# Datasets and skillspaces

> Public page summary · DVIDIA

Canonical page: [Datasets and skillspaces](https://dvidia.org/datasets)

## Datasets and skillspaces

Browse searchable publisher collections and open the skillspaces saved on this device. Publisher references include [Eidon POV](https://huggingface.co/eidon-ai), [DROID](https://droid-dataset.github.io/) and [Open X-Embodiment](https://robotics-transformer-x.github.io/). Their own terms apply; they are not DVIDIA-reviewed releases.

Skillspace cards show their brief, locally saved recordings and supplied creator credit. Local credit is self-declared. Source contributor IDs identify publisher records and are not verified DVIDIA handles. Source attribution remains separate from a person claiming a public account.

The connected operator workspace can browse stored server collections and loaded shared drafts. Its private footage is not exposed by this public page or mirror. Publishing uploaded footage into a public skillspace is not connected yet.

[Add a video](https://dvidia.org/record) · [Create a skillspace](https://dvidia.org/skillspaces) · [Your recordings](https://dvidia.org/workspace).

### Review an existing report

Import an exported classification report, inspect its source and processing information, and choose original local videos for review. Machine category proposals are suggestions, not calibrated skill evaluations or reviewed labels.

Inspect frames and timestamped evidence, review or correct annotations, and export the review for the processing worker. Source rights, project assignment, usage permissions and human review are separate decisions. Choosing a local video does not send its contents to DVIDIA or an AI provider.

The manual POV evidence inspector helps explore a recording; it does not establish semantic labeling accuracy, calibrated 3D reconstruction or robot control. Server collection access requires a separate connected operator session and is not exposed by this public mirror.

LeRobot v3 is the target robot dataset contract. No validated robot pack or policy should be inferred from human footage, a machine proposal or a matching folder structure.

[Read artifact contracts](https://dvidia.org/spec) · [Plan a skillspace](https://dvidia.org/skillspaces).

---

[Agent index](https://dvidia.org/llms.txt) · [Markdown mirror](https://dvidia.org/datasets/mirror.md)