OrangePi AIpro Overview
Hardware Specifications
The OrangePi AIpro is a high-performence AI development board featuring:
| Component | Details |
|---|---|
| Processor | 4-core 64-bit ARM (1 core reserved for AI) |
| AI Accelerator | Huawei Ascend 310B (4 TFLOPS FP16, 8 TOPS INT8) |
| Memory | 8GB/16GB LPDDR4X @ 3200Mbps |
| Storage | 32MB SPI Flash, microSD slot, eMMC (up to 256GB), M.2 NVMe/SATA |
| Connectivity | Gigabit Ethernet, Dual-band WiFi, Bluetooth 4.2 |
| USB Ports | 2x USB 3.0, 1x USB-C, 1x Micro USB (debug) |
| Display | Dual HDMI 4K@60Hz, MIPI DSI |
| Camera | Dual MIPI CSI |
| Power | USB-C PD (65W) |
| OS Support | Ubuntu 22.04, openEuler 22.03 |
Enitial Setup
-
Image Flashing
- Use balenaEtcher or Ascend-devkit-imager
- Download OS image from official sources
-
Boot Configuration
- Set boot mode via BOOT1/BOOT2 switches
- Requires power cycle after switch changes
-
Serial Console
screen /dev/ttyUSB0 115200 -
Network Setup
nmcli dev wifi list nmcli dev wifi connect SSID password PASSWORD
YOLOv5 Implementation
Installation
sudo apt update
sudo apt install -y python3-opencv git
pip install torch torchvision
Running Object Detection
import cv2
import torch
# Load model
model = torch.hub.load('ultralytics/yolov5', 'yolov5s')
# Process image
img = cv2.imread('input.jpg')
results = model(img)
# Save output
cv2.imwrite('output.jpg', results.render()[0])
Performence Notes
- Achieves real-time detection at ~15 FPS (640x480)
- Supports multiple object classes
- Model can be customized for specific use cases