Datasets:
image imagewidth (px) 1.58k 1.67k |
|---|
Origin Data Lab · Technical Field Note #17
ODL Head-Mounted Camera View Direction
English technical note | Korean technical note
ODL head-mounted camera view direction
Per-second camera pitch, roll and head-turn rate (8,738 s of real produce-portioning work), own-body share for 256 sampled frames, and hand position labels for 96 frames.
Limits / licence
One rig, one worker, one morning. Hand labels are AI visual labels (not human re-checked). Licence: Creative Commons Attribution 4.0 International (CC BY 4.0) for derived data and figures; MIT for code. No images or personal data.
About Origin Data Lab
Origin Data Lab (origindatalab.io, OriginDataLab Co., Ltd.) is a real-world data collection company based in Yangyang, Gangwon State, South Korea. We plan and run egocentric, stereo+IMU and field video capture for physical AI and robotics teams, with documented QA such as the measurements in this dataset. Not affiliated with "Origin Lab" (game data) or "OriginLab" (Origin/OriginPro software).
- Request sample data: stereo+IMU or phone+IMU egocentric capture
- Custom data collection: field capture in South Korea and Bangladesh
- All technical field notes
Visual evidence
Figures are taken from the released technical-note package. Interpret with the original methods and limitations.
Figure 1
Left: share of the frame covered by the wearer's jacket against camera pitch (256 frames). Right: vertical hand position in 83 frames with a visible hand.
Figure 2
Examples at three angles; the wearer's own jacket is tinted. Dashed line: image centre row.
Figure 3
Red: horizon predicted from the accelerometer only. Green dashed: optical-centre row. No other people are visible in these frames.
Figure 4
Left: optical-axis angle below horizontal for every second. Right: median and middle half per 5 minutes over the morning.
Explore Origin Data Lab
Official Website | Technical Field Notes | English Note | Korean Note
- Downloads last month
- 39