Intel RealSense · computer vision

One camera. Six kinds of tracking. Real 3D.

BlinkPose.Track turns an Intel RealSense depth camera into a live tracking hub — ArUco markers, full-body pose, hands, faces and circular markers, detected in a single pass and each pinned to a real-world 3D position. It streams straight to a browser dashboard, with a live video feed and an in-browser 3D view of everything it sees.

All modules

6 entry types · MJPEG video · in-browser 3D · real-time

localhost — BlinkPose.Track

Plate I — the Track dashboard: live camera feed, the detection table and the in-browser 3D view, all served from the module’s own process.

What it detects

Six trackers, one camera

Every detection runs on the same frame and lands in one live list — mix and match whatever your setup needs.

ArUco markers

Fiducial tags with a stable ID and a full 6-DOF pose — the workhorse for rigid objects and tools.

Human pose

Full-body skeletons, one keypoint set per person, tracked live across the frame.

Hands

Per-hand landmark sets, left/right aware — for gesture and fine interaction.

Faces

Face detection per person, with a simple expression read.

Face landmarks

A dense facial-landmark mesh per face for finer facial tracking.

Circular markers

Round fiducials for quick, cheap targets where a full ArUco tag is overkill.

Inside the dashboard

Everything on one screen

The whole camera is controllable from the browser — no separate app, no config files.

Live video feed. The camera stream inline, switchable between annotated, raw and depth views.
Marker table. Every detection this frame — id, type, confidence, pixel position and 3D coordinates — colour-coded by type with a legend.
In-browser 3D view. See all detections placed in real 3D space, not just as boxes on a video.
Detection toggles. Turn each tracker (ArUco, pose, hands, faces, markers) on or off live — disabled ones cost nothing.
Camera controls. Exposure, gain, white balance, IR and the projector laser, tuned from the page and saved as defaults.
Record & capture. Record sessions for later replay and grab screenshots straight from the dashboard.

Not just pixels

Every detection carries a real 3D position

Because it's a depth camera, each marker and keypoint is deprojected into real-world coordinates — markers additionally get a full 6-DOF pose. Downstream tools get metres, not just pixels.

detection types

6-DOF

pose per marker

depth-deprojected xyz

JSON envelope