This PR enhances the DANI dataset card by:
- Adding `task_categories: image-classification` to the metadata, aligning with the dataset's core purpose of distinguishing AI-generated from natural images.
- Including `library_name: datasets` to indicate the primary library for loading the dataset.
- Adding relevant `tags` such as `ai-generated-content`, `image-quality-assessment`, `real-vs-fake`, and `multimodal` for better discoverability.
- Integrating direct links to the paper ([D-Judge: How Far Are We? Evaluating the Discrepancies Between AI-synthesized Images and Natural Images through Multimodal Guidance](https://huggingface.co/papers/2412.17632)) and the associated GitHub repository ([https://github.com/RenyangLiu/DJudge](https://github.com/RenyangLiu/DJudge)) at the top of the card for quick access.
