Product Overview
An add-on board for Raspberry Pi 5 with a built-in Hailo AI accelerator, connected through the PCIe interface. It enables on-device AI workloads such as object detection, image segmentation, pose estimation and facial recognition for robotics, automation and smart-camera projects.
It integrates with Raspberry Pi’s camera software stack, supports Raspberry Pi OS (Bookworm) and works with all Raspberry Pi camera modules. A stacking header and spacers are included, so it can be fitted alongside the Active Cooler.
Key Features
| Feature | Benefit |
|---|---|
| 13 TOPS Hailo-8L accelerator | Real-time object detection and pose estimation |
| PCIe connection to Raspberry Pi 5 | High-bandwidth link to the accelerator |
| Low-latency, energy-efficient inference | Fast results without a large power budget |
| libcamera / rpicam-apps integration | Camera demos work out of the box |
| Stacking header and spacers | Fits alongside the Active Cooler |
Applications
| Application Area | Example Projects |
|---|---|
| Smart cameras | Person, vehicle and object detection |
| Robotics | Vision-guided navigation and picking |
| Retail | People counting and queue monitoring |
| Agriculture | Crop, pest and livestock monitoring |
| Industry | Visual inspection and safety-zone monitoring |
Benefits
AI on the device
Video is processed locally on the Raspberry Pi 5, so there are no cloud fees and results arrive with low latency.
Works with the camera stack
Integration with rpicam-apps means the demos run with a Raspberry Pi camera straight away.
Bring your own models
Models from TensorFlow, PyTorch, Keras and ONNX can be compiled for the Hailo accelerator.
Specifications
| Specification | Details |
|---|---|
| Dimensions | 65.0 × 56.5 × 5.6 mm (max) |
| Operating temperature | 0 °C to 50 °C (ambient) |
| Model sources | TensorFlow, TensorFlow Lite, Keras, PyTorch and ONNX (compiled for Hailo) |
| Ribbon cable | 30 mm, pre-fitted |
Compatibility
- Raspberry Pi 5
Why Choose This Product?
It’s the affordable entry point to hardware-accelerated AI on Raspberry Pi 5, with plenty of performance for common vision tasks.
Beginner Use Cases
- Run the object-detection demo with a camera
- Try pose estimation
- Count people passing a doorway
- Learn how AI inference works
Professional Use Cases
- Smart security and access cameras
- Retail analytics
- Robot vision systems
- Industrial inspection stations
Please Note
- Raspberry Pi 5 is not included
13 TOPS or 26 TOPS?
- AI HAT+ 13 TOPS (Hailo-8L) — the entry point for edge AI on Raspberry Pi 5: object detection, pose estimation and image classification.
- AI HAT+ 26 TOPS (Hailo-8) — twice the compute (26 vs 13 TOPS), for larger models and higher frame rates.
- Both fit the Raspberry Pi 5 only and work with the Raspberry Pi camera software stack.
