reCamera is a fully open-sourced, modular AI camera. Featuring 1 TOPS @INT8, built-in YOLOv11 & Node-RED, Linux-based, supports custom AI models, and a 5MP camera, it’s a smart eye for any system. reCamera is the go-to AI camera for YOLO at the edge.

Features:

Feature 1: Built-in YOLO11 AI Models & 1TOPS NPU Computing Power

4 YOLO models pre-installed:

• Object Classification
• Objection Detection
• Pose Detection
• Instance Segmentation

Licensing: reCamera does not include an Ultralytics commercial license. For commercial use of Ultralytics YOLO models, please visit Ultralytics YOLO Licensing page for available options.

Feature 2: Modular Design

reCamera adopts a modular design. You can choose the components based on your needs.

Sensor Board features a swappable and interchangeable camera module design. The current version comes with the OV5647 sensor. If you need higher imaging performance, we also offer a version with a sensor board with GC2053 reCamera 2002 Sensor Board GC2053which has a 1/2.9'' CMOS sensor and supports changing lenses. We also offer M12 lens options for different FOV requirements. More sensor boards are on the way - feel free to share your needs at GitHub · Where software is built

Core Board provides system control and computing power. It has a RISC-V SoC and a 1Tops@Int8 NPU. Also, the Core Board provides Wifi capability. We offer individual Core Boards allowing users to purchase the core board separately and build their own custom reCamera:

• reCamera 2002w Core Board 64GB with Aluminium Frame
• reCamera 2002 Core Board 64GB with Aluminium Frame
• reCamera 2002w Core Board 8GB with Aluminium Frame
• reCamera 2002 Core Board 8GB with Aluminium Frame

Base Board provides connectivity. It offers interfaces including USB, Ethernet and debug serial ports enabling easy integration with external devices and systems. We also offer a version with PoE and 3 GPIOs extension. Alternative Base Board for PoE and GPIO extension: reCamera 2002 Base Board with PoE

Feature 3: Fast Development/Deployment with Node-RED

Node-RED enables low-code, visual workflow building. This allows users to build and manage application logic quickly with minimal coding. 

Based on Node-RED, we provide product-specific nodes that let you deploy a model and see results immediately. For example, simply drag in a Camera, Model, and Preview node to complete the application flow.

Feature 4: Fully Open-sourced Hardware/Software

reCamera is fully open-sourced so users can modify software and AI pipelines without limitation. You can train your own YOLO model and deploy it on reCamera.

• Hardware: module board documentation is fully open, including 3D files, schematic, PCB files
• AI Functions: inference pipeline, NPU runtime integration, YOLO11n models, and Node-RED AI nodes are fully open.
• Applications: Node-RED flows, C++ SDK, RTSP examples, and Web UI resources are openly provided.

System: Buildroot, U-Boot, Linux Kernel, device trees, drivers, and system services are open-sourced.

Feature 5: Multiple Interfaces Supported

reCamera outputs data through standard protocols such as RTSP, HTTP, WebSocket, and TCP/UDP, virtually any online platform or application. Also, use MQTT to connect to other IOT devices.

Feature 6: Add external enclosure

reCamera 2002 measures 40 × 40 × 36.5 mm, and the reCamera HQ PoE measures 40 × 40 × 53.1 mm. The compact sizes make them compatible with many off-the-shelf waterproof enclosures.

If you want to deploy reCamera outdoor or in wet environment, we offer IP66 and IP67 enclosures.

More enclosures will be added in the future.

Feature 7: Visual Application Development

reCamera can provide real-time visual intelligence is various scenes. From smart homes to agricultural fields, industrial facilities, smart cities, reCamera brings powerful on-device AI analytics to every scenario.

• Object Classification: Classify the entire image
• Parking Occupancy Detection: Identify available and occupied parking spaces
• People Counting: Detect and count foot traffic in crowded areas such as subways and malls.
• Restricted Area Intrusion Detection: Detect unauthorized entry into hazardous areas
• PPE Detection: Verify whether workers are wearing required personal protective equipment
• Object Tracking: Track targets and visualize their movement paths
• Fall Detection: Detect falls using human pose estimation
• Smart Fitness Assistant: Evaluate exercise form using pose estimation
• Defect Detection: Segment surface defects on parts or equipment

More demos waiting to be explored, such as a turn your reCamera into a motion-sensing game controller.

