Datasets:
Improve dataset card: Add metadata, links, abstract summary, and detection methods
#3
by
nielsr
HF Staff
- opened
This pull request aims to enhance the AIGIBench dataset card by:
- Adding
task_categories: ['image-classification']to the metadata, aligning with the dataset's purpose of evaluating AI-generated image detectors. - Including
tags: ['ai-generated-image-detection', 'deepfake-detection', 'generative-models', 'benchmark']for better discoverability. - Adding explicit links to the paper (Hugging Face Papers page) and the GitHub repository at the top of the card.
- Integrating a concise summary from the paper's abstract to provide immediate context.
- Integrating the "Detection Methods" section from the GitHub README to provide more context on how the dataset is used.
- Updating the citation section to use the
@inproceedingsformat from the GitHub README, reflecting its acceptance at NeurIPS 2025.
These changes will make the dataset card more informative and easier to navigate for researchers.