Skip to content
Free Standard Shipping for orders over $30 USD (Most of USA and Canada ) - Size and weight limits may apply.
Free Standard Shipping for orders over $30 USD (Most of USA and Canada ) - Size and weight limits may apply.

reCamera 2002 64GB Open-Source AI Camera

Original price $205.09 - Original price $205.09
Original price
$205.09
$205.09 - $205.09
Current price $205.09

reCamera 2002 is a small, fully open-source AI camera for people who want to build their own vision projects. A 1 TOPS NPU runs YOLO-style models on the camera, and Node-RED lets you wire up detection results to other systems with little or no code. The whole stack, from schematics to Linux image, is published by Seeed Studio.

What you can build

  • People counting and occupancy checks
  • Detecting objects and sending results over MQTT, HTTP or RTSP
  • Learning edge AI with Node-RED flows and SenseCraft AI
  • Prototyping a custom camera product on open hardware

Key specifications

Processor SG2002 (C906 RISC-V cores at 1 GHz and 700 MHz, 8051 MCU)
AI performance 1 TOPS (INT8)
Memory 256MB
Storage 64GB eMMC, microSD slot up to 512GB
Image sensor OV5647, 2MP, 1/4 inch
Video 1920 x 1080 at 30 fps
Field of view 65 degrees
Ethernet 100 Mbps
USB USB Type-C (2.0)
Extras On-board microphone, speaker, 4 white fill lights
Power 5V 1A
Size and weight 40 x 40 x 36.5 mm, about 66 g

Compatible software and platforms

Node-RED, SenseCraft AI, RTSP, HTTP, WebSocket, TCP/UDP, MQTT and Frigate. Linux (Buildroot) with a C++ SDK.

What's in the box

  • reCamera 2002 64GB
  • USB Type-C cable
  • Ethernet cable

Buy separately

  • 5V USB power source
  • microSD card (optional)
  • External case for wall mounting
  • Commercial Ultralytics licence for commercial use of YOLO models

Setup

Plug it into a computer with USB Type-C, open the local web dashboard and deploy a model. Mount it magnetically or on a 1/4 inch tripod thread.

Specifications and package contents are from the manufacturer's product page and may change between production batches.

Works well with