Observer view — fused live scene from every camera publishing to
/ws/detect.
To publish frames, open / on a device with a webcam.
connecting…
0cameras
0people
0markers
1
Intrinsic calibrationLens model solved for each camera.
Needed
2
Extrinsic calibrationFloor corners mapped into metres.
Needed
3
Publishing cameraAt least one calibrated camera is live.
Waiting
drag = orbit · scroll = zoom · right-drag = pan
What this view shows
The grey rectangle is the floor — the same one the four corner
markers were placed on (per ADR 0012). Each detected marker is
drawn at its world-frame XYZ in metres, with an arrow showing yaw.
When two markers belong to the same person (ADR 0049: chest+back
or hat+chest), they fuse into one body avatar; the arrow then comes
from the chest↔back vector — a much more reliable yaw signal than
a single marker that may have rotated on the head. The teal wedge
marks where the camera is in world coordinates.