Plug & Play AI Camera — Deployment Guide

View source on GitHub
AI Assistance

Need help choosing hardware, or ready to deploy?

Prefer not to manually configure devices or enter commands? Let AI Assistant size hardware based on your site setup, or run one-click script deployment once ready.

Self-service Consult & Deploy
Preset

Object Detection & Environment

Explore foundational edge vision capabilities with general object detection (YOLO) and weather classification. Watch bounding boxes and confidence scores update in real time over WebRTC to experience INT8 model efficiency.

  • Hardware: reCamera (SG2002) or reCamera Pro (RV1126B); Home Assistant integration optionally requires a PC or reComputer R1100.
  • Recommended Apps: yolo-detector (80 common COCO classes) or weather-classifier (weather recognition).
  • Quest:
    1. Hold a phone, mug, or hand in front of the lens and observe detection boxes and labels rendered within 50ms;
    2. Slowly move the object or partially occlude it, watching the confidence score shift to observe occlusion robustness.
1

Upgrade reCamera Console (SG2002)

Install console version 0.5.6. Existing 0.5.6 installations will be skipped automatically. If using reCamera Pro, skip directly to Step 2.

Prerequisites
  1. Connect the camera over USB (address 192.168.42.1) or place it on the same network.
  2. Username recamera, default password recamera (older firmware uses recamera.2).
  3. For brand new units, enable SSH under http://192.168.42.1/#/security first.
Troubleshooting
SymptomFix
Connection failedUse 192.168.42.1 over USB; check router DHCP list over network
Wrong passwordDefault is recamera (older firmware uses recamera.2)
Installation failedReboot camera and rerun this step
Node-RED unavailableConsole takes over the camera; switch back under Console System Settings
Factory resetHold User button while powering up until red LED turns solid
2

Install Object Detection App

Select the desired application from the dropdown and deploy it to the camera.

Mode

Select Object Detection (yolo-detector) or Weather Classification (weather-classifier) from the dropdown and click Deploy.

Prerequisites
  1. Your computer has internet access; the camera does not need internet access (the host computer downloads packages and models).
  2. Connect the camera via USB (192.168.42.1) or via local network IP.
  3. Default credentials are recamera / recamera (or recamera.2 on earlier firmware).
  4. Console version must be 0.5.5 or higher.
Troubleshooting
IssueSolution
Dropdown empty or incompleteHost cannot reach sensecraft-statics.seeed.cc; verify network and refresh
Login rejectedUse camera credentials (default recamera / recamera or recamera.2)
Console unresponsiveCamera may still be rebooting; wait one minute and retry
Busy promptConsole is processing another operation; retry when finished
App failed to startCheck status in console App page; redeploy if files are missing
After Deployment

The camera is running the object detection application. The app supports ONVIF and RTSP for NVR and video management systems.

3

Open the Console

Open the camera web console to inspect real-time inference and detections.

Mode

Open the camera console to inspect real-time video stream and detection results.

Prerequisites
  1. Sign in with the camera credentials from Step 2.
  2. In the Apps page, click Debug on the active app to view bounding boxes.
Troubleshooting
IssueSolution
Page does not openWait a minute for the camera to finish booting, then refresh
Live video is blackNo app is active; return to Step 2 to deploy an app
After Deployment

Console is available at http://<camera-ip>/ to inspect video, tune thresholds, and adjust network settings.

4

Deploy Home Assistant (Optional)

Start Home Assistant and an MQTT broker. Skip if both are already running.

Deployment method

Deploy Home Assistant on the local machine. Requires Docker Desktop running with ports 8123 and 1883 available.

Prerequisites
  1. Docker Desktop installed and running.
  2. At least 2 GB free disk space.
  3. Ports 8123 and 1883 available.
Troubleshooting
IssueSolution
Port 8123 or 1883 in useStop existing Home Assistant or Mosquitto instance
Docker fails to startLaunch Docker Desktop
Container crash loopEnsure at least 2 GB of memory is available
After Deployment
  1. Open http://localhost:8123 and follow the onboarding wizard.
  2. MQTT broker runs on port 1883 on this machine; supply this address in Step 5.
5

View Results in Home Assistant (Optional)

Inspect camera live stream and detection entities on the Home Assistant dashboard.

Mode

Connect Home Assistant and the camera to the same MQTT broker to auto-discover detection entities, then add a video glance card.

Prerequisites
  1. Add MQTT Integration: In Home Assistant Settings → Devices & Services → Add Integration, add the MQTT integration pointing to the Step 4 broker IP, port 1883 (leave username/password empty). Skip if already configured.
  2. Point Camera to Broker: In camera console Integrations → Home Assistant, enter the Step 4 broker IP and port 1883, test connection, then save.
  3. Inspect Entities: In Home Assistant Settings → Devices & Services → MQTT, locate the camera device and entities.
  4. Add Card: On the dashboard, add a Picture glance card to overlay detection entities onto the video stream.
Troubleshooting
IssueSolution
No data on brokerConfirm broker IP is reachable from camera
Entities unavailableConfirm the app is actively running on the camera
Card has video but no entitiesVerify MQTT integration status and entities list
After Deployment

The dashboard displays camera video and detection entities. Switching apps updates entities automatically.