How to get the dataset

For legal and ethical reasons, the dataset is only available to academics.

The dataset exists in two flavours:

• the anonymised version, with no video streams, that can be freely downloaded from the Zenodo platform (7.2GB compressed). Once downloaded, unzip it somewhere and follow these instructions to update the dataset with the latest meta-data.

• the full dataset, including raw video streams, can be obtained after getting suitable ethical clearance from your university ethics committee. Due to the size of the full dataset (several terabytes), we share this version by shipping to you an harddrive with the data. See the Application procedure.

Besides, the meta-data of the dataset (demographics, details of the recordings), as well as the video annotations, are stored and maintained as a public GIT repository.

Once you have obtained the dataset and updated it, check how to get started with it.

Dataset excerpt

For evaluation purpose, you can download a short excerpt (10 seconds) of the full dataset here:

Dataset excerpt (65MB)

Citation

If you use the dataset, please cite the two following references:

@misc{pinsoro2017, 
  author={Lemaignan, Séverin and Edmunds, Charlotte and Belpaeme, Tony}, 
  title={The PInSoRo dataset}, 
  DOI={10.5281/zenodo.1043507},
  publisher={Zenodo}, 
  year={2017}, 
  month={Nov}
}

@article{lemaignan2018pinsoro,
    author = {Lemaignan, Séverin AND Edmunds, Charlotte E. R. AND Senft, Emmanuel AND Belpaeme, Tony},
    journal = {PLOS ONE},
    publisher = {Public Library of Science},
    title = {The PInSoRo dataset: Supporting the data-driven study of child-child and child-robot social dynamics},
    year = {2018},
    month = {10},
    volume = {13},
    pages = {1-19},
    number = {10},
    doi = {10.1371/journal.pone.0205999}
}