Improve dataset card: Add paper/code links, abstract, sample usage, update license, task category & tags

#1
by nielsr HF Staff - opened

This Pull Request significantly improves the dataset card for Face4FairShifts by:

  • Updating task_categories: Changed from feature-extraction to image-classification to accurately reflect the dataset's primary use for facial attribute recognition and fairness evaluation.
  • Correcting license: Updated from afl-3.0 to cc-by-nc-4.0 to align with the explicit "RESTRICTED to non-commercial research and educational purposes" statement in the dataset description, ensuring legal clarity.
  • Adding comprehensive tags: Included fairness, robustness, domain-generalization, face, facial-attributes, and benchmark for enhanced discoverability and to better categorize the dataset's focus. The code tag was also retained.
  • Integrating key links: Added direct links to the paper (https://huggingface.co/papers/2509.00658) and the GitHub repository (https://github.com/TurkeyWatanabe/FairDistBench) for easy access to related research and code.
  • Adding an "Introduction": Included the paper's abstract as an introduction to provide a concise overview of the dataset's goals and content.
  • Providing "Sample Usage": Detailed instructions and code snippets from the GitHub README have been added to guide users on environment setup, dataset processing, and running the evaluation code, making it easier to get started.
Ready to merge
This branch is ready to get merged automatically.

Sign up or log in to comment