Deploy Vision AI Models and Apps in One Click

Use SenseCraft AI with reCamera to deploy AI models and AI applications in one click. It supports common vision AI scenarios such as people detection, object recognition, intrusion detection, counting, and real-time monitoring, helping you quickly turn reCamera into an edge AI camera for smart security, retail, industrial inspection, and automation projects.

Specifications:

• reCamera 200x Series : reCamera 2002 : reCamera 2002w : reCamera Gimbal : reCamera HQ PoE
• Processor System : Al Performance : ni1TOPS@Int8
• SOC : SG2002
• CPU : C906@1GHz + C906@700MHz
• MCU : 8051@8KB SRAM
• Memory : 256MB
• Video Encode : 5MP @ 30Fps
• Camera : Sensor : OV5647 : GC2053
• Resolution : 5M(2592*1944) Pixels : 2MPixels
• Chip : 1/4" CMOS Sensor : 1/2.9'' CMOS Sensor
• Pixel Size : 1.4μm : 2.7μm
• Output Format : RAW 10
• Maximum Frame Rate : full size: 15fps : full size: 30fps
• Aperture : F2.8
• Equivalent Focal Length : 3.46mm : 2.9mm
• Field of View : 65° : DFOV 93° HFOV 84° VFOV 54°
• Distortion : <1%
• Lens structure : 5P
• Storage : eMMC : 8GB/64GB
• Expansion : removable TF card (up to 2TB, not included)
• I/O : Ethernet : 100Mbps : 100Mbps with PoE
• Wireless : / : WIFI2.4G/5G BT4.2/5.0 : WIFI2.4G/5G BT4.2/5.0 : /
• USB : USB Type-C(2.0)
• Button : 1x Reset Button, 1x Boot Button
• Fill Light : 0.3w White Light : /
• LED Indicator : 1x Power indicator,
• 1x eMMC status indicator,
• 1x CPU status(User LED)
• I/O Port : / : 3 I/O Ports 487,488,490
• Audio : Mic : On-board Mic
• Speaker : External Speaker
• Motor : / : MS3008/MS3506 : /
• Dimension : 40x40x36.5mm : 68x112x71mm
• 40x40x45.8mm
• 40x40x53.1mm with lens
• Power : 5V/1A USB Type C
• Operation Temperature : -20～50°C
• Operation Humidity : 95%

Hardware Overview:

Applications:

Integration of Real-time Heatmap with Grafana Data Dashboard

The project conducts target detection using the YOLO11n model built into reCamera. It performs fuzzy processing on personnel targets and displays a heat map effect. All of this is handled locally within reCamera! And reCamera will send the detected data to the locally running Influxdb database. Then, Grafana reads the data from the Influxdb database to display real-time data dashboards.

AI Remote Wireless Monitoring System Based on reCamera and Wifi Halow

An AI-based remote wireless monitoring system using Wi-Fi HaLow. It combines edge AI vision with long-range, low-power wireless communication to enable real-time detection, data transmission, and monitoring in remote or infrastructure-limited environments such as farms, factories, and outdoor sites.

Video link: https://www.linkedin.com/posts/seeedstudio_howtobuildanaicamera-outdoor-wifi-activity-7401880305940553729-R57w?utm_source=share&utm_medium=member_desktop&rcm=ACoAAGCDEqUBKdDsqK6EqGXClNocxAAtw3IA_T0

Tutorial: AI Remote Wireless Monitor System with Wifi HaLow | Seeed Studio Wiki

AI Parking Management with reCamera

This project uses Seeed's reCamera to detect parking slot occupancy in real-time. It features multi-frame stabilization to avoid flickering and displays results (free/occupied counts) directly on the video stream. The entire workflow runs automatically on the device via a pre-configured Node-RED setup.

Tutorial: AI Parking Slot Monitoring Demo with reCamera | Seeed Studio Wiki

Play Subway Surfers with reCamera HQ PoE

Using its built-in pose detection model, reCamera can recognize your hand and body movements in real time and convert them into game actions, no controllers, no wearables, just pure gesture play. Wave your hand to jump, raise your arm to attack, or lean left and right to dodge obstacles.

Video link: https://www.linkedin.com/posts/seeedstudio_howtobuildanaicamera-recamera-yolo11n-activity-7399730949401526272-UdZ4?utm_source=share&utm_medium=member_desktop&rcm=ACoAAGCDEqUBKdDsqK6EqGXClNocxAAtw3IA_T0

Tutorial: Play Subway Surfers with body sensor control | Seeed Studio Wiki