Specifications

SoC

SOPHGO SG2002 tri-core RISC-V and Arm processor

Main Core

1 GHz 64-bit RISC-V C906 or Arm Cortex-A53 core, selectable

Minor Core

700 MHz 64-bit RISC-V C906 core

Low-Power Core

8051 MCU running 25 to 300 MHz with 8KB SRAM

NPU

1 TOPS INT8 accelerator with BF16 support

Architecture

Three boards: Core board, Sensor board and Baseboard

Core Board C1_2002w

eMMC storage with onboard antenna and external IPEX antenna connector

Core Board C1_2002

Additional SDIO and UART breakout to the baseboard; no Wi-Fi

Sensor Option 1

5MP OmniVision OV5647 rolling shutter, 2592x1944 stills, 1080p30 video

Sensor Option 2

1MP SmartSens SC130GS global shutter CMOS sensor

Sensor Option 3

5MP Sony IMX335 rolling shutter, 2592x1944 at 20fps and 1080p30

Baseboard B1_Default

USB Type-C port for power

Baseboard Options

Vertical mounted Type-C, PoE and gyro variants (B2/B3/B4)

Storage

MicroSD card slot supporting up to 2TB

Target Applications

Robotics, healthcare, smart home, buildings and industrial automation

Overview

The Seeed Studio reCamera splits a smart camera into three replaceable boards - a Core board carrying the processor, storage and optional wireless, a Sensor board carrying the image sensor, and a Baseboard providing power and connectivity. That separation lets an integrator keep the same compute and software stack while swapping between rolling-shutter 5MP colour sensors, a global-shutter sensor for motion capture, or a PoE-fed baseboard for a fixed installation.

Compute comes from the SOPHGO SG2002, a tri-core part that pairs a 1 GHz main core (selectable RISC-V C906 or Arm Cortex-A53) with a 700 MHz RISC-V minor core and an 8051 MCU that handles low-power duties down to 25 MHz. The integrated 1 TOPS INT8 NPU supports BF16, which is enough for on-camera detection and classification without shipping frames to a host.

Two core boards are documented: the C1_2002w with onboard eMMC plus Wi-Fi and Bluetooth for wireless deployments, and the C1_2002 which trades wireless for extra SDIO and UART breakout to the baseboard. Sensor and baseboard variants including a PoE version and a gyro-equipped version extend the same platform across robotics, building automation and industrial inspection use cases.

Key Benefits

Three-board modular architecture for sensor and power-path flexibility, 1 TOPS INT8 NPU with BF16 on-camera inference, global or rolling shutter sensor options, and PoE or CAN-capable baseboards for fixed industrial installs.

Applications

Robotics perception, industrial inspection endpoints, smart building and access control cameras, healthcare monitoring devices, and distributed edge vision networks.

Request a Quote — SEEED RECAMERA — MODULAR RISC-V AI CAMERA WITH SOPHGO SG2002 AND SWAPPABLE SENSORS

QS Compute — global B2B supply of AI computing hardware, edge AI systems and accelerators. Volume pricing, 15-day sample lead time.

Get Your Quote →

Related Products

CVITEK CV1842H Edge AI Camera Module NeoEyes NE503 Hailo-15H Edge AI Camera Arducam IMX500 AI Camera LooperRobotics Insight 9 Spatial AI Camera