Start Here
reCamera Pro is an AI camera built on the Rockchip RV1126B SoC (quad-core Cortex-A53 @ 1.2 GHz, 3 TOPS NPU), with a 4K camera (SC850SL, 4K@30FPS), a 6-axis IMU, microphone input, a 1 W speaker, 16 GB eMMC plus an SD card slot, and rich expansion interfaces (GPIO, UART, CAN, MIPI-DSI). Everything on the device — live preview, AI detection, event recording, sound model training — can be operated from the built-in Web UI, and deeper integration is available through the HTTP API and native SDK.
This page is your task navigator: pick what you want to do below, or follow the default getting-started route if this is your first time.
Default getting-started route
New to reCamera Pro? Follow these three steps in order:
- Quick start: see your first AI detection — power the device, log in to the Web UI, and watch a live detection.
- First task: detect and record — turn a detection into an event recording and play it back.
- Access the device over Wi-Fi — move from the direct connection to your local network.
What do you want to do?
1. Getting Started
2. Build Your App
- Preview, capture photos and record manually
- Adjust image quality and low-light performance — includes the 0.3 lux low-light sample
- Watch the stream in an external player (RTSP)
- Set up OSD and privacy masks
- Configure detection: classes, thresholds, schedule
- Configure event recording and find your recordings & manage storage
- Train your own sound model and trigger capture by sound
- Send detection results out: MQTT · HTTP / UART
- Trigger capture from an external device
- Sound-triggered alerts in Home Assistant
- Make your first API call
3. Use Your Own Model
- Choose a model deployment path
- Upload and configure an RKNN model
- Convert ONNX models with SenseCraft
- Convert models with RKNN-Toolkit2
- Validate model performance on device
4. Develop Your Own App
- Choose a development path
- Build a minimal SDK application
- Terminal, SSH and debug UART
- Develop with AI coding agents
- Camera / audio / inference pipelines
- Peripherals: IMU data · MIPI-DSI display · speaker & volume · GPIO pins
- Examples: tilt & shake detection · visual wake + speech recognition
- Experimental Debian 13 image
5. Deploy & Maintain
- Pre-deployment checklist
- Network, time and access control
- Back up and restore configuration
- Firmware update and recovery
6. Reference
- Hardware specifications and interface diagrams
- Web UI field & button index
- Model compatibility and output formats
- Downloads and release notes
- API reference
7. Troubleshooting
Capabilities at a glance
- AI vision — on-device object/person detection with configurable classes, thresholds and schedules.
- Sound sensing — Acoustic Lab for training custom sound models; sound can trigger capture and alerts.
- Recording — manual recording plus event-based recording with storage management.
- Streaming — live preview in the Web UI and RTSP output to external players.
- Integration — HTTP API, MQTT/HTTP/UART result output, GPIO/UART external triggers, Home Assistant example.
- Development — native SDK, AI-coding-agent workflow, experimental Debian 13.
Where to go next
- First time here? Start with the default getting-started route above.
- Bringing your own model? Go to Use Your Own Model.
- Writing code? Go to Develop Your Own App.
- Something not working? See Troubleshooting.