Skip to main content

ArUco Quickstart (HVS - Raze)

This quickstart gets ArUco tag detection running on a single HVS - Raze camera stream.

What You Need​

  • An HVS - Raze running a HeliOS image
  • Printed ArUco tags (this quickstart assumes 36h11)

Steps​

  1. Open the Web UI:

    http://<device-ip>:5800/

    If mDNS is working on your network:

    http://helios.local:5800/

  2. Go to Devices and click Register stream.

  3. In the register modal:

    • Pick the camera device.
    • Pick the backend (for on-device cameras this is typically Libcamera).
    • Pick a mode (format + resolution + FPS).
    • Click Register stream.

    Notes for HVS - Raze:

    • The built-in camera sensor is OV9782. Start with the Libcamera backend unless you have a specific reason not to.
    • For ArUco / mostly-mono work, prefer NV12 if it is available in the mode list.
    • If you select NV12, also select the luma decoder nv12-luma (it avoids extra decode/color work when you only need the luminance plane).
    • NV12 is cheap to handle (luma plane is directly usable for many CV steps).
    • NV12 avoids MJPEG decode work compared to MJPG.

  4. Back on Devices, click the new stream card to open it.

  5. Open the Controls tab and set noise reduction to Fast (if available) to suppress sensor noise without heavy processing.

  6. Go to Pipelines and click New Pipeline.

  7. Name the pipeline.

  8. In Start from, choose Templates and pick:

    • ArUco Tags (template id: daedalus_aruco)

  9. Click Create.

  10. Go back to your stream (Devices > stream), open the Pipelines tab.

  11. Click Add, check the pipeline you created, then close the assign dialog.

  12. Drag the pipeline into the 1x1 grid slot.

  13. Set Output to frame.

At this point the stream preview should show the pipeline output (an annotated frame when using the ArUco templates).

Tuning (The 2 Settings That Usually Matter)​

  1. Open the pipeline tuner (sliders icon next to the pipeline) and set Decode > Dictionary:
    • For this quickstart: 36h11
  2. If detections are unstable, reduce motion blur and improve lighting before touching thresholds.

Outputs You Can Expect From The ArUco Templates​

The built-in ArUco templates expose multiple host output ports. Common ones:

  • frame: annotated frame (overlays + tag count)
  • json: detections serialized as JSON
  • detections: structured detections
  • mask: intermediate mask/debug view

If you need to pull a sample over HTTP (without the UI), the API supports per-stream output sampling:

http://<device-ip>:5800/v1/streams/<stream-id>/pipeline/outputs/<port>/sample

Other Ways To Attach And Tune​

  • Stream settings (backend/mode/codecs): open the stream, go to the Stream tab, then click Apply stream settings.
  • Camera controls (exposure, gain, etc.): open the stream, go to Controls.
  • Pipelines: attach/detach and pick which output is live in the stream preview from the Pipelines tab.
  • Other stream sources: Register stream also supports Netcam (remote MJPEG stream URLs from peers) and File (replay from on-device media).