Proven across
Multiple Industries

Product Overview
reCamera is Seeed Studio’s product family of AI cameras, designed to bring intelligent vision capabilities to the edge. The current reCamera product portfolio is shown below.
reCamera Pro is an advanced edge AI camera that combines computer vision, multi-modal perception, and flexible automation workflows in one compact device. It runs AI models locally, from YOLO-based object detection to LLM, VLM, STT, and TTS models, enabling responsive AI inference directly at the edge.

Designed for customizable AI applications, reCamera Pro supports diverse triggers including AI detection, audio events, GPIO signals, HTTP requests, UART commands, and scheduled tasks. With industry-standard protocols including RTSP, ONVIF, MQTT, HTTP, and UART, it seamlessly integrates with existing NVR, PLC, and IoT systems.

With an 8MP starlight sensor, interchangeable lenses, built-in microphones, IMU sensors, SenseCraft AI, and an intuitive Web UI, reCamera Pro empowers developers to quickly build, customize, and deploy intelligent edge AI applications — from AI vision prototypes to real-world automation solutions.

Feature 1：Run AI Models Locally at the Edge
Built around the Rockchip RV1126B AI SoC, reCamera Pro is a powerful Edge AI Camera capable of running a wide range of AI models locally. With a 3 TOPS NPU and quad-core Arm Cortex-A53 CPU, it delivers real-time inference for computer vision tasks (such as object detection, classification, and recognition）, VLM, LLM and speech AI applications.

Run different AI Models Locally
Supported AI models include:

• Computer Vision Models (YOLO)
• Vision Language Models (VLM)
• Large Language Models (LLM)
• Speech-to-Text (STT)
• Text-to-Speech (TTS)

AI Performance Benchmark

• Vision CNN: YOLO11 @ 43 FPS (640 × 640)
• LLM: Qwen2 0.5B (W8A8) — 13.63 tokens/s
• VLM: InternVL3-1B — 6.08 tokens/s
• STT: Whisper Base with real-time support (RTF ~0.40–0.50)
• TTS: MMS-TTS with real-time support (RTF ~0.09–0.18)

Compatible with Popular AI Frameworks
Support for ONNX, PyTorch, TensorFlow Lite, Caffe, and Darknet, with RKNN-Toolkit2 for efficient model conversion and deployment.

Explore Tested AI Deployment Examples
Explore AI models successfully deployed on reCamera Pro, with reference code, setup instructions, and benchmark results available on GitHub. More deployment examples are continuously being added, with a growing focus on cv models.

Feature 2：Enable Multi-Modal Sensing for Smarter Event Detection
Vision alone isn't always enough to fully understand real-world events. By combining audio and external sensor data with visual information, AI can gain richer context for more accurate event understanding. reCamera Pro combines vision, audio, motion, and external sensor inputs to understand real-world events from multiple perspectives.
With built-in dual microphones, IMU, and expandable GPIO and UART interfaces, it enables multi-modal AI applications beyond traditional camera-based detection

Intelligent Event Detection with Multiple Inputs
Dual Microphones
Enable always-on audio monitoring for intelligent event triggering.
Detect events such as:

• Glass breaking
• Abnormal impact sounds
• Custom voice keywords (e.g., "Hands Up")

Built-in IMU
Monitor vibration, movement, and orientation changes in real time.
Applications like:

• Device Impact Detection
• Shock Event Alerts
• Device Orientation Monitoring

GPIO & UART Sensor Expansion
Connect external sensors and devices to expand AI applications, like:

• CO2 sensor
• Temperature sensor
• Radar sensor
• Human presence sensors
• more...

Feature 3：Build Smart Event-Driven AI Workflows
Turn AI perception into automated actions with flexible event-driven workflows. reCamera Pro works not only as a standard IPC camera, but also as an intelligent edge device that responds to AI detection results, audio events, sensor inputs, and external commands. Define custom triggers and actions to automate different scenarios without additional controllers.

Flexible Trigger Sources
Supported triggers:

• AI Detection
• Audio Event
• Sensor Event
• GPIO
• UART
• HTTP Command
• Timer

Customizable Actions
Supported actions:

• Capture Image
• Record Video
• Play Audio
• HTTP Callback
• MQTT Publish
• RTSP Streaming
• Save to NVR
• Send Alert

Feature 4： Embedded WEB UI For Quick Setup
Configure, deploy and monitor your AI camera directly through a browser—without complicated development tools.
The built-in Web UI simplifies device setup, AI model management, workflow configuration, and system maintenance.

Web-Based Management Features
Support:

• Main Stream / Sub Stream Configuration
• AI Workflow Configuration
• Model Output Management
• Device Monitoring
• Terminal Management
• Privacy Mask Configuration
• System Settings

Feature 5 ：Build Custom AI Models in 1 Minute with No Code and Few Samples

With SenseCraft AI, a no-code AI training platform in the SenseCraft AIoT ecosystem, reCamera Pro makes custom AI model development simple. Train, convert, and deploy custom vision and sound classification models with no code and only a few samples. These models can be adapted to a wide range of real-world applications, allowing AI to learn the visual and audio patterns that humans naturally recognize.

Vision classification models
Vision classification models can learn many of the visual judgments humans make every day. By training with your own examples, reCamera Pro can recognize differences in appearance, quality, condition, and categories—enabling applications such as quality grading, object classification, scene recognition, and custom inspection tasks.

Create Models Like:

• Inventory Status Recognition
• Queue Status Recognition
• Material Level Monitoring

Sound classification models
Sound classification models enable AI to recognize the sounds humans naturally identify—what a sound is and whether it is normal or unusual.

Applications include:

• Glass breaking, alarms, horns, dog barking, and emergency keywords such as “Help”
• Machine abnormal noise and equipment condition monitoring
• Cough pattern and voice condition analysis
• Custom sound event recognition

With reCamera Pro, you can combine audio and vision for richer, more intelligent analysis.

Feature 6 ：Seamlessly Integrate with Existing Systems
Connect reCamera Pro with your existing infrastructure through widely adopted industrial and security protocols.
Designed for integration with video management systems, PLC automation, and IoT platforms, reCamera Pro helps accelerate AI deployment without replacing your current systems.

Supported Protocols

• RTSP, RTMP, ONVIF, HTTP, MQTT, UART, CAN

Can be Used with:

• Video Management Systems (VMS)
• NAS
• Network Video Recorders (NVR)
• Smart Buildings
• PLC Automation Systems
• Factory Automation
• IoT Platforms

It also works with Frigate & Home Assistant.

Feature 7 ：Easy Development with Flexible APIs
Access the full capabilities of reCamera Pro through standard HTTP APIs.
Developers can remotely control devices, retrieve AI results, manage workflows, and integrate camera functions into applications without relying on proprietary communication protocols.
Check API on Wiki

Feature 8 ：Support Flexible Camera Lens Options for Diverse Applications
Adapt your Edge AI Camera to different environments with flexible lens configurations.
reCamera Pro supports:

• M12 Lens
• CS Mount
• C Mount

Feature 9： Build Vision Automation with Natural Language Commands
Transform natural language instructions into automated camera workflows. reCamera Pro is powerful enough to install lightweight agent like NanoBot, PicoClaw, ZeroClaw, without manually creating complex automation logic.
reCamera Pro provides an official API Skill that packages camera controls, AI inference results, and common device functions into Agent-ready tools.
Examples:

• Send me a message if someone walks by.
• Take a photo when you see a cat.
• Start recording if someone enters the area.

Specification

Hardware Overview