Improve dataset card: Add paper, project page, code links, task categories, and sample usage
#2
by nielsr HF Staff - opened
This PR significantly enhances the dataset card for the OpenDWM data packages.
Specifically, it adds:
- A clear introduction explaining the dataset's purpose in the context of the OpenDWM project and the related CVD-STORM paper.
- Explicit links to the paper (CVD-STORM: Cross-View Video Diffusion with Spatial-Temporal Reconstruction Model for Autonomous Driving), the project page (OpenDWM), and the associated code repository (GitHub - SenseTime-FVG/OpenDWM).
- Relevant
task_categories(text-to-3d,text-to-video) andtags(autonomous-driving,video-generation,4d-reconstruction,world-model,lidar) to the metadata for better categorization and searchability on the Hub. - A "Sample Usage" section, extracted directly from the GitHub README, demonstrating how to use these data packages for layout-conditioned video and LiDAR generation within the OpenDWM framework, including basic setup instructions.
These additions will make the dataset more informative and discoverable for researchers interested in autonomous driving and generative models.