Overview
Last updated:
Seeed Studio's reComputer R Series brings Raspberry Pi computing into compact edge and industrial systems. The reComputer AI R family includes a commercial Raspberry Pi 5 platform and an industrial Compute Module 5 (CM5) platform; official AI configurations integrate a Hailo-8 accelerator delivering up to 26 TOPS for real-time, low-latency vision inference. Industrial configurations extend the platform with deployment-oriented storage, networking, power, and expansion options.
Use this section to identify the exact product and SKU, prepare its operating system, verify the Hailo accelerator when present, and continue to a documented edge-AI application. Hardware and accelerator availability vary by model, so follow the product-specific Seeed Wiki rather than assuming every R Series device has the same interfaces or Hailo module.
Product paths
| Product path | Compute platform | Typical starting point |
|---|---|---|
| reComputer AI R2000 | Raspberry Pi 5 | Desktop/commercial vision AI with Hailo-8 |
| reComputer AI Industrial R2135 | Raspberry Pi CM5 | Compact industrial edge AI with pre-installed Hailo-8 |
| reComputer Industrial R20xx/R21xx | Raspberry Pi CM5 | Industrial control and connectivity; accelerator availability depends on the SKU |
Always check the exact product suffix and bill of materials before following an accelerator tutorial. A product with an M.2 slot is not necessarily shipped with a Hailo module.
Recommended learning order
- Choose your device. Match the chassis, compute module, storage, power, interfaces, and accelerator to the exact Seeed product Wiki.
- Flash and boot the operating system. CM5 products may boot from eMMC or NVMe and use a different flashing procedure from a standard Raspberry Pi 5.
- Verify the base system. Confirm architecture, network access, storage, cooling, and throttling state before installing AI software.
- Verify Hailo separately. Confirm the PCIe device, kernel driver,
/dev/hailo0, HailoRT CLI, and firmware communication. - Run an application. Start with a documented model or NVR workflow before integrating custom cameras and models.