ZhenShiL

/

FarSLIP

like 1

This PR significantly enhances the model card for the FarSLIP model by adding crucial information for discoverability and usability.

It includes:

• Metadata:
  • pipeline_tag: zero-shot-image-classification to improve discoverability on the Hugging Face Hub.
  • library_name: open_clip as evidenced by the open_clip_train.main scripts and Open-CLIP acknowledgements in the GitHub repository, enabling automated code snippets for usage.
  • datasets: [ZhenShiL/MGRS-200k, omlab/RS5M] to link to the primary datasets used by the model.
  • tags: [remote-sensing] to categorize the model accurately within its domain.
• Structured Content:
  • Direct links to the paper and the GitHub repository.
  • A comprehensive introduction, checkpoints table, and dataset information from the GitHub README.
  • A "Zero-shot scene classification" usage section with an explicit code snippet, directly sourced from the GitHub README, to guide users on how to run the model.
  • Visuals from the GitHub repository to enhance understanding.
  • The official BibTeX citation.

These additions make the model card more informative, discoverable, and user-friendly.