- Description
Details
Equipped With 26TOPS Hailo-8 M.2 AI Accelerator Module
This AI kit is launched by Waveshare to provide a more cost-effective and high-performance AI solution for the Raspberry Pi 5, optional for PCIe To M.2 adapter, suitable for applications such as process control, safety, home automation and robotics, etc.
Hailo-8
the Hailo-8 AI M.2 module only
Hailo-8 Acce A
Hailo-8 M.2 module + PCIe TO M.2 adapter and accessories (can be directly accessed to Raspberry Pi 5)
- Hailo-8 AI M.2 module
- Powered by 26 Tera-Operations Per Second (TOPS) Hailo-8 AI Processor
- 2.5W typical power consumption
- Scalable, enabling simultaneous processing of multi-streams & multi-models
- Enabling real-time, low latency and high-efficiency AI inferencing on the edge devices
- Supports TensorFlow, TensorFlow Lite, ONNX, Keras, Pytorch frameworks
- Supports Linux and Windows
- Supports the temperature range of -40°C to 85°C
- PCIe To M.2 adapter
- Onboard power monitoring chip and EEPROM, supports real-time monitoring of device power status for more stable operation
- Raspberry Pi HAT+ compliant
- Reserved airflow vent, supports installing cooling fan for better heat dissipation of the AI module to improve performance
- Immersion gold process design, anti-oxidation and more durable
| AI performance | 26 TOPS |
|---|---|
| Form Factor | M.2 Key M |
| Power supply | 3.3V ± 5% |
| Power consumption | 2.5W (Typ.) 8.65W (Max.) |
| Interface | PCIe Gen3, 4-lane |
| Certificate | CE, FCC Class A |
| Storage temperature | -40 ~ 85°C |
| Operating temperature | -40 ~ 85°C |
| Operating humidity | 5% ~ 90%RH (no frosting) |
| Dimensions | 22×80mm with breakable extensions to 22×42mm and 22×60mm |
The Hailo-8 M.2 module is an AI accelerator module for AI applications, based on the 26 tera-operations per second (TOPS) Hailo-8 AI processor with high power efficiency. The M.2 AI accelerator features a full PCIe Gen-3.0 4-lane interface, delivering unprecedented AI performance for edge devices.
The M.2 module can be plugged into an existing edge device with M.2 socket to provide low-power deep neural network inferencing. Leveraging Hailo's comprehensive Dataflow Compiler and its support for standard AI frameworks, customers can easily port their Neural Network models to the Hailo-8 and introduce high-performance AI products to the market quickly.
| NN Model | mAP |
Hailo-8L FPS
(batch8) |
|---|---|---|
| yolov4_tiny | 18.98 | 610 |
| yolov6n | 34.3 | 345 |
| yolov7 | 49.8 | 45 |
| yolox_s_wide | 42.4 | 75 |
| yolov3 | 38 | 26 |
| yolov8n | 37.23 | 270 |
| yolov8s | 44.75 | 128 |
| yolov8m | 50.08 | 55 |
| Type | NN Model | Input Resolution | FPS | Power(W) | FPS/W |
|---|---|---|---|---|---|
| Classification | ResNet-50 v1 | 224x224 | 1332 | 3.45 | 386 |
| MobileNet_v2_1.0 | 224x224 | 2444 | 2.152 | 1135 | |
| EfficientNet_M | 240x240 | 889 | 3.5 | 254 | |
| Object Detection | SSD_MobileNet_v1 | 300x300 | 1055 | 2.2 | 479 |
| YOLOv5m | 640x640 | 218 | 4.6 | 47.3 | |
| Segmentation | stdc1 | 1024x1920 | 54 | 2.9 | 18.6 |
| Multi stream object detection (8 streams) | YOLOv3 | 608x608 | 69 | 4.9 | 14 |
Based on 16PIN PCIe Interface of Raspberry Pi 5
Standard Raspberry Pi 40PIN Header, Comes with 2*20 Pin header for Stacking with Other HATs. Compact Size, More Space-Saving, supports installing cooling fan
can be used together with the Pi5 Active Cooler B to achieve better heat dissipation effect For the Pi5 And AI Accelerator Module, keeping it cool even under heavy processing and maximizing the Module Performance
Real-time monitoring of device power status for More Stable operation
* for reference only, the cooling fan is NOT included.
Weight: 0.006 kg
Quick Overview
Hailo-8
Hailo-8 AI M.2 Module ×1
Hailo-8 Acce A
- Hailo-8 AI M.2 Module x1
- PCIe TO M.2 HAT+ x1
- Standoff pack x1
- 16P-Cable-40mm x1
- 2*20 Pin header x1
