← Back to Sample Lab
SignIQ · curated sampleInteractive viewerEgocentricHand posePrivacy-cleared
SignIQ Egocentric Human — Pick-and-Pack Activity
Egocentric human capture (head-mounted camera + 150-D hand pose + IMU) · Egocentric human task footage for robot learning
Episode Summary
InstructionHuman pick and pack task (egocentric capture, hands visible).
450
Frames
15.0s
Duration
30 Hz
Control
1
Cameras
190
State dim
190
Action dim
5
Subtasks
success
Outcome
CamerasEgocentric (head-mounted)
Modalitiesvideo × 1 (head-mounted, 1080p source) · 150-D hand pose · 40-D finger angles · 26-D gesture probabilities · 6-DoF IMU · task instruction
Format exportsLeRobot · MP4 + JSON · Parquet
LicenseDemo data licensed for inspection; production deliveries are SignIQ commercial.
QA Health
9of 10 checks pass
1 warn
- Schema parsedPASS
- Timestamps monotonicPASS
- Video ↔ pose frame countPASS
- No NaN/Inf in posePASS
- Consent for model trainingPASS
- Face blur / PII reviewPASS
Camera previews
Schema preview
observation.state
shape [190]
- [0]hand_joint_000
- [1]hand_joint_001
- [2]hand_joint_002
- [3]hand_joint_003
- [4]hand_joint_004
- [5]hand_joint_005
- …and 184 more
action
shape [190]
kind:human pose (mirrors state for export consistency)
control_hz:30 Hz
Interactive inspector
Synchronized cameras, action vs. observed state, joint-space trajectory, and dataset health — driven by the live bundle.
Loading episode bundle…
Why look at this
First-person human capture during a pick-and-pack activity, with full hand articulation (150-D pose + 40-D finger angles), IMU, and auto-derived motion-energy spans. Consent-cleared, face-blurred.
Best for
- →VLA reasoning pretraining
- →World models
- →Affordance priors
- →Human-to-robot transfer
Capture & license
CaptureSignIQ curated sample
LicenseCommercial — sample for inspection (consent + face blur)
See full dataset details →