Wrapping a Gift - Egocentric Human Demonstration

A single continuous egocentric demonstration of a person wrapping a gift — handling wrapping paper, tape and scissors on a work surface. Captured with a head-mounted camera plus two forearm-strapped wrist cameras, all 1920×1080 H.264 at 30 fps. Packaged as a HuggingFace LeRobot v2.1 dataset: 1 episode, 8,554 frames (4.75 min), QC-complete with 0 Tier-1 FAILs. The parquet carries per-frame hand-pose annotation — 21-point 3D and 2D landmarks for both hands with per-landmark confidence and visibility, handedness scores — plus semantic action labelling: action phase, verb and noun with an LLM-generated reasoning trace, and a detected primary object with bounding box and hand-distance. 59 columns over 8,554 rows. The episode is further segmented into 58 annotated action windows. Notable for deformable-object manipulation (paper folding, creasing, tape handling), which is under-represented in rigid-body manipulation corpora.