This novel dataset is gated due to privacy concerns of the individuals whose faces are present. Lanfrica, as custodian of this dataset, is committed to ensuring that its use, while open, respects data governance principles and protects the dataset creators' concerns.
This Nigerian Sign Language (NSL) dataset was created in collaboration with more than 20 signers. In order to protect their images and preserve their control over their personal data, we are providing access to the dataset on a case-by-case basis, and asking that users of the dataset not disseminate or further re-share versions of the dataset.
In order to balance the needs of open and reproducible research with respecting the rights and intentions of the original data owners and data subjects, we adopted the following approach to sharing the dataset:
You fill out the form below, indicating the use-case for this dataset.
The form is reviewed by a team composed of the dataset owner and members of the Lanfrica dataset ethical team.
You will receive an email to fill out a legally binding agreement.
Finally the dataset will be shared with you. We are quite accommodating and will collaborate with you to share the dataset in the format that is most convenient for you.
Examples of such research projects may include:
Improving sign language modeling with the dataset
Benchmarking analysis
Commercial applications that benefit the creator community are also welcome. There will be discussion on the terms of the commercial license of the dataset.
Access to the full corpus should meet the following conditions:
The identities of those in the dataset will not be used explicitly for face recognition or any other privacy-infringing form of machine learning.
We are working on our own ethical charter for Lanfrica and will ask users to be consistent with it. In the meantime, we ask that uses be consistent with the BigScience Ethical Charter which reflects some of the values we prioritize for our project.
If you have any questions, please send an email to info@lanfrica.com