-pin)
Display
1x HDMI Type A; 1x DP
Fan
1x 4-pin Fan Connector (5 V PWM)
CAN
1x CAN
Multifunctional Port
1x 40-Pin Expansion header
1x 12-Pin Control and UART header
Power Supply
DC 12 V/2 A
Mechanical
Dimensions
130 x 120 x 50 mm (with Case)
Installation
Desktop, wall-mounting
Operating Temperature
−10°C~60°C
Included
reComputer J1020 v2 (system installed)
12 V/2 A power adapter (with 5 interchangeable adapter plugs)
Downloads
reComputer J1020 v2 datasheet
reComputer J1020 v2 3D file
Seeed NVIDIA Jetson Product Catalog
NVIDIA Jetson Device and Carrier Boards Comparison
The starter kit for Jetson Nano is one of the best kits for beginners to get started with Jetson Nano. This kit includes 32 GB MicroSD card, 20 W adapter, 2-pin jumper, camera, and micro-USB cable.
Features
32 GB High-performance MicroSD card
5 V 4 A power supply with 2.1 mm DC barrel connector
2-pin jumper
Raspberry Pi camera module V2
Micro-B To Type-A USB cable with DATA enabled