Sample Datasets#

To demonstrate Aeon’s various features and capabilities, we also provide the following datasets that complement the tutorials and how-to guides.

Short datasets#

These lightweight datasets are curated excerpts from full experiments, typically sized in gigabytes (GB). They are selected to support quick inspection, pipeline prototyping, or tooling validation before committing to full dataset downloads.

Full datasets#

Note

The links to the full datasets are temporarily unavailable while we resolve access issues. Please check back later or reach out to the project maintainers if you need urgent access.

These datasets contain complete experimental recordings captured across multiple acquisition machines, with sizes generally in the terabyte (TB) range. Each dataset is labelled by the experiment name (e.g. social0.2) followed by the name of the acquisition machine (e.g. aeon3).

“raw” datasets include raw data acquired from all streams used in an experiment, while “ingest” datasets contain only a subset of post-processed, derived data (typically full-pose position and/or weight data). For most analyses, using both “raw” and “ingest” datasets provides the cleanest and most complete view.

  • social0.2-aeon3

  • social0.2-aeon4

  • social0.3-aeon3

  • social0.3-aeon4

  • social0.4-aeon3

  • social0.4-aeon4

Precomputed datasets#

These datasets contain outputs derived from full datasets that have already been run through the upstream pipelines. Stored in compact formats (e.g. Parquet, pickle), they are designed to support tutorials and how-to guides without requiring users to re-run heavy preprocessing or analysis steps. While downstream of “raw” and “ingest”, they prioritise ease of use and clarity of analysis over full provenance traceability.