Data & collection
Data provenance
Data provenance is information about a dataset's origins, the activities that created or transformed it, and the people or systems responsible. It connects delivered files and labels to their source records so users can assess reliability and trace changes; it does not by itself prove ownership or permission to sell the data.
Also known as: dataset provenance
Updated
Origins, transformations and responsibility
The W3C PROV framework describes provenance using entities, activities and agents. An entity can be a source recording or a derived annotation file. An activity can be capture, redaction, annotation or conversion. An agent can be a person, organisation or software system responsible for an activity. These relationships support assessments of reliability and trustworthiness.
The PROV primer also describes derivation, revision, timing and the procedures followed during an activity. A newer file can be identified as a revision of an earlier file, while retaining the relationship between them. The vocabulary is useful for designing a delivery manifest even when a supplier does not serialize its records in a formal PROV format.
Following one egocentric episode
Consider a hypothetical head-camera recording that becomes a training episode. Its provenance can connect the original capture identifier and recording time to the applicable sensor calibration, the redaction process, the selected time interval and the exported video. A hand-pose file can additionally reference the estimator version, input frames, confidence values and any human corrections.
If those poses are later mapped to a robot gripper, record motion retargeting as another transformation. The resulting targets remain derived from human motion; their presence in an action field does not turn them into commands measured during robot execution.
Stable identifiers, transformation versions and input-output references help a buyer locate affected episodes when an annotation or calibration error is found. A dataset card summarizes this process; the detailed provenance records let the buyer follow individual files and labels.
Provenance has an evidence boundary
A documented chain can be incomplete or wrong. PROV provides a way to express the chain, not a guarantee that the underlying statements have been verified. It also does not determine whether a collector had permission to record a location, whether participant consent covers the intended use, or whether a seller can license the result.
For a commercial release, link the provenance records to the relevant consent, location-permission and licence evidence through access-controlled references. Buyers can then distinguish the technical history of a file from the separate evidence supporting its permitted use.
Sources
Related terms
Data & collection
Dataset card
A dataset card is documentation that explains a dataset's contents, collection and processing, intended uses, limitations, and access or licensing terms. It helps users judge whether a particular release fits their task, but does not independently certify its quality or grant rights beyond the applicable licence.
Data & collection
Egocentric data
Egocentric data is sensor data recorded from the viewpoint of the person or robot performing an activity, most commonly with a head- or body-mounted camera. It can also include audio, gaze, depth or inertial signals. For humanoid learning, it shows hands, objects and actions from an actor-centred perspective.
Data & collection
Sensor calibration
Sensor calibration is the estimation and documentation of parameters that map raw sensor readings into physically meaningful values and known spatial relationships. In robotics it can include scale, bias, distortion, intrinsic camera parameters, sensor-to-sensor or sensor-to-robot transforms, and timing offsets. Calibration does not remove all noise or drift.
Models & learning
Pose estimation
Pose estimation is the process of inferring the position and orientation of a body, object, camera, hand, or robot relative to a specified coordinate frame. In three-dimensional robotics this is often called 6D or 6-DoF pose estimation because the result has three translational and three rotational degrees of freedom, even when orientation is stored with more than three numbers.
Simulation & transfer
Motion retargeting
Motion retargeting is the adaptation of a recorded or generated motion from one body to another with different proportions, joints or limits. For humanoid robots, it maps source poses or trajectories into robot configurations while preserving task-relevant relationships such as contacts and end-effector paths and satisfying kinematic, balance, collision and actuator constraints.