This PR improves the dataset card for SeCVOS by:
- Adding the `task_categories: image-segmentation` metadata to properly categorize the dataset for video object segmentation tasks, enhancing its discoverability on the Hugging Face Hub.
- Including a sample usage snippet to guide users on how to easily load and explore the dataset using the `datasets` library.